diff --git a/CMakeLists.txt b/CMakeLists.txt index c4a2d1d4..326c2103 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -199,10 +199,21 @@ set(LIBTORCH_LIBS_LINKER_ARGS ) cmake_print_variables(LIBTORCH_LIBS_LINKER_ARGS) +set(CHAI_LINKER_ARGS + -M ${PROJECT_ROOT_DIR}/lib + ${BRIDGE_DIR}/include/bridge.h + ${BRIDGE_OBJECT_FILES} + -L ${LIBTORCH_DIR}/lib + ${LIBTORCH_LIBS_LINKER_ARGS} + --ccflags "-I${BRIDGE_DIR}/include -L${PROJECT_ROOT_DIR}/build" + --ldflags "-L${PROJECT_ROOT_DIR}/build -Wl,-rpath,${LIBTORCH_DIR}/lib" +) + add_executable(TorchBridge ${BRIDGE_DIR}/lib/Bridge.chpl) add_dependencies(TorchBridge bridge) add_dependencies(TorchBridge ChAI) +add_dependencies(TorchBridge bridge_objs) target_link_options(TorchBridge PRIVATE ${BRIDGE_DIR}/include/bridge.h @@ -228,9 +239,27 @@ add_dependencies(TinyLayerTest ChAI) target_link_options(TinyLayerTest PRIVATE --main-module layer_test.chpl - # -M ${PROJECT_ROOT_DIR}/lib + -M ${PROJECT_ROOT_DIR}/lib ${CHAI_LINKER_ARGS} ) + + +add_executable(TinyBridgeSystemTest + ${PROJECT_ROOT_DIR}/test/tiny/bridge_system_test.chpl + ${CHAI_LIB_FILES} + ) +add_dependencies(TinyBridgeSystemTest bridge) +add_dependencies(TinyBridgeSystemTest ChAI) +add_dependencies(TinyBridgeSystemTest bridge_objs) +add_dependencies(TinyBridgeSystemTest TorchBridge) +target_link_options(TinyBridgeSystemTest + PRIVATE + --main-module bridge_system_test.chpl + -M ${PROJECT_ROOT_DIR}/lib + ${CHAI_LINKER_ARGS} +) + + # chpl test/tiny/layer_test.chpl -M lib bridge/include/bridge.h build/CMakeFiles/bridge.dir/bridge/lib/bridge.cpp.o -L libtorch/lib -ltorch -ltorch_cpu -lc10 -ltorch_global_deps --ldflags "-Wl,-rpath,libtorch/lib" # chpl --fast -o vgg test.chpl -M ../../lib /Users/iainmoncrief/Documents/Github/ChAI/bridge/include/bridge.h /Users/iainmoncrief/Documents/Github/ChAI/build/CMakeFiles/bridge.dir/bridge/lib/bridge.cpp.o -L /Users/iainmoncrief/Documents/Github/ChAI/libtorch/lib -ltorch -ltorch_cpu -lc10 -ltorch_global_deps --ldflags "-Wl,-rpath,/Users/iainmoncrief/Documents/Github/ChAI/libtorch/lib" @@ -241,15 +270,7 @@ target_link_options(TinyLayerTest -set(CHAI_LINKER_ARGS - -M ${PROJECT_ROOT_DIR}/lib - ${BRIDGE_DIR}/include/bridge.h - ${BRIDGE_OBJECT_FILES} - -L ${LIBTORCH_DIR}/lib - ${LIBTORCH_LIBS_LINKER_ARGS} - --ccflags "-I${BRIDGE_DIR}/include -L${PROJECT_ROOT_DIR}/build" - --ldflags "-L${PROJECT_ROOT_DIR}/build -Wl,-rpath,${LIBTORCH_DIR}/lib" -) + diff --git a/bridge/include/bridge.h b/bridge/include/bridge.h index 8b42b7f1..7ac0e956 100644 --- a/bridge/include/bridge.h +++ b/bridge/include/bridge.h @@ -20,12 +20,14 @@ typedef unsigned long long uint64_t; void debug_cpu_only_mode(bool_t mode); typedef struct bridge_tensor_t { - float* data; - int* sizes; - int dim; + float32_t* data; + uint32_t* sizes; + uint32_t dim; bool_t created_by_c; + bool_t was_freed; } bridge_tensor_t; +void free_bridge_tensor(bridge_tensor_t bt); typedef struct bridge_pt_model_t { void* pt_module; diff --git a/bridge/lib/bridge.cpp b/bridge/lib/bridge.cpp index 8c69d330..e94dd741 100644 --- a/bridge/lib/bridge.cpp +++ b/bridge/lib/bridge.cpp @@ -14,9 +14,11 @@ #include #include #include -#include #include #include +#include +#include +#include @@ -65,6 +67,8 @@ extern "C" void debug_cpu_only_mode(bool_t mode) { } else { best_device = get_best_device(); } + std::cout << "Debug CPU only mode: " << (debug_cpu_only ? "ON" : "OFF") << std::endl; + std::cout.flush(); } extern "C" bool_t accelerator_available() { @@ -81,21 +85,10 @@ torch::ScalarType get_best_dtype() { } } -int bridge_tensor_elements(bridge_tensor_t &bt) { - int size = 1; - for (int i = 0; i < bt.dim; ++i) { - size *= bt.sizes[i]; - } - return size; -} - -size_t bridge_tensor_size(bridge_tensor_t &bt) { - return sizeof(float32_t) * bridge_tensor_elements(bt); -} void store_tensor(at::Tensor &input, float32_t* dest) { - float32_t * data = input.data_ptr(); - size_t bytes_size = sizeof(float32_t) * input.numel(); + const float32_t * data = input.const_data_ptr(); + std::size_t bytes_size = sizeof(float32_t) * input.numel(); // std::memmove(dest,data,bytes_size); std::memcpy(dest,data,bytes_size); } @@ -103,16 +96,40 @@ void store_tensor(at::Tensor &input, float32_t* dest) { bridge_tensor_t torch_to_bridge(at::Tensor &tensor) { bridge_tensor_t result; result.created_by_c = true; + result.was_freed = false; + result.dim = tensor.dim(); - result.sizes = new int32_t[result.dim]; - for (int i = 0; i < result.dim; ++i) { - result.sizes[i] = tensor.size(i); + + std::size_t sizes_bytes = sizeof(uint32_t) * result.dim; + result.sizes = static_cast(malloc(sizes_bytes)); + for (uint32_t i = 0; i < result.dim; ++i) { + result.sizes[i] = static_cast(tensor.size(i)); } - result.data = new float32_t[bridge_tensor_elements(result)]; + + std::size_t data_bytes = sizeof(float32_t) * tensor.numel(); + result.data = static_cast(malloc(data_bytes)); store_tensor(tensor, result.data); return result; } +extern "C" void free_bridge_tensor(bridge_tensor_t bt) { + if (bt.created_by_c && !bt.was_freed) { + free(bt.sizes); + free(bt.data); + return; + } else if (!bt.created_by_c) { + std::cerr << "Warning: Attempting to free a tensor not created by C code." << std::endl; + std::cerr.flush(); + } else if (bt.was_freed) { + std::cerr << "Warning: Attempting to free a tensor that has already been freed." << std::endl; + std::cerr.flush(); + } else { + std::cerr << "Warning: Attempting to free a tensor with an unknown state." << std::endl; + std::cerr.flush(); + } +} + + at::Tensor bridge_to_torch(bridge_tensor_t &bt) { std::vector sizes_vec(bt.sizes, bt.sizes + bt.dim); auto shape = torch::IntArrayRef(sizes_vec); @@ -129,7 +146,6 @@ at::Tensor bridge_to_torch(bridge_tensor_t &bt,torch::Device device, bool copy,t return t.to(device, dtype, /*non_blocking=*/false, /*copy=*/true); else return t.to(device, dtype, /*non_blocking=*/false, /*copy=*/false); - } extern "C" float32_t* unsafe(const float32_t* arr) { @@ -229,8 +245,8 @@ extern "C" bridge_pt_model_t load_model(const uint8_t* model_path) { -bridge_tensor_t model_forward(bridge_pt_model_t model, bridge_tensor_t input, bool is_vgg_based_model) { - auto tn_mps = bridge_to_torch(input,best_device,true,best_dtype); +extern "C" bridge_tensor_t model_forward(bridge_pt_model_t model, bridge_tensor_t input) { + auto tn_mps = bridge_to_torch(input,best_device,false,best_dtype); // tn_mps = tn_mps.permute({2, 0, 1}).contiguous(); // tn_mps.unsqueeze_(0);//.contiguous(); auto tn = tn_mps.permute({2, 0, 1}).unsqueeze(0).contiguous(); @@ -243,24 +259,12 @@ bridge_tensor_t model_forward(bridge_pt_model_t model, bridge_tensor_t input, bo // auto tn_out = o.squeeze(0).permute({1, 2, 0}).contiguous(); auto tn_out = o.squeeze(0).contiguous().permute({1, 2, 0}).contiguous(); - if (is_vgg_based_model) { - tn_out.div_(255.0); - } - - auto tn_out_cpu = tn_out.to(torch::kCPU,torch::kFloat32,false,true); + auto tn_out_cpu = tn_out.to(torch::kCPU,torch::kFloat32,false,false); return torch_to_bridge(tn_out_cpu); } -extern "C" bridge_tensor_t model_forward(bridge_pt_model_t model, bridge_tensor_t input) { - return model_forward(model, input, false); -} - -extern "C" bridge_tensor_t model_forward_style_transfer(bridge_pt_model_t model, bridge_tensor_t input) { - return model_forward(model, input, true); -} - // std::tuple get_cpu_frame_size(uint64_t width, uint64_t height, float32_t scale_factor) { // // if (best_device == torch::kMPS || best_device == torch::kCUDA) // if (accelerator_available()) diff --git a/demos/CMakeLists.txt b/demos/CMakeLists.txt index 95ff2ed5..419523bf 100644 --- a/demos/CMakeLists.txt +++ b/demos/CMakeLists.txt @@ -3,4 +3,5 @@ add_subdirectory(video) # add_subdirectory(webcam_filter) # add_subdirectory(torchtest) -add_subdirectory(torchtest_bridge) \ No newline at end of file +add_subdirectory(torchtest_bridge) + diff --git a/demos/video/CMakeLists.txt b/demos/video/CMakeLists.txt index 6c8ac78f..78584fe2 100644 --- a/demos/video/CMakeLists.txt +++ b/demos/video/CMakeLists.txt @@ -99,4 +99,6 @@ add_custom_command( "${CMAKE_CURRENT_SOURCE_DIR}/style-transfer/models" "$/style-transfer/models" COMMENT "NOT! Copying ${PROJECT_ROOT_DIR}/examples/vgg/images to $/images" -) \ No newline at end of file +) + +add_subdirectory(cpp-model-construction) \ No newline at end of file diff --git a/demos/video/chapel-webcam/incomplete_sunday_afternoon.model b/demos/video/chapel-webcam/incomplete_sunday_afternoon.model new file mode 100644 index 00000000..2576c9aa Binary files /dev/null and b/demos/video/chapel-webcam/incomplete_sunday_afternoon.model differ diff --git a/demos/video/chapel-webcam/lib/smol.h b/demos/video/chapel-webcam/lib/smol.h index c02e605f..c212905c 100644 --- a/demos/video/chapel-webcam/lib/smol.h +++ b/demos/video/chapel-webcam/lib/smol.h @@ -31,11 +31,8 @@ void chpl__init_ndarrayRandom(int64_t _ln, int32_t _fn); void chpl__init_smol(int64_t _ln, int32_t _fn); -chpl_bool acceleratorAvailable(void); int64_t getScaledFrameWidth(int64_t width); int64_t getScaledFrameHeight(int64_t height); -int64_t square(int64_t x); -void printArray(chpl_external_array * a); void globalLoadModel(void); chpl_external_array getNewFrame(chpl_external_array * frame, int64_t height, diff --git a/demos/video/chapel-webcam/main.cpp b/demos/video/chapel-webcam/main.cpp index e2f68c18..eb2cbf99 100644 --- a/demos/video/chapel-webcam/main.cpp +++ b/demos/video/chapel-webcam/main.cpp @@ -3,41 +3,37 @@ #include #include +#include +#include +#include +#include -cv::Mat new_frame(cv::Mat &frame) { - // cv::Mat rgb_uchar_frame; - // cv::cvtColor(frame, rgb_uchar_frame, cv::COLOR_BGR2RGB); - - // cv::Mat rgb_float_frame; - // rgb_uchar_frame.convertTo(rgb_float_frame, CV_32FC3, 1.0f/255.0f); - +std::tuple new_frame(cv::Mat &frame) { cv::Mat rgb_float_frame; cv::cvtColor(frame, rgb_float_frame, cv::COLOR_BGR2RGB); rgb_float_frame.convertTo(rgb_float_frame, CV_32FC3, 1.0f/255.0f); - // cv::MatSize size = rgb_frame.size; - // std::cout << "x " << size[0] << " y " << size[1] << " channels " << rgb_frame.dims << std::endl; int64_t height = rgb_float_frame.rows; int64_t width = rgb_float_frame.cols; int64_t channels = rgb_float_frame.channels(); int64_t pixels = rgb_float_frame.total(); int64_t size = pixels * channels; - // std::cout << "Width: " << width << ", Height: " << height << ", Channels: " << channels << ", Size: " << size << std::endl; - chpl_external_array rgb_float_frame_data_ptr = chpl_make_external_array_ptr(rgb_float_frame.data,size); + std::size_t chpl_start = cv::getTickCount(); + chpl_external_array rgb_float_output_frame_array = getNewFrame(&rgb_float_frame_data_ptr, height, width, channels); + std::size_t chpl_end = cv::getTickCount(); + chpl_free_external_array(rgb_float_frame_data_ptr); - // cv::Mat new_rgb_frame(height, width, CV_8UC3,new_frame_array.elts); - // cv::cvtColor(new_rgb_frame, new_rgb_frame, cv::COLOR_RGB2BGR); cv::Mat output_frame(height,width,CV_32FC3,rgb_float_output_frame_array.elts); // frame to write to output_frame.convertTo(output_frame, CV_8UC3, 255.0f); @@ -45,61 +41,10 @@ cv::Mat new_frame(cv::Mat &frame) { chpl_free_external_array(rgb_float_output_frame_array); - return output_frame; - - // cv::Mat rgb_float_output_frame(height,width,CV_32FC3,rgb_float_output_frame_array.elts); // frame to write to - - // cv::Mat rgb_uchar_output_frame; - // rgb_float_output_frame.convertTo(rgb_uchar_output_frame, CV_8UC3, 255.0f); - - // cv::Mat bgr_uchar_output_frame; - // cv::cvtColor(rgb_uchar_output_frame, bgr_uchar_output_frame, cv::COLOR_RGB2BGR); - - // return bgr_uchar_output_frame; -} - -/* -cv::Mat new_frame(cv::Mat &frame) { - cv::Mat rgb_frame; - cv::cvtColor(frame, rgb_frame, cv::COLOR_BGR2RGB); - std::cout << "Frame size: " << rgb_frame.size() << std::endl; - - // cv::MatSize size = rgb_frame.size; - // std::cout << "x " << size[0] << " y " << size[1] << " channels " << rgb_frame.dims << std::endl; - int64_t width = rgb_frame.cols; - int64_t height = rgb_frame.rows; - int64_t channels = rgb_frame.channels(); - int64_t pixels = rgb_frame.total(); - int64_t size = pixels * channels; - - chpl_external_array frame_data_ptr = chpl_make_external_array_ptr(rgb_frame.data, size); - chpl_external_array new_frame_array = getNewFrame(&frame_data_ptr, width, height, channels); - chpl_free_external_array(frame_data_ptr); + std::tuple ouput_package = {output_frame,chpl_end - chpl_start}; + return ouput_package; - cv::Mat new_rgb_frame(height, width, CV_8UC3,new_frame_array.elts); - cv::cvtColor(new_rgb_frame, new_rgb_frame, cv::COLOR_RGB2BGR); - - return new_rgb_frame; } -*/ - -/* -void new_frame(cv::Mat &frame) { - cv::Mat rgb_frame; - cv::cvtColor(frame, rgb_frame, cv::COLOR_BGR2RGB); - std::cout << "Frame size: " << rgb_frame.size() << std::endl; - - // cv::MatSize size = rgb_frame.size; - // std::cout << "x " << size[0] << " y " << size[1] << " channels " << rgb_frame.dims << std::endl; - int64_t width = rgb_frame.cols; - int64_t height = rgb_frame.rows; - int64_t channels = rgb_frame.channels(); - int64_t size = rgb_frame.total() * channels; - - std::cout << "Width: " << width << ", Height: " << height << ", Channels: " << channels << ", Size: " << size << std::endl; - - // chpl_external_array frame_data_ptr = chpl_make_external_array_ptr(rgb_frame.data, ); -}*/ int mirror() { @@ -116,9 +61,12 @@ int mirror() { cv::Size original_frame_size; cv::Size processed_frame_size; + double last_frame_fps = 0.0; + double last_chpl_fps = 0.0; + while (true) { - uint64_t start = cv::getTickCount(); + std::size_t frame_start = cv::getTickCount(); // Capture a new frame from webcam cap >> frame; @@ -137,7 +85,9 @@ int mirror() { // std::cout << "Frame size: " << frame.size() << std::endl; // std::cout << "New frame size: " << processed_frame_size << std::endl; - cv::Mat next_frame = new_frame(frame); + cv::Mat next_frame; + std::size_t chpl_delta; + std::tie(next_frame,chpl_delta) = new_frame(frame); cv::resize(next_frame, next_frame, original_frame_size); @@ -150,10 +100,19 @@ int mirror() { break; } - double fps = cv::getTickFrequency() / (cv::getTickCount() - start); - std::cout << "\rcv::FPS : " << fps << "\t\r" << std::flush; + std::size_t frame_end = cv::getTickCount(); + + double frame_fps = cv::getTickFrequency() / (frame_end - frame_start); + double chpl_fps = cv::getTickFrequency() / chpl_delta; + std::cout << "\rcv::FPS : \t " << frame_fps << " chpl::FPS : \t " << chpl_fps << "\t\r" << std::flush; + last_frame_fps = frame_fps; + last_chpl_fps = chpl_fps; } + std::cout << "\nLast FPS: " << last_frame_fps << std::endl; + std::cout << "Last CHPL FPS: " << last_chpl_fps << std::endl; + std::cout << "Exiting webcam feed..." << std::endl; + // Release the camera and destroy all windows cap.release(); cv::destroyAllWindows(); @@ -167,24 +126,15 @@ int main(int argc, char* argv[]) { chpl__init_Bridge(0, 0); chpl__init_smol(0, 0); - square(3); - - // int64_t array[4] = {1,2,3,4}; - // chpl_external_array array_ptr = chpl_make_external_array_ptr(&array,4); - // int64_t sum = sumArray(&array_ptr); - // chpl_free_external_array(array_ptr); - // printf("sum: %d\n", sum); - - - int64_t matrix[2][3] = { {1, 4, 2}, {3, 6, 8} }; - chpl_external_array matrix_ptr = chpl_make_external_array_ptr(matrix, 3 * 2); - printArray(&matrix_ptr); - chpl_free_external_array(matrix_ptr); - globalLoadModel(); int code = mirror(); + // std::size_t start = cv::getTickCount(); + // std::this_thread::sleep_for(std::chrono::milliseconds(2000)); + // std::size_t end = cv::getTickCount(); + // std::cout << "Total time taken: " << (end - start) / cv::getTickFrequency() << " seconds" << std::endl; + chpl_library_finalize(); return code; diff --git a/demos/video/chapel-webcam/model2.ipynb b/demos/video/chapel-webcam/model2.ipynb index 09366591..3cabe64c 100644 --- a/demos/video/chapel-webcam/model2.ipynb +++ b/demos/video/chapel-webcam/model2.ipynb @@ -228,6 +228,70 @@ "execution_count": null, "id": "d526ae6b", "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/var/folders/jy/f3n1221n2pq22hjdxzygxggh0000gn/T/ipykernel_83750/3818043664.py:6: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.\n", + " state_dict = torch.load('incomplete_sunday_afternoon.model', map_location=device)\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "torch.Size([3, 1080, 1920])\n" + ] + }, + { + "name": "stderr", + "output_type": "stream", + "text": [ + "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers). Got range [-0.29142156..1.7656863].\n" + ] + }, + { + "data": { + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgMAAAErCAYAAABDzICRAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjEwLjEsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvc2/+5QAAAAlwSFlzAAAPYQAAD2EBqD+naQABAABJREFUeJzs/cmPZVu35Qn95lzFLs45Vnhx/d77Va+KiO9FFkG+TAilkBIQbf4d2vwvNOkg2jSQEB0QmRABGRGv/KpbemFmp9h7r3LS2Dczuw/pk7LxfEgu75i7zI7tvdZYY44xlpiZ8Rmf8Rmf8Rmf8Rn/ZKH/Y38Dn/EZn/EZn/EZn/E/Lj6Tgc/4jM/4jM/4jH/i+EwGPuMzPuMzPuMz/onjMxn4jM/4jM/4jM/4J47PZOAzPuMzPuMzPuOfOD6Tgc/4jM/4jM/4jH/i+EwGPuMzPuMzPuMz/onjMxn4jM/4jM/4jM/4J47PZOAzPuMzPuMzPuOfOPw/9gv/D/+b/x3djNFHtHvMVuajkpshURDXWVOiNqVsRl9W7qJyp8pX44GwKGPvWGuMYSY3j/ZEi4npOLEk4/xyZWVjmBW60kywwSMxks4d2zKHyVCBbgdyV8wX/NSwLqwYN2cs1lEXuJ1XkIIE0DBis2cTx9VBjgE3DSCGE2OMnTc07i839LwRMqTV6EVxUYiTpzRYTchaMQ+pQK0OqR7fG8GM0XkiHksrg0DsEXEj21qpMWHe4yZYUmHVjJ8jxymiqqS10VtDgzEfAsWUpcPjYST2ylYKW2tUaTAbogFlIrAyRuNZGh/mB75PI2sUGAaKRDQEXMp8/TrSY+DffHdhrcY9lb+4O/Of/7MFJ4Xnl4Ftc+jVIT++MFwq8+oI54BkOAaj50JTR9GKxE7zgjCQrxXpCapymh/QK+R6ZQqBbYXNKtFHvBhyA4sFNwfW0rm0zO2QKPce//YVr9+9ZpWNZTB89FA7rIH87crw/oXpssLWiXXCFQ/ViCjFlBwa15ZgDDyVyjZm8tgZ39zTVVlrJpVGvW34XBlQXk8zr6MQbwnvhWEYqLfOOW+4MbAsBScR14Sxe6IpEgtdOtUr72viWTPJO9wXnl//p18gCfrLla3e+LaN/C4ZCvQQuP9qYpiFtwfhr76I1Msz/6e/C/zbP3jeusLPHwK1F57PK4WB+dWR6BtWVuJ4ZLXCzw6Z/2zOuBz42IX3LXDzA6fDHbMD1xKH5vDLSlpesGrIrUHeaHSSOJbkuN2MvhqNTjxEhuDwAq4VYhWGLPjFcLdK9MK6QjVh6UKiUZuBeZoKRY2klXD0fEiVSwA5Tvh5gGgMU0S8oIPQ3f58FzVsFFTBNU9NC90JqxuJ/oV/+UZI5w23Cgfr+Jw4VSU/ZZqDOAaKOfoQaINDppFknqUI63UjrIn768rxWqBVhsFjF2FLjSmMtOQouRG9w4qnN6XVhTF26I5aFFsFpTAHR7ARCzO1FNqw0H0m+86H0rk6pXahkrk7OMq1kEon+hHbKq/ikZAHtBsR4zQOyAq9NXxo+OipyZMwki6YCKbCapkyFJZWWZxnUaUHQ6nQCnP0tOyprXEcR+wG0jIuGIdpZC2VhYweHRuepJ7WKm0tuFI5hgFvDnJHamMeGn41fFEGZmbzHFTQGvDTSG2FLVzJVlla5WXLEByDj8RqzDYwuxFXBK+dKQZaE7bcWWTDx45ZYdNCHhoSHGsy1ubAK96EsSsuK7ELE56TO6J4rC9sveKPgSVXPsoV9+D52BrPoRDezGQV+uigVOyW0FvBt4Ya6FIJXZhlwFePdJjnQC2dYpDomDSGQZBiiCiNQO1G6Z0YjVoLW0kcDkdKgiyN+eQp1sihET3ctgu3+5mLgUuVkBs+Kcc88Lg6HpLnFAN3cSRuSq6ZGhMLmRY7N4E8CC44rrdEap3UCvPdSEZY24of3P6MOmE4eNrWuLXMNp74tk3cdCZHg9T53/8f/7d/PDKwiOHiQOoV0co4QRkcKw2VSvAOf7xHuzGWQn5OIAPmJj70xjxvVDXicMflmvGS8FLR08wPLSMjtCMEPZC2SioZ7xX1SuqZdqyYXGhhQtxEk061DTcJEj25Q6XRhoGunmVZ6XcN9Y4435OKkYOQvWJDpDOxNQdOUevUlnn12PG2chxnKIKvgZePz5gGihfKpKw9kQ26BYo4mhfalpi8gzCTcsPXlXFuaDywZrCeaK4QDxMZx6aFPnmKOixOpAB127Cx4xzoMPPUKsUHLHj6MFDrDWLA4ai9kpynxhnvR17r/mBVMT51x3l95nD8AsI9161g2ZicEcKRtSoijdaMrW4QPSFGlqy04UBNCw/a0MmhPYIqqXdGPLk32gEKGZzh/EQVR8MgdJwdcESua8HfCSHe8+l6pvqGuojhKC3jJ3DDxLVV6tFTHeg8smrncBzRozLFOy61c86FVhsBZTwdKC/PxElAPS9LQS1xvL/jclmQUikN6hj5RCEfDGZHj5GPqSKDR6cRmQRRxbYCeNroaKGhg+KJNBG26NlW6M7BfGDdKqE1atk4eEfwke4dSSqdQPAH1ubQCUwak3qSKRqPFO659YHrrRGd5+RmtmxcbKGflMkOhDAhc6TpxnM1rsvAeTuBi6QlcheV1hO3OFFaBnfGsxKqcnIHklN0cNzdT2hpxBaIXbCcOR4fSWkDX2Az2BKTRSwE2slz8xvShBoiOg9QE90J+ABOueUXDmOiu8gSA9dSWNOK+plSha0mOg0Lis5HXlrj7CNlnqh+YMvG8eCxEHDuSDFHVUdVo1hFmxLUEXpFVPDOM/TI6G787JVhw4ZumWMH3Yy0zLQwoiVSDPR4RzWPH0duPfNxW1laZ5wCD6PjMFbCnHF2ZL0W+mzoNHHeGi12fFTWbjTX6OtGjI3shD44kutoWwjWsPlA65laFmQWwhS51M7qK1v0LFTMFO9HPuULGgScY62Z8TTxfdkYYmHIhTlMXGwFb0hdmNxM60pWR60bzXcseLZSYRAYHOcqrNpp0dPahted3HxcNrwfcIeB93khTkI0iMPIOd+wCD0qWTrNd4o0imV0aIyjZzGjtoLVxDw5mglucAiKTwuHbpQ401xhWz8hA8QxcquZl56wB08uGVdXTuOBW92QuhDVOM0Har1RqFRX8eNAo1PZqD0hIbJJYwlGc50QB5zzPK0boRbu4kwLM5ftE6F1XBB6iFzzFRuUfAz8mF7oY8QOM5sN4Gc26+ASfgKVQM2NumZ8FKoqqTW0r8w+UMVRB2OxSiXhHJQwYRGWnGn1Shg8TjyfbjfC6OEQeV5fiNExjAd+qDeI4LxjywmnSqnKGiMyVEarDKJce8fFgkfR8UDVBbEGsmDRs0rjZokyRdo8sqWN7jckDmg8cKZTteOGiaU3eumoG7itHVWPzjNLUhY5sLoja2tIrf+oPf4fTQa2XwuVGz1kumu4CHhHFgiqeKeoT0QnDFIZ3ICp8FS+J9jClYpzgshvGYIRnKd3T+r7/6fOo+JwGhEzWlZSLnQyGhqDj2hVLsuN69IpBMRFzHm2CltxKBG8sOmGtgXXM+oCH8uPtF5QN1F0ojkwN1FrIHWl9EI3sEPl9S8813ah1Iyp0ruwtYVGARHURXodWLMh0jCrqAlZYLEbrhUOFLJ13ucXei9Yh46CD9QKKiNNZ/r4wOA8MX4g9AxeaC7y3FbWqtx6Y5wG7vxGsIXCrkj4YeLpsvFxWXk9HnntEkkc1Qe+jhO/snt6euJpfUZ/9ooYO9Gfie0Dny6N4bXnuRwoLfL2NUwHzw9rJbsbX/688Ngat5x5um183CrNBl7FESlneruBhyoedKJ3QQDtnmpGqRvegC6kXMil4MKIoCgZ1yrqAuYcRRpmI+pPXE1ZLLDoyusvA00HSg5sa+BaDVxlfpt482Vla5lbutK6YBjfpffUpSCbkTbH1jpyOuCmSHeOTQ3vPC0ra490F9GvAkKgTcKruwvE837SRdia8ZQda40kc3Q6aiv3uoCvaAh8fy68nCPGibVPvFwiVmfi0XiaHCkamYkn7/j99ornFshu4e27NyxSuawvBNeARtHEt3XhW/+KOk+0Hvj4lOk6EXxAG+jaebkV6iJIiPxydqwkau+kZqhT5jAj3ljqyspGP4z4qVDZgE5NC1orbclI8+S1cU7KykiRgeKE1w8RHzr4SuFGb2esnNlK5pw7C5GlRXKG5XbGaieXDuLobmJpldzuSfVA04msIxlHngv14Om1M/kBFaBWyI2tNjgaf/KF48E5rBkfls501/izX3yLtn+g1kSumVRH0suvePnhjvKU2FpA3jxw/+aE6ZnQn3hdC18MMI+V2M4M5UJvlVQiZZso60h9WpGtkruQunJNG70KrnkcUKj7BkbBW0VaobNhYogL+GEmZcdWHN2NdARxiuAo9RknhoojlwSqIGdoBdcrrjeEhnaDJggGJvsf52nSdjJgHedHzAVqU7p5mggVI/gJddDrikfo0mh6xlrDmxE0As+oQseoOMSPNEnk3rFecSbcesfEMKt4gZsFdAW/CZYaKp3gOp3fU/qGGz06BLoodBD1pAr7FTeC2QdEK05A1VAPSKfXhogAQtkarg2ITDQfWVC69/gA4hboK70WIjD6AakbQQ1ngomQ8066WhhILoIf8YMg/cLBQcPRfKD4SO+Gu2303MjVsO5AC6qJyRcGXbD0HsNowSFBEPHkdiG3SqMj3XAItITiER+p2hAVaAb9CXWKRCUSGG6KT8Y2Nl5GpWSYxRj6C4NtvGw3vu+7EkSvjBiDU0QcguCnwBoPfHe5YuKZHBycsSzfU53sh9kGzo+oBlrJtDaQUkTOhZYHqjuwxXtuLRP0j0wG8mvPJo1VPT04xBdcMFDhLhgaoIpQrFOdcLHOGDp3UyXzkeISzq+YbpylgIDTAW0ZRBE/MQaHk4T2TmvK6DpIolijtIDUgO8Tcz6S+x2X1NlSJiXPtgVCuCO3xHMGTFAJOO1EClGN1JVbj/tGbCO1TqSmbM1hoszekIdGcJ01F57rjQQYjcF3MMGHSNAHajsiHVrzdPW8rIlzarx2jccx0Wx/6VMtdFGKQe2JvgakRVYmLhq5I/HnbxJjvmLekSZj6UdyGjCZ0Bi5v7thbcUHo7kE8p6fiePnvObr1rl7eqKtIyqOo3zPfVpwpVJOryinD4z3K86e0FrRuwNd3vDvnw78f771zNqZw8JfvG10WYmysF0yt1A4T/A0BLJsHI4/chdf6LJQXSFYRPsjrd6RqycjpFaoLeNJOOmIKjkp55ShFuawoTTMRfCeWj25GbkquR+oDeZmcIJXo6PdPD1Aap2NjNfCeHgmkDhbZdXCuj1zTcaaPHZtBB0Z/YRoplhiDJ436nBb59M5kNvEzUeSd/TBcx8Sv/7Zwhy/RXxB1bHUmT/8eOTcJhzGFBsPYaPpB1YuVJ8Zv5j55vdf8u36c67pDc0GYo8Eb5w2eK2Z1M68Lwe+vwVSvfKrtzd+8ZCpeeVxOPNXXznexQy18l+9O/G331/5dsmkeM84HelVSD3z52/e8fz9j1hSWgE5dI6jUrVSNbIVT+sVV1a2y4WXXHA0cBceDp0pVKJLXNdEPi8MoeFJJPF0O3HxAx/ChGnn7r4xPbxnmj4Qw43en3m+PtFpUI1W3/Cy3LPVgb4WwlYYS+fWBq4Ilz7wtDqsP9D6gV6GfSOwxGCJ+0n4ecg8UimXT2ypcmsj6034cvZ8MTxTk6KckOGFw/h7WvkbCAudStRXdDnAd4HJoLfKh8sLr/9EuZ+/Y9AfCC7TbaWz0vIF+orzBozk9Q3L8obh3uHWG6VHXjbHtXSsCV4qKa901+lSaFLQutHzBeRK9x40UovjVu4we0PUyKCekjc0ZLpkRBShYWL03igt/3SaL4R2o7VMFU9lolvDSgF2lRMKqgUTQd2IcweUASeeYNCaY+1G14CJ4kSACpIRDG+Nwa+YCB0B8Yjq/nOppzXDY7haiD6ylkLXjgnQMnIEtykewaSS8wumNyQUNBoaPeomgj/QS4cqtN6oFJBEJ2FaEOkQoPeG0JHe8a3jKrTywFUeKMw0GTE/Qb8x+hfmsVHqgnYY7IyPGXONYg3rHWue0iLmH1n6SGHG64G/vB8ZLs9QA+YDqw9cRZHWkOj4WCtXc5gURAciL7wLZ958kcAFVlEcBdrK2pVLd2y94Q2kVEbfkV5RbRA7TUFwWAejE7ww6Uy5BA6bA1W6FPJhJIlj7iuTLVA+0TXT1CjdaH6G4vhqCvzpceQhLlwunW/sHVebOfnMu2nh+BbW1FhqJsqE6kjtio5w2xrVCeNTJbYATWghYhoxtj8uGbB5pjNz7kb2wjBU7mbwlogh410HHbmkAipsFjmXzOlhYQwd5AUfb9R+o0qnkRHncShDGJF+I7qElw0nlQBgmS6F3g3KQHcnajlRncPshFkjusbDo+MuZh6GH2l95Lc/DvzuOXDODecWnFwJPlLKK55uR7IcaRzo+N1CqUqzzlorQWBujU12Nv5UOtMsHAahV6itUEhcy0DOgvWGeeXqJi6u8fMjvJ4bqQhbONBTIXVjyYnb1piHE6Ec2fqR1geih3vtDLVRcqU2wc13qB/pVbBWcbExeWi+UPqFyjOZRi037sdfcu90nzvi8KXi1xdcAdsci5s4vO5Y/oCTRqSSNyO2xnG8Z4ow2Q88xDOtFbZcKC4SwkTDyOlGmODh8UrkI8USnRsaRiaEcp1Rf2JpK4TCQOOojV4TuSvaZ16uGVrl8GC7gqQVU09tActKaCtSVpyNCEc+5oXH08Tj48Q6FJ4XxXrg3gUeJWDSubUCfqX0JzabWRlRCaj43YchAZ8SdyyMpYAOLDpiEmn+QIkDq/dM4cLDXefdVOktYd5xPc98Wkc+cWAePDFkTneeIUHJDWzFhcAYAtv5jpf0iJhQnDC6xqecUL9wOkb0euNPHz/xr/4i8av733CIHxESoxdOeUB+jLAW/urwwMs/+wv+b99NfFcKw6ycbCZIZ54y33Oj4WnW8WXhfhbCWli2Qg9H1rpRsnHzmRcGRAZKS/h6YwoF1y/4nLC8MDrBiIQx0jb4lOEP1TjORnDPfHH8hjj+FtwLuW1sreDoRPGUJdC2AwXjOC4MNIKfWdOBazvxrHfchjtandASCS5wPwQedOHAmccQ+NqN3J9fWD59w5qMYHdYmlhflOHdQpTOTMPchnOJWjeQFVGjlAvl/JHj+YCtSmLCthujCMfxCScJ6zfMFlQrqoWaL5jL+/hBNlLdOMQ73L1xvcGZe56DJ6pw6CuBQLH9wGI544ZEuA+UEujOkbLRQ2DLB7TPnGzA5YrhKc6osyAUpHeCelpKEBX6rjloSkyTcmvGkw5ca8NFJUpDbcO5QvQN1bhL+9YILjB02X0CtdJQrrqPKsU6HsG7QC+F2YRphKZKxpGa7sRAA7lBpxG7MTtjco7BHOeeuaL04JgGYZiVViu9JzhFWi+IdkLcxxdVJjIzZkbw+4Gp0TDpuyKgIGq03ih+ooiipeKqEUyQNvBxPVDkC0SONAbMNSLf8256z1J3gvRucgRZyCTWLqQWGeIbSj2Q2sC1PZD5EurEpW58NShGpzhFVSjVkfo91n/yFHlPkQTeeKp3LPk7vvxZYyCTbx4hUd1KcoGrzbyUirOGaxt3DqbJ00zxoyf3CtKgZ0Qd1UVSV6a7EaPymsJ9z3zjOr9tgWv3vPE/jXR7Ig+Rc3Ekd8dGwJnwnx1nvtKN79Z7PshbbnYAeebV1NBjIBb4+Clj/hVaR3IxqlTy5IjjiLqFkYpQ2cxI6piG4Y9LBlTv6dVR3UgiIlbormL1Be8yk+9UN7HVzHw4UHNg21YEZZQfGY/3YI2Oo9lCExAfCXhGJzitKI1eNpx25Cdjn5mhVtEh0otD9A6pjyzrkWYR80Jl5W668hCfwDqPf/KKv2yvWJYbh+lCdBe6Hfib7zrnv45cu0HcqL3izNG109zC0RXu/IHxvHHeEmN8g1NjCp0v7gptu1Ixtt7IpZPbka0JvTuSCmNU7qPw4ISn28ZzGSniWFJma9DjSCwDh7XTJbLieBgCDyoEHEkDhYmjKVtXOgGtDTXHIXjMF1J1LD2Qe6cZ0DvSB1zauLOGjkeg0ssL0SovP1bWaeBnX3+FLxu2eISBwQvHyTHExsEbXhq1Ks0cjs4glS+OnZ99WXEh8Wg3rEI2sFbovRF1YXIBK4oB0zgxAFMd2WoiaaT5idA2NCgPRyUOoA68V9Ky4PjI6DaCE9YaOK9fQhoRApSO9MIQFZPAl6Pjze2Omm90B8+aifcnLi+eLo+sL5AZWJry9jFyGo1hrYwS0GniUE+EfCDGA206UN2Ik85BEpM2OpGOZ/YnjuHA7bry5iR8edo46EcOh04fB7xrDBLgjbHm95zcSu+Or9++4TS+8O7wAz97WAiyYj0xxGcO8Q8E+UCXDXWRMdzTrgO+fIWY59Cf+YuHK9+0n/P9x865FGK+8uUp4PqNfEssfSIPjlkMaZmAJ93gB124Fkhy4ImB96lhrvN6MJ5uSvjyxDBuCFfiBFsd+bQ+ci4Tejjwqg+4duGff3Xjf/5nP3KI37G0J5JtmBeOo6fXjKrAUXlT4cM5c9SNhzCRLq+w9I6t37PZA00GGsI4VP70MfAzO3PKLxidmjoMHamZkAuNiA8elYHvf7jwz38hBCnMZSVrQZzhvFIb1FpJRWDr3DlPFiGLElph1szkKqaRUhvWE2bXXSIdPNV2QqHWmF3kzo84g2Qj43lE7ER0jrt+gb5QfWfrCz06ggSURFPHrUaWpJQ24OURqY9InZG8Er3t3qXi8K5BSsQIrq4UMapmtI04i4zVYaaIPvChNQYv+HxlZOGk+wZTG9R4z8KBtjq+HGasXBAz0IEuM7cKPgRoBSkdtUZoiWiGmWM15dyE1BR1cff3eKHUytGMeIEheZybWIPnHD1Z4U4bQTfMOsiAxhOqBbT+ZKY70tqBQMBqw8iYjzgf8Fpw0nACC56VO25bRxqoOU5eGN3AsnxF6V8x6IHWjNoazQ+c6x3XPDP1lVErQzjylDPfLY1mEw+84zD/knVzbCnQOeArQMQPmYLSfOCSAh8uDuyE0xETh1BxWsk9Uzmy4dnyC6/GzFpGslWQlSKeax9IdFq9EXWipw3nJ8wGanFkCupBemI+HllLoSGk7nj7+p7Xzx95ZRuPVlnkLf8unyjmuKYTMSqXa6WGA8k/8NQ7d7byMASO2wfsNmP6CicHUo/8/XPkuwRbMvraeDMdeJhOlF64lRV1E57I168Lr92V23LmN/bAogOPo/xxyYCzE2oD2me8n9CeyMuN7gwhMShQlaM/MLkDxTvMDdRq4J+ZNTJNjpRfQOM+p8IR1OH6AlZx3tElAgWnEFXoXRldBBuR+MDL9Q1BfkaQR+rg6K7jWHE9IdloDYq9cJwaX729Qf0OdQXz9zjeEtcjp/vG21MnpQuXrfApNc4ycNSJOxXGfsfNVi4tcNTIvav84njBH1d8vCG+clsjf/u7gffrHTI4ZDS8j3wt8EUcCaXy6Wlla0q2iYYiOmCp02unaQcBL4JXI0yKH05sxUg07ikMOMQrgX1BslZQOWEqFFa6Djg/AA1QJO6zdUzRu4mxO+5T4DffFsbjka8fBqw2cjZEM11f8OLRXOnNkHCHj5FpAB0as35AwwecNoYaaW6AXml4LA48uBFNDtPDT3PCwpGBkA8MZtwKpKx8EWZ0cNwrKIXDYaKnZ/yhIe2FVj/ggifGEaeRUt4gGax0vBkOQ1vhYRLuGmy906NjGCYsRu7vRn77h4Efyh21jrgmuLlx0A1vEWcN9YH7+9cMT3esbkCix/tp//0OldCeMe/J4rmfAv/Tn2eMxJevLkz+R7S94F3Hu12Sdf2IPmb+J3/2TOojYiN34/dQv8PxG5QPIAUXPeKNXjdKypTUKebpFQZGkIBpQKxy9AEF3saZJReivPDn7zr/xS88cZ34r99Hvm+N7kBV2D3unu/aTPKPnK/KZgOJDd8Lm1VutfLvlsKnd4+YjLTueP8tWH6L9o4/dP6jnzf+5KsbXz7+lhi+YStPOOmoCF4a8xiQPtO64GzlFw9PnLQyaWLsnc2NfMh3DOvMXVecZqaY+OoRfuaEu1unLYVbE/ow09RAKqqV4IbdzOk916Q0rRxPjaNVrtrofZfMXW/E6FiWjDolDg5KJDRHUGFUQVKj9UKYwq6UqSHthgsVp4LqBC0yRGGSglbh7dD5s0Nguj0gpfNYlC0Zy1ipTijdEacZR6FZ4Ifnkcst4FTxMjGWB1wJSFNaUHIPrIsStTNaRVrmKA8sJdMsMYRAtIWhZKLBsR3oTUk4TF+hfSHlhWkoiEHKM5vOdAOaJ4pSe6EVxfr+DLdqmA27amQN72dcSSidUQPSlc3AmmDqKK3ju3KqjZ95j9zAtSMHm3nxjkpj6xutCb4F1HafgiBIz3TppOzwPOwjk1oxKYSQCe6Gb2dmp2iDYifOL0eSHQiDR+hM9YYfRwZ9wPIJ1ybUGmoe0SOJNzzliMqVhyGS8o33z52bzMQQie6e9nIPFjg6x+wWfvVofOkypSv9OPD+RfnmcuTjNjPGE9odIkatG0E96iJ0xxw621IQJ0x64noFpnd0GrXmfZwdjlRLVNdZO7weJ8RHzlZpveJ9w9pAt8pmnaZCl8DDcOB+u+JS5efhnr+L97z0t6zJYcmRa2cYjjA9si0vcHji7qCEnNA6M3FkcBO6Kt+unbMIzgV+fjxgGOZmxvtAf7nwqcBC4xfO83ZK/JrC399uZDcwun9cg8A/mgx0VQzFi6PZ/mCkHshNCfcDrm2ErpgLBA0EOl5gK695n/4ZqY78yc892G5eisERQ2DwASuQ60rrmeBAnRCcEHtD1VFNMAKNiYN/zbrdEfwDua6kUDBzBB0I3UP3FIk8L4ExjjjibjaplWl64s9/8Yo34ZlD/R1tzNS7QA0z5zpRyj3ux4B04RgcI44gjikO3PkVbRXhGaeFw6Hgvnzg6/KIjBupP7HmhccyMLWKm4S/urvjuwv8eO18yrK7ztsVawu+e6KrnC+d/6CvsdzI54F4mvaxhFsZyo04HvCq6E+fuZkjMBI64DzaAUZwRrUbcnD4PiC3DUpm8o5WBr79VHjz1sjhyi02cnA4d+Rw+AJnd+QqpHJgqZGredaYuX+7ofIJlUpzhdoSzVZijKgfGMQTPVxqJsRO75UYPT4EapsIznGwytv7wDA4HmahtI2D7X6ThtLdQOoeaMTYmKShi0B3u1PWd5wYnsYx3AjDB7JPxCGj7gk/fORRT4zbO/Tbe94/K4OHsayMseJHQ3F0H3B1BjxeYYqRSuEwVo5zZrQz3SWCq0zye75+pcSYUL1gfaO1TmvQOqCG9onYF6z9HT1EfHiLtiPkT1j5gVIuEBzNItIDGgY8A9Eia+o0DB3uMH9AAELktghqhT+bE6dZGPTC/+pfJd6enljfj9yGvyD9sOLDiqgjdWVT5eMSuPRAjPD1kPjq3nM0oy5n7iyzyMrtMvP9NlHaPXrr/NwrPt1oVN5Nha/vv8XLN9T8jHgjeE/rBcQQq0Qb6K3R7XvuTj/w5SzU2hjCK5b5NaJfcgo3vnjozPXMrC80vbCkV1ye3pHtRNG0J0uGishCGBwmEJxgQBNHscxwXLgPRl3Tbvwl40Mnt0p0geSVOvxkwMsOTCEbUhWrnpobwyHgXccs01nx6vZDiBdaqEi7MYwDlp/58m5irG/IdULaRs0OnGDBsdYTzxejq+fjs+P55ghD5BAMv1uaUAzCgZvN3JoDSVhLqAg9Z4yMK50YIvSR2ke29YXZdd4Ez9vmuYinSaC4yLUGhlYQFbYWqDKgvf0PJj315O5IJpRUCGHcDxq909pKPEasdLQ1DqLch5FzymwIhpJFSAIfe+fnU+Ck0DfIPWCMNPbYptSO9wFnATPBzKhkmmUgUlugdcGHARhZ64qocAzKUG8EPJs7QT2h7jUVYaSi4kk58+roefrxmUkaqGc1z9ZhaSObvSH1QNA7lvVC64E4PhBU6RV6F/wQ8JL4178Y+MLeU2831i5kAk+L52VxqE60JnRlj6r3xJtj4eEQWFNgWSCXjhfHFMCHExdmSm84uxK80a3vcdJe6VYYvEOlklXYxOFQelWGePxJlVL6T4qmExh75U4yrwZP3SKVQLJHCjOxn9AyEm1Ee0E1I3HmdDjxePNcWmIaHK905Lwm1Cu4mW048k1qXJ9Xmt6x9M5Wb5y7Mc+VV23hKw1sKkj+x+3x/3jPgO0GIrQwTgNJjawOywptJVhCuqMF4Xm9sVZ2B7KOuHJi2R55uDtzd7qn90SuUG0iF8/jMUK6YfYMJEQ7Xh2DC0hzKJ7KPet6T+FEdYEhdh5oLGTWlvFq+K6gjhYGco68LJ3XD68p5YwGt3sQ4hNBB1z2tJxobcWp8GYKZDJtnOHY6OsVZSTMSlfHlgOud5wUXBRU3D5PJKN1xWQhDAPRrmgpBKtMsvCzhyNv7idyddAU/8Ez5JFkwu/Whe+b499+N/PDU2TtBywGHk5Xfv2ryM+OBS+V+tMLp60hHWprnPxALhnyE2aPXFPipWzooLx7M+L6BaTvykNwyCxoMLozctid0WPrTP6EyZHSI90GnI5sL50frgbjxNs3B1ovlLYivuO9MsaJOTwytXtyM6xdCdwwZ5RmaDhAH2lbpZbKoJ1DvGPLHUGYtNDqFWOh/yS1IQ1Y8aGRxCil062DGAioJaL8Hjf+A2oLLtyI8UqMjZQT99MvibrPQA++MfbE6CqEgvjImgPZJnxQ3hxA/MLiI2+HK+NwRduFttuo8VEJPe8mPAHCbi2pBTDoHYSEuA2TgOlC44bGE1RPdxHcPaXvhifE4ZgIEhGpiDScBnLdcHFFNVLGI9ut8ct45hiuaHGsvnGcLqi+58/+5JHw7sJ//CshPZ15U1d+vG18siMxRKZivDoU/vVD5W79wLEVpnrjTiqf6oXfH97x7Tazfiz8ajS+dBdsu7LUyKcfOu/eeXyo4PZlockek8IM54TIhW27YVboVnd1QpTmEsMp8B/Pb/l1nLm3BV+vlH7jSeGvbw98zJ4t7bHDphncirmCHzxePUrD0cE63VY0fmAMJ6a+YT2hYY+CTiHQ0sCiHXlY8KPCs0OyJ3ZlsP0de/roOExHovth97j0DNXhPAQt+MFwfVc0J9W9W8C94WyvqbKrTslFPp6FH14GnipcaDg/M8wOFzqbJbpl5qNSbwuDnNh6YPWF0d8Q26h9V5fqtiClMfmBJMKPOjOPJ15ZItL5k2B81zIfTcljZPMGkgkq3KqjVmPCaAJr72zOc5HItQWsCoNzKELPBjLwfMvczxHaFRVhoDG5wFoLTQzxnoaRvOOmlSkaLjh8bnSD3gZWMrM6aoNgRtSBZrAJ+1hSPJmCVyX3jgSHyMSSKw9+IvZEMOWgneCNmxoRx4DixJG60fwTb14H3nXheoW/bzdWGbidhcbM1gsLM32aWN+fqSKM04luQg+Nta4cxxuvgueUClc38O0W+e6iXMpEcON+cK0Z9ZHSMveHxq+/qPh+45ML9J4xc4gKTjMaE1VHfBW+jCfO9YKMge6EWjqhVCRvOA9KRJ0n90J0gvbK5AK9gVjdvQQh0msnAHcY8aQ8pZU/LEe6HjCOtBZoZaCUQFaDMDC8GfhiMoYiqBnOIv1aWW4XxjphbeS6VhZzdK90F6jpRhgizkClECUTXEPbH1kZoHfoYFRK30jRaCHQu6cWw0WhA0spLCIUU9RHsnVEOt4NtDqgbt57AlrgsgZyEnLZeHt3j/dPtFYwKiZKaSPWXnPNA5f1yNbfIPFAjZ27eGFoVw4Ozqnh8YjOaPtJVpbAVu5YSyZE20tywoQm0LLusRh3ItvAttzz8jRzPo9QT3xx9EwPRw7rJ476HrXO0jfm8IS5TNM78uJY10aNN+ahY1KIQ2fWTj/fdpZeCkk91y7UbgxETiM82kdsS/jpniBfka4jeQwEmfj+5vnhx4Fvr2f+6i88f/pwpWzwxf0DjkKQxuQE0446CGWl6swiI0/F8+M3jmcx/sXbAbdVBud4nRrxBJo2tP50ehiMQzSOU6WvB7JNpATZlFQV4w7LC1Eg9YZYp2P0DpNOHPSAlD2TP0wdkQ1Rw0ngfFsp2RFd5BADxRrajErAx5XoXzB3pfYrpZ/xruxuda+UYtA95/MNxoAGT8Rh7SM+fEd3v0O5onIFqZgFHH6fBYaBwzjyMJw5yErQTlOjOc9iexrmdO/51dvEHD5Q/cgoz0hYEcmoZKoWzCC4XVYVARFFxVA6tf5EDrTT+gre09HdSCUK/ojajWYbzu2x1cSMtAeOEuj9E+IFI1PkE3XIaBjZ3JXXd46vVHAl820euLSAmcOWwixP/Nnxv+VfPBT8l0b/EPl/3d5yvt2jbuLojb943fil+8DgMrV27mJkkMS7GnnuJ57eK4ODu3jhSKFPgQ+p8vTBcfrhxM9+/sAQOsUytSlOHSKZIXRGSbS80kyxBt0McZ0mG6fDE/fXbynLPcOyf5YyVMZ5wOKBHzYll0jYFN9W5F1Ejh5niknEMph0oirON3xYsJa4mzvKbXelW0M10DF0vOGnj8xBSQfH7WlidhOshllH9TWXy8Y4sx8CugeBVq6IeNQ3AoKtDaudIJ7JZZ4t0Q+OLXs+/ag8vYxcGMmjgiZ8VKxnrAvFKr0V0uCIQwdLuJRxYmy9ktvGbIlZjV4S2jpqAQRWhYYSq+fYNx70hgsO5w78rgnFT6w6slney6AcaNswUWoIXJtylcDiJ7R1QutM3pBuiI/civFSC49eUG27BK+C94FqjVYLwTlwRndK6yuGcZxmxhS4WEDdiErDa2cUhdJIXeg+AoHgwfVOt4Tze+GOqsO5kdvSeHQzUhODJKI0zHdUKkqmiZF9YO0FL5kv7oR7hL/9OLP1O9bE7j2YJp7twHW7UpsQh0fUTxiR6+XMrVWkCAHBtUwqgd98bDzLHS1GjqJ4a2hXMg11whQb9/4KeWVjYowB2kiTSvCNwRtUo3fD0zm5Abxwzle81D250Ru9Vsw7mjgyQhQYHXjpFCdYA/NCd44eTgzq+aKfuYuFp9cPfPsMmzM2NqIHb57aPM8VaCMXAj7AOIwchhOHVHDuBTlG3qiQLh8YdOTvSqC6wCVVYq2cpoZLHa8QyQz7y/XHJQOiiun+stVeSdVItkfGluLJsdOjUorQxCNe0Q7dOl0aIRpOHdviCdNAaQO5BZJ5ztuNu+lEiJHoI4jQ2sB2+5L3T19xLSOmIz3MpPVGlEIQAVuRJgwSCeGIlESgYmY4UbZ24Pnq0GGlR8gpcLATok+Ia4ifKesdz9fX/P7jiffXkS17ohkP88pf/nPPv3r1HdF/z3G6orr3B5S8cFl3c0+X33MalajLPsftQvYLpXvWeuBjD3y/NFbpRNf59TTzaoXlcuMbO/LbMnBePRoEVxshTGxNSNnzb77ZmPyBX+kL27UDI9Ve0NAIUgkovYx8dwk85ZHFD5zTxKfvP/LFl2em042lZwTh4AyfEoeeUS1sHabgOE2J1i58fPFc04EikZtTeuwMThhtb3cUAlVAgjIwIHlvTGwURu+gN0QbQeG5RG5n5VoLTZSsB6Ts6Y83bwWvL/T+kWYrQSuu1718JUZqu+dl8fSt4Uy49hU7HDhpxvsnevmIyYJoBYXWGqUIrs4EUe5m4yBnor+gEYoOXNLEp0VYxzNfvj3yq/vfMbvvqRKwVtmWxDh7sI5Y33shxCHBI96BeiTf8ALEXZVWEbBCK2cIM0YFrXvjpQu7dNiMbq84Pz/Q+4SdjOAq2AeCW8A1dF6BGeoV2QyfG9Ymgp+o1WjVEKcEWQmyoG1B1ODtW+RT4enTSsV4PUW+mq5Eu/LDrZDcifuHE5I39LywfNq4ZMef/+UdfzHc0JcbHy+exVXC6Lh8Evq7I+PdlbYkpCecT0S/ELXjpTMFz7LVnShVMNcJszJaRuoZXx6pdsT3MyrgnVKc5+MWSfWI70bYNr751Hj95oHer9A93nuOKnhxqBOsZ1SNwwFUEyKGOE8uHekbh8MZ00x3zzz+8ityVGq5Q/I9SENkoteBVjy4gHNKLTecKKoeo0NLWNv7TjQaV1Zw0OeBl6dEfgoMNhFi2H0uzmF5YR4ivnd6VzqOS86MCgfN+Ot3VHF8ihWNyqUIJp1Xh73Bs+rItTs2VtBODkIpRuiNWY1H1zmLchGP95FaN5zccHnhNHp6T3T1OBdRC9TW8BG21oCM+J2U5q58ujbm055QqL3RTakiPzWuKtYbzjtM9w3NrKHtzKsxUFLF/zQStv+uZG6eKBmWrqgEVlsYR6WmSqei4hD6HmXsJ0ppDFIJHaIBFIoom2tsplQ5kKwylSvOrjzEytgDefWIHYg6si4bv/m+cX2upDogOaLNMdCYdaAUx/LyREoVs8qWGuIiWGWe4HRwjAXyBi8VUq1oTQTZgCuDF1yLdFO68wzS8Jsh1fAu0iQjoQMJtOwNpk24qDJ6x9oyqxmox9N4dIGMcXGCyUBzj7QIvTmOzfEv+sJ8e+H3bLyeIqkUOh4RCDpyTSO/S566dT7eRs7ccRbPNQ/I4rncVh6OjbePYLwQtsY/pBNrdSytcaTy6DZi38AiePbE0JL+uGTAzCNuf3GzNUq+YUOkITw1x1sckyiD81zLXshg0mhWMTGmUeitk8pMsshWItkmTAM4T24zNQ/4WBE/oNxTti9o2yNDdHvbmwP9qSCjVkGz+++LgWoM6DgiaSNKxLZIaSeW9sj1/MzijRoHvnbKr+IPdJfo0eNc4N1khFS5yxtPduR9mvjdh8TpzXv+2Z9+j+jfYGWhNUWjZ/BHDq8UWzMhTNxPEVLFOFLjCXf/yO2sXF6GXQjXQuXCaQzcTcowK/1u5un5wN/3iWfZG9e6RmqIXM8bfV148CuPbuCda2wvK5mBM++o4cYXD4ZW47Yd+fdPgVQcIo6le1oZuPaNcHhmu944nUZOYUBTR2vCOzi4gToOhLjRT53HPLPUC9f1Qg0zBCNowjdj1GFf/aWDeOgj+TqwLYFsGaKiophWHIJcPOvfZloScvAsh5FPOMCRFscvvjwADScNE1DnEQ/OH7mUe+rGPnoxBYTuN+Y54dyNXgv8JCljYOpZ0wHXX3GnCe+eGd1Hxmlv9NrsxPffH/j+eeDqX7gfFo5vfo/r3zAOB1qP3JZ9QdRgdAERSKXhOww6IEzQM04TItBVEDN6qwhK7x0XPN0C6sAs0pvSi2O5PZLOX1PzTG8Lb956nJ3BPqI+IdporeDkATUDcbtoro7a4bo1Tq8E1WEvqEkKecXSws8Oyp9Gx4cNDjnxOEFNyod64Kwzfx4npuqhJMa8cZQjX9ytfPn4DTxeeGDkZ5JZlkTrgcFVgpw5TVfu3YrKRte858Y7RO+oA+TaiYPti03sTF0QV2jRqP4RasO7F1zwrEU5J6WaMoin6yN/983KL9+OvPkCRnP8eZx4vCY+3SopgYQIckVdw5xQUaxC74q6ihtutJYoBkUKh8OJ26dH7vo9Q/TkHultjzb2PtAt491MYJd2e297oU9QrGW6rOhcqS/Cp0+J7XmvFm+6EbtxaIaTdU+2wF40MzgWC2xSedlWBnflEBQ1R9GB1AuqnZQSnGbkXBFTrpZZrLJh1G64IXJXbJeCc+MYoWtAzOEJOBEGrwQz1EVq3+XneRi55kTwASjkngkuIr3Tu7H2wMcl8fped8Px1mhmWNhn/04V6Z3mPNlFVApWMq69MLkB3LiXiQmI7dHgcTjwkpTWbfdJ9ETQRnOOW804CQQZqCqUGjFLOBNmq4gVrqoU9mIndY4mgfzTOOUYGq9mz+/PBXWKlk4XeL42qoE/TGy1sq0v/PKLE6MJkitPBT6dFw7DRu3CHANbT9wdKn5Z6GtDbYIOTpSWE2IbohteA1oa3QlZHNE6vWSCGaU2XAj0uiG6E8dujh4iT1Y4eKWIUXojivJumnkrmSp9L67Smc0iASO1BSeZL6Sja+HrQ+EL/y3XekdeHVsfaN1oZeC/+U3GqtHcwKXAIgEsMDJzWYy1dd7dF94eO3e+M5+VT9mTtxfuNfPKnnF5JetbkveU0Jimf9w2/48mAwnZNyvxhNEx9kRzytJ3h/PPmPh69AylM6rnUipF92KGUgwnDhhIG2QduWUhdQhBGIMjl8iNE5IDcYx4eYXqa+YwYlEogV16sYoibDUw9SNbNqoTPq2N+f5AEEfZAt4/knPk1jqrHTm3ytIcD6Oji5J7ZhsacWzEi2PqR94dTnyYHH4ZueWZ6xIAhzVPriNLyZTbirqGczAfV6R5VN7Q7YHr7YFP28S1eppF3Em544W3w4bXZ2YtvM4f9jrPh4E388iPf9f4Zr1jtE7QF7Qv/OKw8qfvMv/q52f+6rEw2Ft+eBlJovzmYnzcPO9eDzyMR66XyPeWdpNS3b+vIRiHqTDGH4mvNqwHRntDuXgwh3jZq3mlky+FrQtuKvzi5/B23WisfFoXxBa0DuzFpIJ1QGfW+o5WH7nkhPmE1QWcoU6QTXFP8OVmDFa5NfhhES5u4qYTv/194fXrj/z5nx+BZS9JCQGNkWqP5NuBIQxo72gvlCa0Fdy0obJvvl4CJrsjvTRF2iPXJ8egiRif8HGl6gb+ju9/FL55b1Q30Grn48cz9ssbIpne9yga3bMkZQwCqojsRq3cjL4WRj+gTYA90y39p1dHAs0mKq+4XN9weZ6I4yNjvOM0Ktor7TkgN8eoj2w/zHysK+++Ou7//qcsu/pAazM9vKbNb1kWz1L3mNPz6nk0wQeg7Q7u9gx6KTyk9/znk+M/lImXLPz+myuvHzz9MOKmgV7zXmbTOvfR8W40Rr1x9+o9vfw1fsjUtvDpU2ezR6bphHfPeHkBue0xtK4IunuGJDNGQ/azNeohiBG0IsMGrzJ16cR6YAuZm3/ku/eGEZiG4SciH1jqkd89P/P2Tyd8WRjKDzz6jR5GbmdoNhHiFbTTxNOLp1neN3FpGBs+TpS8N/lhCfEGLdAswuDw6uhJaerpXgnxgAseK/tmYFZAwVwgbQfe3zzffdy4rgPlOaFsuBgZ6h22dtRdidqIrWM20pwnibE4z0cfoCdUFrqLWNtHKNUSN9/43fLCq2nCt33DUudo6rjWxrUVHoLSs9HrXlTjXSOG/aTtJOwJnpYwr/sp1jpaNgYavWS8d9Ar4gzZQFUpAp+KUG6VMHusbwQZyM1Qr3jrnMLAGIafGiQ7TepP6Q3F1NNNQXRv/qsb6F6C1AKQM9S9E6aLUILjVo3j6Fl854wSqzB14ZU3jtJ4yQ4Tz8F7mhWwRlUh4amWOI2Vr0Ljsl1IZVeU/eEAFsjO2HJC2plT8Mi1MInnIoEfzyt3D3tbpJNOGA44V7Dqsa6YdMQpvStb6bTWmXwnWMWHyKUL36yVk3muBkjHWabWBfF7wyca6GqYE1JxXJeNYRgQB7NzPGhnLHv1cxkDlgfS0onXAj2gGoANrDLQ+PXDRvv0DTfn+XHNbBKI6vnm6czjF3eIm2kyoDxQk1CA3Aa2y8jffyzMbyMhVO64MJfO4a7xl7byeP2ArYXNC6l5rI27YvnHJAMFj7mdaae1M4wzGoSM8rJlRJXXfuWxJ04S+OA8P7ZCEkW77r3bKdIksOGoTfYP2JROoRLI7UirA5ckiHneTncct0h10PsGdTfVmArZlOBGegtkK9Te+ebsOMUTrXS26hB1eKscfEO1E1Q4hE63Rm+FrVa8D4yl43NF68IhwOwHggcfda8SLQNmYZcGu1JM8NJxtjCECXwkba85f3zDsw2cvRFj4xCeeAjvOQyfCLogqeF4hOERJ/BWC788BD58/4l3j0/8l7/e+LNX77n33/Lu4cbsO/L0M/7+6R2/SZEsAz9k29vSbsKydUjPHIZnHu4zd1aoW2ErK95WaCsinZo9W3bk8iXXJVBphFcTfig0KqklCi+ovTCHlbvDzHx6IF0OeF6hvHDeVhbx1HLPc36kpxPX1BAqg4uoLWjrlFvDLcZDX5hlL8A524rvHZOBrf+K//t//cTj61/w9tVtrygOAdyJtD3Qs+d4LAxjRVNi7I7WByTfoBtuD/binKOLQZtw7Y62gieDv1GmQtOZT+eJ734c6P5EVYe5wJoF5wS1Shejyt7D753ui7juxKfURm/7pVStwKiv2Itd215kko90Hkh14FaOPH16w+VygsEzxo2/+BIe842TTdTq96hWmvjhfOR4eOTx9P1eSWtKl5H19si3n074eCT7kfMGpcF58ZTqKW3FV6UtFU13kBasfOToYMwnPrxM/NvvM1/8CyG8rTwcPZISXK7QjEErb4KRzoXmKt49g3sB2zg8OB7nhpQLdAMRugkVQDxqINZRt6sxIYABIUJ0IHXBwhk9dbyNrMmzNOXffXvg/XaPDAOdvOfjBcRPPC8Lz2Uv16m3F1zvhEFYCGx5IA4OpVBzB3NY2bstRBzOAb0w+AM9zzjzSF9pPZJVubSCDAPUV6ADaZU9D36qTP4CvaLa6f7AdXnN7779mv/3vxGeL4lhFNQ1NFS671gWVCP71myYZsxFkkCuxqKdFhxLAqfC4iq3egZxiHRWKTQ6vl557UeG7jANrBqoPz1RYh0RwUypXXFeoRiDOGIYKX338WxlY/Cy9z1QkVZRL9SaGYLDkRidp3ZjUyN5uGYIlnfCXRtTGKDDo5/4+Thxp5EXEZJtdO/+h8RMN9QBbleWJu/A9vGkyN6jEAzsJyWiBaUhbK3iPFxd4WCCq0J0mZPvaGoweHrPOGfUZhTxPNfG7AKab7z1H3k4eHS6ZwkzxSlPl0bqjaygfSHYDFVxrSACT/nAb54D0xjoGigo17Xiq8PJXsHT1FG7QY+ULNyNjbsYGMvABxX+Ydt9HUIjyMJj6Gh7ovqZ3B3FNWrvlFwote6kqmSCU+4Gt48YrGGiDN64LYXleWVKDXqjjXtqTcVhRfhn3mD+gX9YFl59/S/5h0+ZrSmbG7nKAzkJlwvQV2Yf6WYYirlX/Pb9mT9/PfDrd4W7e8ff3jKJzj//8Uq83mhLo8VMUUfqgk/9j0sGunpSK+Ameo3k28LdYcArSBi5D4lDX0g5QQ/EMBKlQm949bTmSXZPlbL3KwvEGAlOEDLOT4gcoCUQJdXIlhN304GtKefrbX84xai10byjeEVsJK+d2mVfDPq+6fd4ZRqNV2NnGIyswvvbymOrUDMWFdNGt7Y3dcWZIo5NlLXt9ZQx7JJxN6PVDhoRUZzujVviwXCkPLOsr1D/ipnAeTsTxgt38SOTfMDV2y6ZMdBDwFlEgnE//Z7/5b9c+dd/1vn67u/55f3fofUbmi34EJH2NfAFp1E55JWaP/DFaeRx2HicV371OvNqXJncGctXpELPd/z9t0pOhfu7kZIL1k48f3zLy/evKS+O63ojvZr4+X/yitPdN8xun7FbO5PLhcvWcPpnzPIX2DIiA3gHToTWKjG84NyNEK7kfkPITLFSe6fmvboz2sboKskHJCi3vLGVJ5oe6Pwp//bf3fhf/5cnnL8QdMSYSNe0S/zjR4aWiMX2sdLLCcdeUqU4VNruaO8D1r/E1hPOdZxuGJULxkt65G//biS1E3Eeaaok2RcFr51AI1tBFWrvDMOAk5HWN1RhCnvdb67KJZ34/vpzcg24YCybY7lOuOGw/57cRCvK3TzQg+Cnkdw6W45sa0FlYF2N8wYv22uG35959R/doZagC52RH7+Z+P7HkRQ3huORVT2NxlYdzk30njDLiFOwkSoei8rtupDrxMErPyxH/uaT8eZud0nnwXYyGyZGXflVvPLpx43bRXl8rZQqVHHICH4sPD3fmPSBaYg09tSAdUXDCH3d1YHecfpTcZTsYQuhwtGTLpFb2wnXkgIfzq+54nGHTmwN6Q165TB5mhz49vnCVz8/ELcJc5XYOtMgwIRkBypEDbS1YlV2F7wqXQRcwOqElCORSG+JVM9cq/BDFazdUfufslyv9CoMofHlm2f+5OuOkw38wKfLI//2v3nH9z+85vzicR56zPShoN5YeudpS3QPoyrB2h5XjMaaG7Xvs/bBO8wGLqmw9kaSSs0r4xj3my3pmBkHV/CtQ9kTA6pxv6SGvNMC3cdmDiEYKHsFeumCuZGt2l717fseq3QdnIIzcl2Iso8Tqni2mshS6QI9Vw7d7evwsvH67o43ON74AeueKp0koNKpNdFoRITeBHrAMzB2w2nkqQrdG9ELs/eUzViq0J2iTqhUttqYqBSz/57cnIIyK7SWiNHvBXTOccuJ36dEPERUC6/6GR8EC8Zva+G5nNjKvkZXMw4hkvMVXwW1QIzKuTuen05Mhw13GCk9UpMQBKbBo+YQieSednW5D8QWkNbRrZE25QWHuQmnhTFvPLQLY9+4daHrgNl+B4v9d4cI2+8waF15umT6mxkXHBo90Ud+fMp8fAavAwctewVzDPTBo2Vl3la+ovDsN1Z+y69ffcl/+GFjaw+cf+zU7hjdhEfQ3n8aB1WuWaCP3NbEpE+oXvlqnHlejHvrSALLBZGCDLbHIf9xXOD/jwZC69AbVVZMj3Q7ck2w9ca9Cg9uw7fGWjylwtpW1PW9DEgDpc7c7EDRjebrXrrTCofR8eow7JcM2YjT3fkY/YgZ9F6pSfA60GiU3sGUVBu5XfaLLGQ3HYWxUtoTtVwxdhXhYToSHQR16OwYLjeqq4gPiBPETXyqA2d9xUt3vCT4YCM5CEtSWp1pVemyxytV97Y9MyPnRsHzYYusbeA4DUQ8B+DdfWKSM1IKqiPO7XcH0EdwDqTht+/40+NfczreGPQD/fwDvRpummE8IuURCcqb8Lf8F/cfsXD9yYsRGA+GtvfY9ozkRu3KViJO3nIaX1NSQ3VgmEdyvud2PuI34S2VrXr+H3+j/LB0/vV/deTx7reM8RnrhaCQe+fp9oykSsgjwxCZhkz0N0Jo0DfKdsVGoeMQ50GMWxpY1BHHieDOqHZcMPowsNnIVirCjT7MfLi9xuoRNCF92LvC+zPD4XlXC3pHXCPGiWl73C87cnFvjOwdkQD2NS8ff05Kj0ho+Aej+MjzGvnNHyYu9RVuHtmCsTpjFSXOSnQJZ4XcEiUnKorawCBHcmuoGM7bXmcbD9yuv+Tvf/tnLHbE9IpGYZ4f942AyiSN+VQRyxQDE88tT7AdOZ8X1lxYk+OcMpWRjz+e6P/ihPfPu0dgm7heJiqe3uHpdqOHmeoBUUwSZheagXcjaXRUFyjtgX/4jfLx7HDa6X3gD99BOc2sIfO3U+HucE90Zw4Ofk3nm5fK+kPn7jHQdaS2yjAGdIwwVp62DTfOIEotlZYDJQ2MU0As41zfc5YCokAXzB9YX77g08sraprofjd0uWEgjh1xG2MB1gJBuT8pIQy8bI2lfuI0Z1rLTE4oLLSWUZl+qnotWA+I2E5OWke8gEyU9gWZLyl22G8QNeHDS+KJI+ct8rK8Ru0N4wGOGO3T7wnjysNdp5d7/r//7R1//Yev6BbRsCAhcXPKxQwRT+qOOgUqxuBgdIHeCreeyPPAumZUAr50pBnNe17yinlDvZBrJngPnX0W3TOPcSavhavOdBdxDlozmmRQCGpM2ogl7WtfcDT13HpHemDyyugSjop2I1lBaEQHvu43fHYX6NrIPdNV6babFJ0bUBUsdeZYidZYcTSv5L7HpeN8Qs9XvDhojiAeaXt5EnScKGKdVvdrvdR5xGRXCOreQ1ANRoOigb1YPiC9MwyOrrqrGBpJ0kgS+PF6pVwWXoeJuw6jGWm7kpNhw0Q2yKXtV7a7gaU8M0gjHF/RbomLOW5NGRbj7VFxXrBc0FhQq7S07Zc/iYHul8aJCL1lSvupwdBFMoVqDbQjJaG1gm1IVFBPKmWvaK95r2ASRxXPGeV3H678p693L5EUo1+Fc39EPfQhc3Ab0Qu1gbQ9LhiBcXD8UD/yyy8e6Vn5u7XztG74sPvHujlKVJw24uGAtUSziXU743qHvGHXhuZAkAA9YH0jqhKkUXql7nnhPx4ZiL5gRDY6pd8wcVwW2y8k8olRG700kBENitfE4ApvB0dqRr91ajxwSRs499OtVw3JjV6Vrp7gB6yX/ZfpPVIDSzLWLuh4IHYl9sbWC6UbzjWqvWcYFU9B7JnWnzEtlJpIJXKfR5wLqPN7L/aw13g62fAS2eo9//7bI9+/TGwy7hl39fRg3NLE+08zh/mBwpVuhdHv7WalZUKcWfPIeR348bLw7vBEjIVXj594Pf+Alcs+I9Q7tj6QWkTtjkdXmKaF03QD/xFxz9AKhANNPRJGOnd4XuEnj+/fcJDfIO4CbidWLI6cr+SawPaHpruBkjroFaxj9rA3oy1vYD0wOriPiVdH+Lo/8n/5bqT/P43/xf/sSLAfoRuWDS+O0RxrMz5+UsI28vNfHQn2Da5+QtnNUaUbTiLUiOiBqU8khNPJMzw51BQfA1cv/FiF4u+IztEQVl7R+wlpT+BGSmocQsXsQmkJVCi20RTc6YwPX9HLidK+pNtMtke+/+EX/PjdO4zIPCSOYyW3iY+/D1wuh52JB2heuKnQDL585VDd2LYrBdsvUqmdVhoueHxTOg3BENlb695/M3M7f8Hw8ADxGelXWDPjaWAKhTk+c/Q/jRXKwDVtnJeN7fKIL3fE0sm1Ijnt8TB/5HZ7hNMZ05kPHx4p7QEFSCsNqMeBikCY9r+1gSh9GLgc7vnDp5EP3xq/+aC4Pu7JEpnoyVHKiW8vjf/zcmH9BfwXX40c2guH5ZkvED5+F3jz61cQVsBhavTWGKYB2wIv57zL6nYg9Fe8fHTcpgtvv7igrvxkolRMJ1J9YF2+5Pb8S2x4g0hk1oHYMq9F+PkLIAVYKdGzRcd83EuXijqWnmiHZ0o5I74zagF1iNsvD+ut7w7x3nd1ShtNJ67pC37/8Vd89+kNz8+G6IAvjpYcK8ImQk2BcTxR08JmHxmGwsfnQOpf8odvR/7mD/d8iPf44HBbRZ3xvhkXU3wvRDqDD4zmeUvn3owPvbOYcC2e4j2jNbxkaqt4Ba+wtd00LarUWgkIwQVonaCV+xi5YKwysIpwcY3e9+a6UzTuJTGHwtqN70VZNLJYxJjxveJkBbfiLZF63tUEhWad1iqrdYruI69WCyIeExDnaV3YsqGx09tGZR+3mROaE7x3zIcZ1o6mXfIfvOAsUlpjsULxhqmxWgNr+GHG1UozoUikoDyXxL2HybNHHmtBJBCHSGdPZLlJeDwNXIk7AZbK6WEkWGC9ZVoNrKysm4JEhhGG4EgoSW7gL4RXno/LC+YSP3s98iePnrVk1smTtwW7GOAodY8Im+5vdxLHuTRM2ccc1n/qOtkQS6RWUDOsVIJrZPWkWukK4oVeKk473nkygR9uF+pdJhiwVXo+khgRVdBA0k5MZw4EBhUSSulu92zg+XF75nA88HXI9JLRcEQE/HiHyG6U35a0f58MXFdj2TrbVrDikdYppZNspPeN2n6qhPZG3+yPSwasg0gnYAQte2ezKbUZKhVyYtkSRRzqA2Mw4uDIrTLbzKZHot4xa+dcryz1xhwcwYR1SUzsBTlCx+tA7zO3fGRJA2uFvDUeovLLuxkT41w3mha2fKUIu8HN9mtHAXCOVDtr2piDUqqR6x59GQ+H/WvqwI/f3/PhfE/xB/DDT/M7EEtoHHi6DRBH0IwLninqXsNse8IihEdaORDw3N+tHIdPqHyg1ydoDdGR2o48n1/zfptQjUzvnpjiQs8XartBS7ty4AIu3rPWgetlZnD3vBoNlqc9TmYRcw+cnyYGr+T20y2Sqrg4Yzf2LO98I/fO7eU17Xbg6cMRwpEQDX81pH7gy7sTbK/4298F/pN//pZXw+/pNeH7nrBw4Qt+/+PEN0+Ry3cHnrbX/OWfv2YcLiCGeYe0CijoibS+Ylu+ZmHm/q7gHwVZPGAUHEsfSTZy9J7SClsxWj0ASl4rqQpzYI9TuUxujS6JVBb8ITG7e67Pb/jxqiyqvP80s92+Io73BL/Qh8zqKlt54On7jvMzilGsUsXTVBil8NX9hZQubA1yEKruhTbWKhoL3gUM2Q1Z6nZJPh2Y3MzQKrosWLmhApHKmwch+oRnwwiUte41rIzUDrOf6aky9CujFUw8rSk/fHyNW+HpJbDdviDEe2Ku5K2w9szHbSHJDN94fLjn7V3nJEJtE0/5xF9/CqzLEQmN4DshTuS8S+iWwft7zvLI//W3fyDeKf/qPjPfbjiE5XrkfH1gevXtXhRlnVI3mgvUtjdD1tzxXajVQXvDcjtQambwZb9qu028PD3y6fwFqb1j8q9p7gl3dMR4x1gc8ftP/Im+QWxjdRvvZSAPAQvbbnbrC40LSZ/J7hNNyq7A6SO3IkQ3In0G/f+x9l9dsmVZeiU211ZHmJmLK+KGSlWCKADFJonmaAw2m0/87RyjmxgcTYBdqCqgRGZGZUbEFa7M7Igt1uLDdjZf+yF/QAh3Nztn7bW/b06l30g3zB/Z7Bf8w2+/4e/+5Y5PLxNLfr12Gx1hX0mjIzTjJia8b7Qm7NeKTYHn64nf/WA8nBMt3vBUI9cd0LnnEnzFXCd9Bp+IwZFK450Yw7ZxprPXvJvwTkhcodVX50rrp7O2d1uhKbUpCcd0GHCSUIRE4RTgXJRLGHEy4mJ/sbyPgdulEZtRRbgQWGokt0R83Vw9XmGYDC+N22HEaOT9kVU8Gj2r9X/WSWOMnn1vWApk8TQfSAbbfqFGRzVPqyAiNBMupVClMo7C3ZwI14KzBtL6z2PGpRWSKOn1asG5nSCezU80GamlZyF+zhvH20RqSgiBUeEUu/ZXhh44UctcVWkS2dk5b594d7jlzRiYHg0/TEzDgFH4/jgzX88Mzti8cXEvXERgNI7J8fUBDvmJWpQWxldsu6AcaNrx3T4Ibhy5Nsd5udDUGJxxrQ3nGqovbLpydgZRaK3hqURLROfZtZJCYJgiIDQHe3A8W+DpZeEmZEqBuzfv+fjJWJtQNrgOkZsw4X2/btl2x5WdaA0nI4tOiDaSe+T7uxExYTqOfL5c2NyBXGBdd3yY0BZ4uRovW7+v23Vjy7C1wJ5uadGo8UATxXkjuj/xMKAaaRow51FpFC1UEhi8HRO1eR43yKmxR88OoB2kUOrExkSpAd+OfIgDMsyYvpBYkPpCyWda3AkBSo18/inxcHGsYrTgEeCrw4FZNmrrKdld8utU65DomeINqpWtrSBCpVG0IVqom1AMFjV88JyGO64vtzx8ucGNJ+5Prys0CTQyyJm30yfwV675Qhh3DiESJCBupKmyt8jWEs0Zb98qt8MnRHsPPqvi3Yxz79n3b2jyFS5Ecms09wjygnFmrwthCtQmBEmUcs+Pn++56pHDeODuuOGWAkXQeMOyv0PjV/i4kFxg1yvZdlpt/c5TdjQatR74/PNb9ss9Vx0pUYArX40e0cawXggqPNaZ8zKQhNf1e6C1W7b8nku7ZR0n1ur43U+RD++eeXv3SAjXLiah33e/LO/53T+/55y/JTNxbpV/++Ydc7xiFnDbyrupsEkh+g3XrrxJnxkcJH/qp6wqyLTig1Latfd62XEpM4jn4Cv788TT4z1nO/Dy3JimA8FlvLvgwk4l8tMnpVrHJIPhDbIZDePGZe6nC00b5gZUlGbG4BPerj0R7gGJqIBzES/vcfWW0zh19nobqPsKzshn+PLTzm9+NUDZMAy4kEaPVdfplYtSNfcpPwiRjYry+cuJP/5tYy2Jw2nkzW2lzI1zi7ysI+eSODfPw8+Rf/7ynu/uHP/9Xx04tgUnkSQDl5qIUaFe0egJNwf0XEAbo13Y3YGXbeI//e4zf/bXEEMkTAnZjMvlkeGNw6X0SmFzOJkQt9N0R7wnjoJtBR120Dd8fozcv50Z/M66RfbyBuOGcb7tL/Z2wSbQ4Uz7OOPOibt4wSfjU+tpdufBucxxcpT1kTE8o26huhU/9NPrxpEf/3hkOcP7d294e/rMKA+ghfN64Lf/8hX/+MfveHh+w75NvUWQK5tUUqzUvPVUuTTW9YXdJQgnLp8W2sURdGSeb1A19mtmlzskelyEUp/wDgiRFiLmPKluTK5CyRgRS4ZZYW6BwN7hV953q2SG4AKX/QoePA6wDvlxPaDnaExkJg+bGteQCCIcfGWuwjF7nDpiGiklUURwzuENgnmsTqz1wvs375jdgLHxuWa+5IXiHEUaVncGbQwiFN+DdYQBI2AmVF36VhGPq7190c2GsDloThlk53Y0yA1r/ZQs4tHkuFplRxlEkbJgjIgNeKCIkn3gMxG/7Hxze8O1eY5i/GoQsMROfxnuGFMaaLXizKhtw8oDt37kKz+ylAOSlOkI78uFE3tv4cyBayucS6E4x10KzOxIrjgCeds70CoIW1aa7DgCp2Ok2s7DcmZpV/ALNwflvGauZae1Bt7xLJBFidY4RJBmVBnZXD8Yz8PA6fbEJVeeto2M4/cl8X7fuZ8C41jxh41975mP0jwWRkLLWM08Fc8XfwApnMKBaxMqnkmUkY1M7YFBFG2VNNySfO0NKgq1GWut1PpEGG4IeWRpxh5Gdhl5tsCC4rQxuD+1m0CN1hxZ6H1ql9BmHLzwNngeF0V3z3mvXFdlE6W5kbV6tmq42CfVmOH2kvn3//bEITXWdqXqincbe1to0rAS2Z8POHfAx8a5ZQQYg0BeEElEN7ApNOn3OWaGyMLdzRsu+8Lj8tgn9zAizWOb4ZwRI7wsjetyy6c/HHkqBzKNb++Fu7HhQqbaA1P6HUH+Fic/IMMTMhaMEdo9nsgQJpSBNTtiuDKnhpSNmo1s0FyicMfzy1seH45s0VGGSMnKXzbAXfFpZRwMDcq+VNATl8f35Ot7sp/Zt536ZmMQ6QnkbeJxv0XczCFthGDMEnDAuixdmFTASqWa45oDu65ULqQI3lW6LjIxzAfe3wiPjxv7S0FrD9lYncn6NT/8fMNvHxZetJKlEtbK83ngq7sZrwt+BCNR7Ssuy3c87t/x8/WeVQM/1p0sK38xDLhgfHfvyO2fGabGd28EZ0+MfOYQGrVFfv/YQIS7GZJv5LoBK7hM8JCSMlhlr4rnDb4embgwuEcG2Tsi2oxtHbg+R4bXYGsigvNkg1gr9yOModDM9wcFnZ0QncdVgyZYNTQGJM2oHFkuH0DuGA+OpheQK2HMmCimgYdnx6/2QMSDKKdRyKFiS6bUjfX8hO07wyExz4bUhRCFy/XA0+M7LArzaeBink/bDf/0CXYO+OG2s+cRLnUkf3H8+tHxi8kIKowKzXbCGLic+7pwjRMlRKKHdwn+8PTIVpTHPfBUd4ahdhiJjjgbCHiKgllkGI/obog0XDBMKlv9go8LHAp1ueXx4y95Oh/5/hcvPbyGe1W6Lj10rs/YUPFTYc1K1G9xLuN95XQ782UD8wvJFY5B0emRoT2wnx/xs2EYu8780z/O/O3fv2FtR9692fjv//qGoxtwubIvJ9aX9yxbd5bkFdI00Zpx2S8gjTkJbbkw1kZzngWlNkdrB4Ipd9EwMmEc8a0QPYhJH6hjAgfVeZqB18pJDNHSV/k+UZ3D4o4vjdEpPXKnbKZsTcnWCClRX2t/UQJbMdYg9LJyZ64ck8PV0AmXIQIZaQkRQFfGVjm2gWfxBOeYMKITKolcJ6IJv5wj9VIY5Qgx8NB2nAuIh5uWuVcjTkeucsO5nqgEJlFMFq75ArESXeiDvQJmPbQo8KIZp5Xb40x93rDa8NHRmlG8p1nBmRIdlNyzYXNMiBprUDYRPqvhqYTDkbgX7qJx2TIPW+UiG24a0KBoNCD0Gl3OqFduh8ZN3Whx7CZbt6BjIQ/COTRelsylZbQqwkRohhSQ6LEg7K0hsiAUUgpIDHh75svlEXSjSqNYw+zKHCPX2mhWkDB0hoU2boJHtXEzJsriac1TATHPIQ3UXDg6wQbPT22g7ZlbAec2bt5BaQ63Jy5XY90hO2F1G4sbeLATdZnxEjh4RXVhcopvjc2NPOkZd5rYzs/4Vrl7f8OXlwtqOyHCtqy49kIJQpXAUruQqrhAcSPKgKvg7U/cJhicAhmHh9btgBPC12nng3thL09cy4VFEosNLP7AUiY2Tf1OuIDzhljiZa/8+suVP/8mv1q4jGYQ09wBEOcRmjCODZ8y3gvBT8wOtnNjdxk/3nFK9zwuF7ayM6RGcI4QjNt0w9au+KacxiPDFtmrgaz4VDE38cefA0/bRJau9XRu5RhXzO+of+Rm/Ce8+0eyfqTFDXMg1mj1iJVIoYBvvJk8k4Tez1VBi8fFgSYJC7/h86cj1/3AOTqu1eGaZ88OqiHa8EMPq8nxPS8/fODl8R05nKgh4ix1UIjvNLXS3mPuPU4mrFWavCBeca2nlMU3pmlidjPbrjxuf89pDtwcPNNgTHWET+/ZHm95tJl5crxhYFlO/PDzn+GtEsIBf/iG27uRf3tawT+x7i/UvDHziWm4dAiHbai+4+nje84/veXm8BUlDHy+ZBqJLzniXgrNPfLdf1P497e/J/FHTukFY8OaR/M9j/stmzvhgiL+jJSN5KE199plN0JsuLrjTJlif+lOu4J1LoMzIdrM05PCJhz9iBePekF7VpOKoOuG+YC4hNYKNLRlqoOdSLC3tJpYyoG83PB8PpDzd4z396RwJbfPmH7C2caeGyJvyJcRGkQrmBhjMELw1JhYq3V4UPCgZ7xUbkZI48BeDhSp+Bh5WiMfHyo/vCR2edvZB7sQY4QUKHWj6j3/6XePnP5yZqBSU2SfB9oaWYcTn+vACwM6RWpbeX66su+9jfC0R37YDpSpME9D/7x7j9NEcAMupP5Zq2dC2HHeKKo0KsqFMFZG+zWXl69YrvdQP+LdZ6IrYLnDgWwnDS8QKlobrZ0IQ+zY1sEgNeJt5dgW7m5W7saC368MWz9BM8KmMz99+pb/9Dfv+bx+TZzfsD9d+bs/PvLnd8q0PxGz50RhbCttn2kMvJRMRsAC5ZrYa2MUoAmuCskJg0Q2f8tVIEoGceS8cpgO1P2K1Qahr9cLRnQOh+Gd4V2llu6sUBf7g7ZFKhXnDOcc+17I1P7CaQmoiNP+2QyBGka+VKGlRBwCS04s25XDYcC7geiEtENWY3WNORiRK7f+yIsbuwDOem6itIBx5Pl8ZjhkhlrwbiJMM0Pb+Xl5YgwT90TeWSG5iLjISxaGNDKJUOrKCwWLRpbamQ7WuWI9vOtRhGurrGSmCCVb356qge/niqyKUwEVRAotn0lhpNCFQcSATEeyC0yhMomy1MJeFnToWZ3ojGkQfG5496pYtoYLK3PaufgVVWORjU12NEw85MqSO/tAgHXbaceOuS84WjS8K5zixvubmc8a+Fkb1V94kg0T6RVKjFwqa9U+0AdFVPtwaB5zAXNCtZWqHvUT+MSaC7UWpuQxhKXBAtThgBOgZIZh5eZuYGgHfvrBUS6C+UoOO3sUHjeltVsOpozNcIwIGz6BBmXVnd0MiYmsYLnhI1S30yRTduGoHUWcxbPXxKIJGWZ2elhdvKfp/zZT0f/mYeDtHdRqnFLFJuFx28n7M//Dr5X/3c1PlO2Fx6vyx+fCH16EJZ+obaCECceMeCPXQmuRWjIvV/jyqSCDpxRPGCfUtX5P6E8MI/hU2cNKTMIhTczDxP50z+Ml8/yyUwIMp1tibIg8EELCdEN14ziOqMAcE2HvFcaG0Kyw15lLmaihd3hVG8uysXPFXCadMiks1Lb1dnE1GkrVwl6NVozNChYKgyq3NiF+oNWVLJlikcIbHp9u2PYj5ua+mmtKtIxYgHoLtmB7xcJAbm/49OOR1k5sg+elLIhmzpvD6Xuer/C0DexuZUgw2j3rsjMMnxFdScHjYiWEBbRwSDvH7xXnPEN85esvX/GsH/h9+4b//Ox5GSt/9b3jrUX+5n88smxCmAd++dfCv/5X/8IY/wHhgdrO4MFqAj2Am1G95+M/v+HHHz6w6w0tnrkdlXH2nUORC+ta2CTwrWwE/wmrPzP41rMmznPeNz5nAOE0BLzu0AoeCCaoeLT2e9xcF+Is3IbCYSvUZSWXTLGF3IzNPPs2E7VXP8X1L5RFRQiUHMl74LJGbt5OvV5ad6opay5sZeDT5YaHxw88PJ0owx2SBk6T55f3L9ymz2zlt+T2sX8uZMKa4vIdTo4EA5UuVxITboeRfT7wMA6sOKwuHNUYLBCdJ3hjPkVWGptGPuYjV/8GkSPWPE4ctSnshrmRTR3/8GVhTldux8rjEvnSDtgyUzTyHCI5jkjMnOuKtgOrdyy1IO2G//F3hUF2/vLrxC/ebWSD0u4QS6h6RDvDPASh6EYUA2edECkXwthABrTd96R2eCJ00SuVgnHB2xNaNnbbYPwWf1iQHNmDsA6VHFaivjCMheAbx/ZE3HYYIq0Kl/09f/OfP/D5/AEZ7mltoBTHf/2Xjdne8Z3bkesZ14TbIfEvzxUnymKVPCVcSoi+51o3JG6oFUL2TN4zoGgUftDEWTzHdWP2xhCVG9soTblIo1ivIRdTglaSGOIzpe745JlCw9ZAqxM1ZNZyxYKi1C7WMkPUkULCORjFSOLYW+BqgafF4wVKEfZt57688Kt3iXfDgFfB/MZaCkP2HVOcBFFDnaOXDbfXNsVAKQuqGecqAZid58YPvJQjkcrBZY5AdI5nsQ7fsYpznsbAWhIyKzsVVcNLZ2wcQqI1e91WKE/bjiSPSqQ2sGqYGkoP5TYnEDvXBe0vcomJthXmceT98ZbP18wcHOSFsV25842LOXKpjOJwWyb4xjz1quSaC21IiNtR+9JzH9KoJVOBrYKXnu9qDtbaWNzQhxSfSM4I+8o9G0jmsUXMHJoqz06ozWF0dkJujetemNLE7EYoFe88GfDBddyzK7RYyHtDbCZrZskJs74lFGk0gSABKw1qwQ2NUlZOKRN0RHUiBMHlnqez4NhawKsRAK8OFU/xGfUb3g14ywTv2WslhIw36QKqSVFzSBa8CsE8WzvyAoxhQWVCLOJaIbnypx0GbN/6h3wKqN94c1y5e7fxV1994lj+ljwUDlPim/cHfvkc+X//c+Pz6lmc42l5JrducCv7zs2gTCFS28j5Zet3KeJ6UrZZF2MMfV05erpkSFe2euK8jaw5dIpgEy4VmvOkGLi/maktU1tD1aPWG7yzg2kOlDqytZHn60C1ESfgrJFb4boZexwwC2ReeHcbEQlgAw5Fa0V8YseT8RQ3ARO6DoRlwMUZSYljCgTneL4Enl6UF71SdIXYOPorp3hhDhXWE6pPtKlR8oHz+RYLR1x4YYwbMWwgK0ig2pE9CwRHdpWtFh4/T1yfR379zcRplO4ad6BSEdVOrDGA7fWBMbIsN/yH3574+6cb3Nj4P3//hX/3/hn38MSPNvDRZn6+zjz/tJD+zU+I/RdMz4hVGn1DoeVXlO0dl+Ud69N3zPqWmca6rWTdubm5RSPs9cKuBZXQpR++rwFxYASW7Yq6Fz68uWeej0QzfN36FxRIQ6K1RtNGKRXljBs/kuSAXhzHpOiuvOyVq4+8aOKsE5tzJFcIvp9iCJXkGk4SeZ94fAl8d9+6zStEfDPWvVFIZP2eH//lW57XW/IkTG89x8ML96fPzO6PqP7EVjuZr6FIiIQ4onpHI2Gy0Hq0tj8sc2J3E+vgiTIjXBEU2RuHsvPr+cCVws9lx3jDxsjgIw4I4vDiwClFoMbEy5746Xzh5bxhLpG32k9z4kkGuRYKO5di1GysGihxYg/Cp81QM64/LkzvZ27kwmV76b8HJyT94+vfxhDrbA3X5zRicNS941JLjeQykixglkGhtUqIXeqkrXUr43zBzZ/wwzu2NrNshoXCeGiYFNZcuKuFUBWLnr1MPD19w4+f7/Hptg9X+5UqjofLxN//YWG6D9zXRtPMGGuHIWnkOBx49rCWQgihq4pVOQye1hozQrLC2AraYEszMSa8rTjLvBsgN7jkyipQpAdKgxZCrbTJ2KMSTTn4xn1tXL1A63/fup4ZA5TWMMCHRNOKNOEQHVGVsmWyuH7wSTMqB7LA43Xlf/+98O2s1Na4FuWqkFMkqDElkKXSzGivnBWjQ4ooQjXFuUoRoThhV6G6PoxQhMFDDAGrglrFe8WsYKIoEGPCau6faGs9GOsCqYc7yFrYq7BoowShiHRnRlPGFJjHhLRKbZ5Kfz5uuG5adYHf3B35VjJeNw6i1NwpfPEVlmQIjsrQuvlvmGa2esVH4VodW62vvpBCi4lLWSmaKc71jI51pXPz8FxW0jCiuTG2iaFsDGR2GuaMqj2j8bJknHM9VCiOUntTB2vMbmSIQnOdVJjN0VTZcqG5yCBAbqg3ni7PuFD6AJj6tYlK7Ide3SkoJhVtF2o11HqQcdDA4VqZRTk7zyax17BVqQSGCqMI782zRuUPl89oO7DICWsOQXHOECJahGQOj+O5Brx/w0EGMhGlX+H0i+Q/5TDQlJwb54tRixFHx81pwNWFbV1pVCStOPfAm/uVP1sz/yp0e9LTcuZl7xAK741v7u65cZGfHq6srzWulGAOgWCZVp6p9Nqacx4zeNk2fvv4jFsmzPtu4fKO4ia+PO8Yjvd3hrYMznPd+4skJCO6yDC8JYYb6hLQlhj9QDDXe93maWrk1qB61mViefuecfyItIrX1qfe5mm8rn4ZaGXA1Ql7CazNkeMAMRJuA+8PC8fxR767O+NC4ThlfHjGtHH0N+gKlu7JRVhahyXNX39idGei/xkXz8TQiG1EH7/GuwMmDi+Ri3/HT+fAUN/xsA80B8llNGeCX5nCSqhncBULhSqOWt7xhy+/4h8fvuYqN/z7v9j4P314IDz8AM/GG71F3MBz87w8GVpfSMcKDWoxGiC+Alcsbzw/NZoe0aoEFiZx1DXw6XxmngJRG1ILiuPytHH/HsYEJKHkjcwVdRnv+10s5bY3e9yAyUbTjGqHOjXzbOXMGD7iyg3R3zD4Cj5jaeBjPvD5/I6P5xmLM96U0Tsm74miEBpKBwhtxSNaEBZ8FKI3LEQcB86PCT0PDDJTNtjPGzdf7czhAdEzIQSkJEJwqHauALJx3QrijqxNu5ikJp6/JC7nRIsjBMMY8K7rvstWGOqVr31BEoQ68LfisDD10JrthNdaW2tGdVCtkBF2CQQbKdXRbMMNA/Va8TIykvoDXR3OOVIMtNbIEsjuDfsO++XA//Rfz+Dv+Itwx5waaRCELzjfkcsgnfehRghdOFSbYC0iPrHkiVFHWn7u1ibGfgXYHEL/bpp7YbyJ7PvE/jKybYINDqfaT9kqOJtw1imY5/OBj5/eku2E947BrhgLmwWyO/Hj9cg7NxLnSIwD1zOUzdM0QGmkyVF9/x77kCgIj3XhlOi1RNs5tMoH8/ykRg4JscxMY7TMzRT5uTZKM1YgOAcC11Z5KsZpnPphxXZ+dTrzx6vjodyQg8dqZH4dOvdWuiQuCuCw1nAxvNorAyIRs5HoT2RRtCpWK1Ir1nJnXIyRvXqkCrMrfDvDp6wYniHO7G0jmCE+8UJhGAN7M67m2BTEXB/onJDGCR8mcusn91z6ADSGRJSINsU7T6KDjASQV0y24jA30Lz1QFsMbKXSnBB94u1pIuhGU6g+YJa6Ulc8e6vcAL+cA6e6UCVTW6boSm0F5xIm3c9g+8Yotde2gYsbWXLgYYPdF9aWUcuYXakCmyqtdqeJmuDdiInwlCuH1PDeGGpkqA6jy4N218gol72g0ut7wYF4MGfQOiHT0wcoUuIlK09bw7veZmi6M0WP+IYhnPeVwfXvSt0Wai08SeA5CjE0mq94ClUjWY1sjqsGRD2uKvep8LNcKP6eVTtZM7tAaY578dzrC7/UC98fG//hofA79bhw6I6F5qmWcDrjqyLB2LxwcUdaS6wYVpQg3cPxJx0GRCtYJYWbzqhmo24Ll0U5hiPGQtONJitu/Mwvvj1wbIbbvlCmK3tUzAviIol3bNc3XEbPHAtv31SO0wsxnhEy5iK5TuzV83iuvTaxO2LxjCG9+uD7oFFVWZsgOlKy4sTjXejKZNsxDHXKdS982Qae9ki1gVM4MVhlyQuTc4xOGK13512749NP7/juuw8ELt3GZQ0nAwaoF7wOoDO+JcLesF3ZJHEhIvuFD7/8F96e/gFxL5iutLZgKM2ORP8btvIWlRvMeaw0vDszHR8Z3Y8EHhBfAEN1QMYv3A0HbuOBLBMfr2cu6R0/PGd++uGE1684jd3oNfqFP//uR96PvyPFFaJHwwd++7uv+V/+yw053nCIiW/vMzEWpERMJ8Y5Ig8LJc80BQkZ7AJVcYBvr58DuRDiI3P4BVeLGEZuQrFMrn3b8ynvfCVgCCFMfP5p4d13Jw7vHsh6ZSvbq1d9pGE8LJ+4iwPBBYQZ71bUNlwcOvWwVHCOTa+E8IwPrWuxY2GKIPnIx08HXuyI+ECMoKkDTrw68nqlTA4bKi607h4whdeVqQ8OFiM8C7euskruPvUFblLFyjNVM7hESrBrryCWYnhfqGVnKSNPy3vWUFmqZ72+xXEgTokbb3g7YxIxB6X0UM9RFpw2vj143u+ZH5aKtsAQIsEKfVHZUBUsBpoOlHDEjzNlrZhlmr0QDwdy7kZJZ8Ju2j322jj5ShOjVcH7SNETP70E/uPvlHEsvJ1fGFPlzdtvGN1z/717T2n99OnFwAynQi2N3Qmfngf89I7g9g5vKeBiQ9T1l7MYnjPDcSBLY3nwrHtnhyQvWMmoBeo+ojlQgvKiN3y6zNgwYTS8FbzPuFYo7ojKDX8s7xk2Q0rm83lir2M/0dVGrDBOyl4a0U0UjexMrLbikuG2HcmZD+nAmgsmjiiCieFaIWblYEICVuunQvHCLsKDeUYLqGvc18J7+UKYKs8vgU0S6mcCCigxQhPtdVVxZOUVtd5rz+Ib1hbEHfBhxqvj6bJxHbvGuVkFibjxgK6R2TW+s8ygiSsDrXUsbfDCPJ/4tJ0ZQ8WC0GxgFmOKkWiNN0NgHoyVRPYOhyCuMy9y6FyXWgshNG6jEvaKw7M0YXOQWyMTsHDE6hVflUhgEuE0jqRacS1Dbah4Rhe5iyPeHMLOrau8jR42I+Eo2qi+gu/bDMtCWZTowYthNVNC5OMy8LLNrFYpZaHaCq5QtaD/v7t+cwTvcOJxWnEiXGrj8/nKXVo5RsUWR2FkcYWFLjwyoNSCuErR+rp9679PtBMmg0ErlcjUtwJVcRIgKI2GhExVUIW2V6bRYdL/tp/zSkrCV2NgYiWWjKVbFNhNWF3Cc0D0mcld+RAHnqyjtNtri6kkwbNzclfm8om3Y+BpPvIvlwubBKaQqE04t8zJn7C69QC0E5bqKDpSrDEOjWKK939iUVEuYOYopSdqKYFlG8jckE4XmjVqyyCewkatnxiY0ArNGdIa6lxfl6hCcLx9GxhG5e3pQm0PZFv7Sh3PMN4y1MAo98y747M4Uk141+9pnEATj5JYrTBoxNnM6I5gkNF+mlLrYaEmXHfHUqFJ5SYoofShgiAchsSdh70oZo79fMvz45G725noNpCIaQ+1qXN4ZsRmfIm4UkhNOfqChchaM7o9QXzA/IKZ8Uq3IEij7TuNCYkjgcTgrkT/wCF+QXhCW+mUrwYmOzoo3q5EF5lC5H76yK/vLvzNPzX+/vcTf7i+5aoHDimR28bDYpxODfNn1hL53e/v+U9/84HH5RskHdC2s21nCNIpgLun5Z3gJwYXensj9lARgHRBHaaG2IpjIXp4Wa80CYTkWC7Gte4U8zQST82j6Y6znshfbln+Y+b/9j+84N0DxRx76z3dIApqPC0X1L0BPTLPFbGdEBLeeUpb8X5A1VHbgkweNQ9ZETX87pDdk8aAC0ryjeSsn8Kkgz3u5kC4TcxJMRdAPSZ92nfiieo5iUfHzGNdWN1E1oCVzr7wPvZvvzOC93hxeD9SgsfXBSuO6+WOxd/x6aGyW+JwCv8rYtePgbpeoSkqA6NXnO34VpmHM9+/OfOPZeZyaQxpJtKvxrzzICMmEQ0HioHUDbFMdcZWLsxBCfGGS35hB1pURCqzCm90ozlHqRM1Jpo5ljLyzx/vaGnkm9OVg1/51R7417/8EcdnxBohOExfBwIfWXfHNStPa+b5OfLlcs/7d57bY8G5DdEFrFfksIa4ghNjrwPPS6DK0A14rZ9cg3SP/L43LtV49LecS6RI5xxImvDugqxXVG5pMvClTQyXN0heKelETh7RVwvfbkDhEMH2Be8nmhux1qisqDRMG5Ir7+cTrTXS4HnZl45ubZVkE4OPiO09SYdhvq+9vxQh+JFjbZzqhTcYN+mWJwvgB4orRJQjnlGVpRRmNyNi1GoMwVNzv+N2tUFdce7E3jy//fnKcYTBdrRk1BvjzQB5wvTKJI270VE1sNBIBIYAtEYjUN3ENExYNn5923ibph523s84W8kV7kMPSmcziAO77y8nrcZXfuD7ZoRaKdVYfOJzrTTn2dX306rtuHzmfjxwUscUPLVslGp05eiISCTawOgjxTxxP6PblYDDQvfJEEZ8NLatUhZIBKLrYd5cKuYTuU3szWO+deqfVNR21DWaKeI8SRPptTLXrOC8J1tl0UqqF26j4dJMbgcWRnZrNMv9QOsCVruQySE4F7ECzjyjBqaqmPPsNWIucpGtV/yqkckYimmvQuINE0ca6E24YDxYJVmn7yYp1NpIrrGgrAakiBFosvNGr5y88GATF+7JKM1lIrCYdFJuXbn1ifvk+di01zdVePGRex+I1SFVkQwFT2nWs2IuE1Q6SfFPOQyU4Mk5sBRjl9zvrzfH+8PMt+8PtPqIYb2ZYkIzxST0wJFMKJV1r6w5EtyRQULHZvH6izFHLo7LAhpcT4xnB80zDfDdzYncZnJzfW29V2qAvRTwCS/C7Rg5pRty3Sh1IYn1SdvdcrlGll1oogwBovR0twuuy1c8TFJJ0gMx59rlQ+X4meBLN8DZRNMbit1hdkvQE+JGLBlb2VFpHCYYj7Fji9WB+h4Ek4CXiOfI9nzk+SUyf3tkmjdSrMSUCaH21kVIvMYdKfn1Osv1CVZ1x7YLyV34P/ym8OdvTvx/fjfxd19WmgbOGviHz2/56RqILFyugY+Pb9n5BhlPuDAQysYoO2wVOxuteqpFisnrureBexUz+H6HbOp68DFOiNxyqSO/Ww5kO5BcQ/yJjOHCRPWJj9eR3/1oXMoRkcTN48Yvf/3Er3/1e1yAZCviN5wzoNKyUvJXvHy5ob53TGkjRXAuUUOk4Kk6s22RbffsduAQ73DFwTrzVgKbFPy4M6SV6B0xREavzKHSNLLZBMyYPwKf+oOk7v0UKAvzYaPNkX0dCWqIjHx5TPzy3YDEQiTg1dPMcDiSmzDzjLZTZOI5Bz5ujufngS0FrGTsixLjxp9/G/GHEXdZmKaAL4Yu3YMQrfDtdObd8ci6dQpnHxYUiwFvggOcwWjCbL3atKqg5li0Mcb1/48H9j2R7RXeWkbKyHk48NEczRkxBXy45+fHBRnfEUph/4PyZ1/fM0fBERBRfBTEG87fcF7v+PJcWVsmiudlP3Gxe06XF76+f+A49MyAE+toWgts5Y6Hh5nqZmz07LoSJBHdiaE4uChr9jw64bkFtuoQ1yl552KMwXN3d2RZd7QZ7I1KpOG4iiPfeNreEPX44FF1tJIZQ2QrmRBGRBJbXVEXUCldDCTKEIwohg/xVQ+94aQhOhLEEKnQMoihNFaFEiZURtAr0TaOsvEmHSi7I7SIM8VbIIWutRaDYpVrvTCo8mG6gTyzL8LFCpe0IJx4WkZ+93Dhq7eJ1RTdFJcKx5Kg0K2AzvAxIngODo5h47KeCQESkDCOQ+Td0BjaFXUGoeIEbqPx5xFus+PTWriasTVHw4OB3zdObkPXDXAdH2+BTfuzWRt4TUgbmONAy4pvFTHDhcBuwh49RfvVUlOjMVLbzlodgzgWPF9yhV1pFNYdCkZynbyo2t8bWvvv3flILjtohaBUq7RWsNAx2Zhx8kdqbRiGF8dhSAwSib6iPpMTXOvIp6tnDQGRnksYLTAwkuuGWsTahCMRzRAETyafjeQ9czK8j2xWKaYYoYvx8t6lRtVT9kIajPtjv7ItDh5LwUnDJZCyk0JB3EJJA2vIXZXelDfsHE0ZqPyUI3VIVDGyi1xbQocD1EwqmaN4vrjYA+0CFzU+aeZNMKjaK6JV+rhihbF1zoX8qdsEV3E0gbUau9CnTJn47ZfAL+83YuzBvZ4k9WRVLmKYHjA9UEqmemFTT7ATRT1qjpdtYs23zPOOOY8bEvoa9mrPJ6x6ctq4f++ZvENMiNK5Aq4qNyhHH4mhcjOvjPIZpy8c4gtpLIxpJgFNjXGAU6i89Ts3srMhXF4SZgNeIlZ2rAhIpPpA8R9YW6PtE7VkSr3j+XzgvE34ceaX44HmHX/E8zefMlUdU2x884vEn53egjmcvP77NKIys21fcXn5FY/nDzz/WPjNr2EIC6IXLPe+v0pFPMSUCOKpr27v5h2tBcxcvwvX33I4Kf/+Xyvf/Pie3z06/vAcWLbEH55PaHuDyommBzRMDGHiNE7cDANBPkLrmQJ1iqWBvHTMctlgf5lJNxNGx8MWnTG7YX2Z2PM3/PDyDf/h4T1XuWUKlUNSxHZaUXJWeIzsbSAON2wZXvZ7/vDHA3/xm8Y4PHGcM42MYdTWyBaIFmB5Q91P6CBou+C9Jzntdr9txG1Hzpd7HrknyIxbIdfAHBJWG0E3fvG+klzGo0S9Qi5Ud4PqG37+Ah/uZm6Prjcs6OtZF1bi8Rm2A6uLyEWxlvj4dOC6nTiOPfyXUuz/nZBgMY5ieIyrv+GftgM/PCZqHmkt0PBEeSWX/bjw/iZwJyfehsZwAM0XTPuL/uu08efHKz/9PGEE9rbh6ffNznVnnmtXjq4wWkWrI9Se3q4+8JR3pjGStGCqVFMwZS5GIyBJ+wtxM8LQq2Z7U74scDiMfN7f87K85fZuoLntNaQk/eXc7tj2A3sNLNYYnUeL56ePnj/KDU+PL7z9Px7wzuNMUG5Y8jf88Me3fFyP5JhoomhzwMTsjelpJexdcnPxkaUmqgXEdQFMtchLvSO1ldPoWJ4XTtJ5DqZGcAEXjTUqehXiK0GPal1l6zyhNARhd5EtDrj4+lC0Lk3yXgkpotpQjCKGaWOgVwKzKiqVasqGcq2Z4hPNj1jp+t5JlIBDSPjgKVLZpGejPI1Q4KSJqRa+BoYw8FQKFzz/IpU8e0oeeXx55v23R7I4lmuBy5l/dfcLtidhL4WzGZd1Y9WV736R+P7UeJRCti7GIhjH6AgVWq5UX6lO8c7RckVMGIF3hxHZdlppNCcYoVv1PFSF4j2rOJ4l8aV5RASxxug9TickRnzb8VjPzVCx4CmuB1Rraz33o57NAsvlwuQjdR3ZN0+THs7DjcQh4+nSuT17BMNK68xSazjbiP716oSGE+3ODA+lVpZ9IaWZ1gqnyfPtITEYePWE0fOQCz/tV1bvaa0SfWaolUNTTq2hMfHIwIWBLIJLni/7whh9rye2hmrDB4dpJUtjp5FpZKdIU8wJqkrYM37yjClyrZm1FT5Lg+CwCgM7kjYuFGqAZp6oI0UjvlamuTGuGVGHc/1KqoWZRXaSH1GvzHljLr7juZ1xVkWqw7RwkExoK9ICmONcFdWA88Jof2I3QW1CofQ7+KI9yRkbRQM5S5+mXSK8WsW8HPj8Arp1k1Nxwi7KZoFIIu/CskPWCZ7vmcadX/9yhrCyFaW2ieQTrSqbHvj8bHztpQc+xPV7beeI4hj2/kCd4oLuvwUemMaFMTWiu+PGT3z3tvGSz9wOEzc548rGGkcuhzsel72rLs2zauFRdz7TCJeBl/YNcxwIJjw/JVYdaTIR1sIwCiF6HuobflcObE1hrfzd3154rO/5b//yjmFs1Dbx+cuJ5fIVOX/Py/UtezlwSpmy/0DkpadSEYKLnY3vBR8GXIuEQXqCtmy0ZoQwUqX/ngPPJP93/NmHZ757+wvO+R3X1fPzE5zzyO8/ZV50YGkra2vkvX/pfzyfuAt3HCal1R484pQol4G1BX7+9C3T6RFrz9QsqB7Rco+uJzR/TeItwoG8N+4OMA+Zsp1RUUxnTvHANPSfpRjsFjkvE8EaUc6Ir7Rs1AYuCGNyuBKo15H1/BZxnqqPqBROd2NXl2oGCTzuR7LdkUPipsFRV65t5UqkaCKNleSVllei5n7SdUZrld0f+PiUuD8mrF0J4jBtxJSJ8wI3hm2RXBrFCl/OkZ8eT/zy9ECTDW2V4AR55UqIRfZ95lM+8cNnx2N7i3c3NG1gRhHwpXL5Ag/VE4Abt/DX7+D7u8q0nLkZ4E6ufDs8c2TmWiPVAy71oUIr4mAQZY5GeqXAxVpwCotLrBXKnkmTIbYharimDGnm0jwXV9jcSI30awN9zSM0R2mNrAPLdsJ0wrsrLggm/WWhWTlMhXmGy3Vk04ayU0RQG/j5+Z5//rQxzjv7Xnl+OfDx+Za1vsfioVNCTYkScDKTysqclx7grcIu97yUkSYjZoJznmqAO3GtwjwrY6ikCtL6MCLNqEMju8weIlV3nPQevmqjSEG9J5lQ/MDZzYwxgzbctjK4QKjKYehhuN0Ja+gP+NEUb5XkpDMMRDDneN43LtORWxcJWohNsZapmnCDR1HURzKtE/VeV7rBJ2aDYAp1JXnHsRbuS+FTK1QnlJa4bluX5TjlvHUHwHh7T70U6rWxZwOEennm3YfArEeeVscq0l/sZhQL/2toTpyAZqpK78gTqEVJ9M+hvCblq3pyNWgNq3CwRDAlxAmk38sjyi6Nh2XnEBtmDZWKoyLBEwLsVQlecCh7Kwym+LwwhcBNbrx/DTietw1cJQ2O0cPZHBcCAkTrrYfWCkil2Y5zDacNla78juIgejatWMnczEfejQNvI+jWSaBFHQuORZUjK18Z7DmjyXOMA27RrjUnUtSBd+ymiDbO1hgPE7YXvCi5FiQE1v3a+0DRdw5M8NRSmKQTTk2U+qp0rgqLCI/ZmJwwDgZaKcGxooiL4JTsB6Qamz6zz0LJjZPdkgxqWdi9sOuADMIdmTpFvlwLSmJziRZvaRW+jcoxXbG9cS6RxR3ZrFeUvfyJ2wS/fnsg10jxDj9kjrMivt/VDWFm9COEAmGkuYHrw8j1emBrHi+eQkMj4B1byaybY2+eTR2VEd1u0HRhHl8ffG7CJ0dxjqU53O7xe+Hp4qnRUUhUL5xNOTeYTMnljNgX1D3TuPaHWTVEDhzHCN5zkDv87pFSEb/z5k5Yw8jbaeQ4BqajIzWPP18571da25huJ/wa+PjFeA6O4yFxO0RG9wJux7cbvHkcQnGOS5n5Tz/d8Jvvf8m72GjlRF6+YT2/pdWvyPnI+VqIo2HtSGWmlD7lmkq3wlntvX4JiBWCGBYgpISp9auY6nvX3S4E/5HBe6JE7sIN3xwGHh+v3Knj00tB3Tv+6Snz03Lh6oWchd+n7xk0c5SRISjNlJ89PDvjP/92Y7zdmf1njmmC/YC1E2GLDNzz6zv4v/zmZy7lka/fwhgXtnxlryO//fIOcd/zsCqK9RyJBJbNEcxeE65GoFu8TLvIsZZAbhO2HZkmhZbYWsUl4xB2hMIwOCx3AFSUym1tuO2RWhw5veXz3tDWSZJxHHE207ZAi4Fq0mln3ve7+7Yh8lqLY0dmY31JPG+JvTksQJHIlXt2+4lWXrrAyPoXW9ShOvPy+JbHi2ewgTEcKCUyuEBrfR24mWCW2J/BSuSTG/i8Fv7iXvjrm4lT+szYdm7ahdu0s6rSxIMznGx4hUSiGYDQ+j6e0zhy3neu1TjqiKyF4jN+arhmNBX8eKDlQMYoraJmBCp300byC1PceXeIfJgXZq/UcsCFK8qCSevhV6vcvLnlu/d3tPIVVSKXvZGl9hqZHfmf//4E43dct0ouM5LegDsQGXq+wgIxeFo2IiODOlLMOPUd88pEwffwnIuU1vBxoiksZWOePOX8zBx6HUwwQs7cR8/iM9ki4ug0PC2sCM07zDx7NKqODBy5oTCsGwlljJ6hVsaT45M6snT8a3IO1wzvHClMrNpQPFtTHsqVQ2u8wbjdFy56QsOMJMPH19S9NqoqyYfXdoPrDQvvKCVjIRIQprwRykoLiSIDe25kHMv1zGDw8PyE2146Xc/fEKyfMn3NDB7UR6I59lcp0laWThAUMDqMLGqj4VAvlOpp1QgiTCGyN6hmrHjO2vDSvQR4z3sZaGL9dyICGE0C5711ul/u2wFwXYYWhCUYNWcW64bAQGXQ3LvytuMtgE84MVxR3qwbJ6k07/k49JbE5ABTRBQlI77ngoL1v2WIkSQRkwCkrjEmch8jvl7R1t8xay5UCQR1fN0WUoXn6HjIhTIIY/CYdf1m8N1K2yoUH3mpGULBD/G1MWTsbSe5gV2VtfQQp4pANHKt3MWAClxrZXGBpopH8KpsQZGTUGvhRY2sRoxGODo2uSCWGV3jvVvwT3DDDbP5fpDGscoIofHBNW7Cil0zL3qixgNrm9hFmNuVXw8b4eUJ32auGnl2DkIj/qk3A8FqVz+Gxrubyt1U2LVRw8Aw3nCI9+zlEcHj2y2u3VPlls0J27Zh0feqX2i08voLdZ4iSkHwYaaZsezd+DWcJvZ9h+gQ8cju0RK57MJThc0HmhfW1lDfYT6P588chzOZFRscTQ3NhtkZUSWmRBqOyDWBRVyA4AoHKcx6ZZoq4jOTbdyfCtr6F9tbRQcj3o3804tR1kyTwGVPUBYOcec37zfObUVNeFhWVF9QvSXoV+wreJ2p16mfwM1jxfH0RfjPf3fgN//qO4ZD4Gl5QnVCxBjGyMFF0CuOR8AQgWoNNYfWgjUIwSH06xOq8eXJ+PIEWjOH+cSH25kP8sDMF/7qzYnfbQuZyHXNPF/gMzec18ZkA4Pu6NEIbqMUY3ueOB3v2R6OnPN37JoIOhAV0vjCv/v6hRyfULuwt503p0Cpb/nh48LPlzOb3iMuUeiymO0KpkeE0FdrThijIj5i9i0/P77hSd5gF2O8PeFdIHmQ9tLv2NyAU2EYG60txOY5rJW5PGOb8lxgGo4cJ89h3AhiuPSGK8rKTrJH7qed29NH1F4Q2THXaWt4I5fA53Pgap7xduitA19xfmfb1y7KaaV/ya3hvLLXwMPTxJLhOAmPy4qXgUjrQhWFVaF5IalHc0AIfGZkVeE0RH49P+F25Vgr396sfNbCrr77DDQzGLRdevhvq0Rv7OtOMWHygbEURI1qI/uSOc4TFgDvOXthiwVXN47hzDhE/t33jr96/5F3pwtTemEcCqYrrnxi2zxunnAsIBVQSrvg4u/5cHrPcNdoEnnMiR/WgJhHxNjrgeUiqPCqQD4gMlDVQUhUg7ytvIld4R1NOpTnJtCuBfWhlxpVcc6BOSpgMlEt9K3k5AFFWVEqIo3RQRwge2N/RegFbzir1CIM/sCK8IRjcidEV2LkVcOrJAvgev/+QODFQfSd21EcCB7oMBgFrqY8kRGpxKMxbZnddkQc8+Ap1sjnhgu+98/dK51viGQv1KjkXDgGYWiVwIoMARh4fHrGDY5qA641rnujXkBZcd4TygzNYQougJsKJ/HcDpGtXqjuBW+ZyMboFerIthxYKjC47osRCFRO0w3rktmaskfPzzoQfO7Bxnxmvh15oyMXP7GpUpripW8UTBxbqx0D7oTaNrYNdtcrnUUmmoPglck32nIl2MQ4HDhbobiu7x32zI3Aw9aFYYsltlI4xEQKAVc9W9upWvFuwIX++dpaz6OFNOLdiDRjCpWWN8yFvgExjzphEOOoG7451jDiY2JV7ZRAJ33oo89QVQ3Ds7vEooXZGrMTvFZWAkECk3d41W4WpRIkchDhW2ncrDvLdOCfqtLwhCBozah1KFWV/lmpwaO+f1ZVSz/8+o68Pg6NZAdsj0TzXDchhBHLBWHjdsz8MioPND7WzMbE5mcu1Sj5M3d+werCyEQNfUPiq/xph4G9GltRxlPE+ZUvjw9cqhCio7094Ih9/Z+VQsT5mZAiQ65ECexleQU5wJ7hWsC8oeIwb4yDdOWmRar2H8Jw7LWxlsa9GjONW++wOFBN2VvFTJlC5SZV1rYyOevM8rKCT4wqr6yAipNMLV/wLiLDgZA8ZSnIy4JPF/LyDONO8xtGg+axElnqDbsbkPnEV+7Ep4ed33155Le/zXz3fuSbW89fHR7x/kKzcx9q8pk36ULAON44Lk87+zZwrZG1bly2gC8z5/MbxrsLX//yicaBzY7s2dEuwoDn2/dvmEN+PaEJa6sYgeh7YEl0A6z7GTTz6ez57U8HGAYORH5z67jxlZBXTvLMv/1qxvsuSdmzowbPjw8PMHgcT0T3xBAvnI5P3N4s+O2en//4HT9tX/PgGk09cclMY+Wr7yLOnVHbsObYloWtKHdx4oHMY6lU10jJcTvAnU+U9cgwp84sMCWYUPUdP/zue37349c86y1SGnzZmCZjnjMHrVAF1URdN3x64n4y4vXI4bkw207Wwg1HPl0CUhvUCy4auCfuPlRO/gsf5BMSX3A8k/URpxkwzA0gRimCSeP45okpNtR2XPzM7fQH0I+YFJyPSAUvgVaMdRfWHEECg5YOigozQcFZHxy6w82oW8XViCPA6njSwo/PK5dTX4EfKHw3bfzTNbOXA631k662DflfMzaKOWVblSbgvWPCsbp+Snc6cQQOc+VglXF45t39lX8TM3GICDN//W3h6/hfiP4TJheUQhNP7rQjmvaXmKlgCOIdjsI4fORmcsQpMdkd7eUN5WHD2kK0M9BQGQihr51bU1wKZFN2HOYTH7fCX4w98AYO9QGbHEF3DqEQLHLNO80dKTp0zj69JpydowQlisPKlb21fhJsCjQaPZQWQofmSKn4UglyZPegoVBE2KsxekW23AFRzUOEpMLgPFN0mFX2unfduoH3A2ZKbjtbGvksgSCCToW6PTJNI5KE0TkOS+RaQbXicXjz7E44x4HiocWdKIZLmTEWDpNQMxRN6F6obu4aXedhmilb40hD20rZKy1fSENknHcGP1FK5uXyQpYLeb3ibMFjeLlH3S3ZBUatBEY2yXgKWAPNzBGUxnM4Ag4vwmAZvTxg0fB+wruJVhzJlNEa67ozppFWM2INJx5lZy+NhmdXqNaD2YGK7JnoBnyouNCdNoYyp0DKK6P37HtlDYKzHgq9iZGjH5k0IwlIPbN2ySunaDhzJH9EwkyowlI2QvTsNeNcpKrvwUJrJAfO9Yq04qhO0Fw4xMhoXblWa7eEu9Dzbsk71Bq+KYM4pCniHWYOL57oQ6/VtszshVtXuMkbh0F4ypHVByKF2SCVneaUQkbd69CKsVmlohwldgJkK6TomPyVvUSMiczIxQxxxuSGTlaUHa8rb6aJsguNBNJoxRgkU6kkfUH0hkp41R7/CYeBvFXwEz4qElbcBL5E1DrK8VZa3wr4xK6OMClfT5WxrWh+IYRGFeFpi/zXj46VuVf1XCY4Y/SVXFoHXQgsuYtgdjxrFT40xW8XbvxMFc9wGnkqmSY7d4eNwX/E/JkCvTvsA5WO4nQ9qoq3SqtXFs2cX0bWCusCQ/XEUMA2trbTfEQ5YYwsdeC3D2/4qd6gUrmdHHfTxGkd+I+fdv7nj2f+m+8L/913Z476kdFvhAO8eyMc5wDV42m8uUs8v/O0S+LyUln3nVYEF26o7YZkDXEwuoHn5lhdYCkjP3wxfv3+DcG90uusULV/AUMaoLZ+gmtXag38/BB5fk3H5m3jN4dKJJNi79d6C4zeEXTBlwvpeOGrP3/C7CewB/b9ma1ecb7i/Az+K1o5EGTmJMaeM3n1/HSd4aj84vuBGM94HGU7E/0TH45CW9+Sr4HD7RtOg/LduPKLENCnA7yZce4M5rA4sV6/5enLL6nljhThZMLhvHN7dByPjnGMiN7y/LKyZUBW1pxxZkTfNymzL8xUhuqoa4XBegDINo7TFed+oNrvaf6CxQhq2FJ6QKqHqhEqY3ohpBfi/IBzT4j/megfMatAQpkJ4Q6tI8XueFxuONdA9TBSOM7GIyulue42p9fZ3A4jY+/6F8NZRKuw7rBqY7VCisK7uPJmKTw8wbYaozemGBDrxLgswktTNIb+/9wUvKB+YK9gtfD9uPLX333h7fTCm8OFOZ0JfqfowNP5A3fDLQdpIJ5cA1ve+xAeZzyC6YZ3V6D16ygxmjRC2knTBR9HRj9z2A0pBV8XjnHj7eg418b+SnRbX2l5VZVFoWJcEQxPco4mkaUFzt4jIxyavdr8GtcW2GRAidRSEIUxGofJ07JwmA88nxsZT65Lz3D4AxWhtv7XHGMgqVGKkfxEwGNi7IORU8G5vomUIKwiZPGMZhxyZaQHua4W6Pd2AaxhJIieZ82IKb4thJA4xcggQhUPbSRfKsMQiUHw0qvOTQPFAkp3F4TQmGZjHBqbgWaPqVAVYjzyWBtJCh5lrAVzAz70xL0PhtgzUjPBJlyteO1OBdQhrqBtRYcrzQaceWYXOsCqLVzXF5ztxNqoamgcQCKDuyHmFacrdb/wuHykhVuCOqIsRMusKM0a8zTR6oIYJIRWdiROr5sZ63f91njVDeC081rEO3zwDG7AlZ0BJdTa8yLJQxVqrQQ3cDzc0UypaeBpXwgh8n4KHPAII5ub2dTzx5cXyq1i+kR0nmyBohXnFZwjeYeo9pBkEJrzaM7Mw0ywwLkYEoY+GFNx4lAtVFOCdwSk559Ugdifd33cQPpPT8DRrhvvxjd82Xcmb0xOCaY0H9ml9GwBglkPQRqVqwm3EvGiOMsMwxWNkZIDV43sHmLyBKFvSqy3S2rKRAKydj+CyYBpY/SVwTZMMvXVlvknHQbqqoSDg3rB2hNNKk0qtRUe1sy7YaDljZagxUI67rzBSOdHmlx6BtM30v1MmL7hvE9ElOgrhY1z3nnYMvoqGwpWuWbluVWCzQwasbLR2k6RxngDf/muIf6J0/wE9kda+0yx/icy13ujUxwJOVBxnR/tJ35/ueUPX97zlIUQAn928JQp8/Me+I8/wKclEuOROSXED+R6wOJEHCouPBPsE1/fe36/3fDw8jV/88eFu8Hz7XBlnuHDjTFHgwZlARcjgjCNV977xLuD45PL/M8fK1uZua6RQaZ+ly6eNCfWPPAsRhHlnGemeKLZFTWPSGDdIk/ZMQ2BQ3SM9pafPn7Nj58P1Cn1E4kIrlXYF3IulAb7UCg5kfyMmsd4Ymo/I/kjjZ10GGh7Y9OVUgf8Hri58dz4QpIXbLvws4N/WI784Y+Vb74ZmaYV2s6UrgSXGaaRr46NzzM4vzG7C39++8DteuX60z3D+1+Rbh87AlVnrH3g/vYd76eJeD2Tro/ourL+bBy/nhG/gGscxxNSoGnG+0StgqUE4hhSYGpGVOXlS2M6nWhyQ27CVM7cHa7dM08lt/YKphJKUcwq6gznMtNwxtIZib9F5TONtYNGJCCuPwiEiceXr/j85T1/+PKWnRHfMsdp4EY7IjZLYXNCwzpVc404czga3jlKNYIFXh436l8ILVaiNe7GC++vV3770NhtwBTKviJewBUIrm/NzGMKyffq1wa0NDGw8t/+4gv//V/8L0T5SNMre80UMQoTaRRqO+HiDVZBWiToketZOGtkfzbef/2BIxegK8L1lZEhYWU8QWn9ARgQBvW45khUpiCkNPJMYC2KDoFilWKCGiC+q3idx8eh8wHcQCs9nDw432VUB2EIiRUhirGdd/xeGL2g+4VxNNRHwnZksYxj49YXolWCBR6a4DgwB8/IBm7ALYkQISajuchLK4S5txNa69vLG6vcVBjLzkThJgU2l7iYEfBEgeAi+EjRPkjWtjLoyts4c2qNf/xYeHmZcP6A04bT8krscqy1G01NBsounMLE4RDwh8hjyb0+mAaiZnbd+bIp708D87pw50dKU84O8mZ8+bSTXMbtSpoDVXt7JKROHxQtVF3A71yKcDk7tvIRjcZaLlQX8KocckUalC1T9sqNiwylELhiKbCpQ0cHrVLdzhqUrILyyk8YDtR1wzXlNIxsBqKNSRtj8lQzovUa5xRdv26OAxKHTgJNAYr0gcdHTAvadoZhQrWSd09ujr0pq3lGhA/xyFzh2gT1ASeJVm54eMqcpgO02Ln9BuoKu4/MogxbxWvnHWgI7K3iUN6PnmP0bK2hErhsC1ihWRcx4RpeOtDL+v0CIq6LO8VTpWPia/DsVcjRXmv2G0JBgpF9V0OXZv0AIj1wKU54KTvvhkQ013MYtnIcjMveKJYobsCJomLUWqnaIBmjrgxMfNFK8ZkcAk06ZnsMGecXWhs6COxPOQyUKlhuyPqMnS6ogja41sLvvqx8O884hfN54axXDsfEu5gJ2TodsGaa7VS/k4bE94c7ZN2Az7Swcns0vnYDe6kEhGCFLQifN0fdDsQ2UhHWZiwq7C8Xvv9QuDldGfzP7PVj57i3SHOOva00dYxxxu9QywVCYMlHfvvxht9fb6hDYFTj+zjylDP/6aeB/8fHCU1HhjTApaLs/PLmkV+dfuDr28JX6cq79IkogR+u3/HT/hfIfMt//fxHPsobYoR/f+g61OtuXJ8PrLlxPN3z7t03HM8rJ678xe0BS4n/5efI8/nAjw9fMw1XTnPG+TOnqRLiytIK1WaK3bPnhvcBHyZSPKIaeLqcydOB2/or/uWHd1j4CrVA3jccFatAUTRnCCN1Vc5n4VqM+TTw9ZsjgzhUYx80mjHNE9E1rDgePynr45X7NzDEF0RWvjrMPMvAfhX25V94c6/deqYZcZ6tRNQikgwLivcr98Mjftt53v+M//LP76jTI+J3tFSCf8toiUNZOS1P2MvPbEvmuh55+qNw+72hWjCNiCRKU/bXayRLAZnvsPYJaxveGc/nmXiZkeFEmALX/QdOhw3zrYf/MFrbXtWenVuh3kgObk47Va/scqG50qtManjXCD4TQ0XyxOePH/jdj7es9UTwwuA9xSDFwDH09XqpvZvs9x4+260wiH99KYJUz3ZpEALpTvHrypSEd4dM0IwyYRJBgdKQoK+pZQevJ+1sO1M6IhoIPvBhTvzldzvH+DOtPSAhs5YFFWOeb6m6UnYBO0CtHbPtHa0qtR54fDEsBN4c7jB2tG6Y8yAJiYrQGPWWst1iGhhcf6EHczgtmK00m6hFKQ4sjXQGn2Ao7w/GITS0CNl2hrDz3gp7CRyaYDWzWebu9I7RIuu6wcsTriotCnH05JY5LxWYGUPiaC+8z8/E1ght4NLG1/v+nRB30NwlLxoRzTR59WOg3EyRIQuu1O5ZcA4fI652aNPX88Q1F3bt+mDnlKqVkAIlbx0qpBXXdmJxnB8jGk50sG9l8AFjI4uCD6gGsjmyJkIxbp3x4dZw6wuXXPD+SIuJcwucW+UK3A2Kbk/4emByA1uDbR1I0UOOrLogww6y9cAdrWOFEdZr4fyQ2C4T0kB1IU0REaNYo2npJE4fGAewdcFZ6w6EOFB3I6tiKNE7QowU64IqqdZfUBZ7pVAqMSSKVqIK0xApbmCQK63tiBsYfeBWG7fjgDfPMqw8EdAiBDUcQpTI0XtcEfaqPWDZHNEi90545wYoBXOBTY3VGkUC1PlVLNcrpNEFakhs2npITyLB6IONN+rrVcB3c8HngjfH83UjWAOt/boZ19sBAuDwbujZOR/79sMHWsns0mhhoqiy1Qqh100dCq5vupvr/BwBqtb+HDJjE8fDthOHkckLqsbJGbXtJJlYX/mWIkarjYMMzHVj0pXt8hHaPS+zoMNIWbvbxNfGRCDgUP0TZwa2WuGyMsyd/S9o185Wx87Iy9Y44Mk1cmkOW8trGjkg/tgxkNZo0jobuj7jc0DxiGukuDGOG/fzTLsYWipjqMynAXdzgMc3PG4r13LlsRTmUvBuQ/QJswstbygbfuiBJguBagPRRSQI0SJNFF8TQVP/w0rCW2F2wtgc60shp7fIcE91AXWVD/GJ//tfvvBnt39H0J/xvODjFWtv+OBOHYhyH7g04eeWmNpXfLfe8Df//IUvdWK1I3txhI+FX70N/NtbuE8Vr5lvbyL/8Ame1zf8P/9emU47f/1nn/hq/ojTJ6KruF15XlY4OJQZVUGz6wlbF7m9+aan/PmO2obujxdDggMVnraVmxTxTvHeqO1M3iPnJfHzVfnjFf78z75lkplAZsovaF1pTvDxHcv5A48vBz4tZ371rvEmDYSUcOvOEIUpCZQrtW1og0sOXJaIqRHHSHbKNBvJl77OLpGXx6/Jy3ssLCALIQhHWbk9r/h1QXNvT7gKTz8Jt9/dAAsmO9UqWQJ7LkztBfEOZkfbB9RGINLkhs9PhqbAoW28/1BRaahVVPo0rtUw5xD3Gr7EIV5w8kKwjaKd52CpI1wR8FFwTqh74sunwJon3HGilUpxieZeAVqsqC/Mg3DrQILnbx8rm0V8DL22WAtqSi7CRqDNBQkVJwHDIcmhe3fXO4RWV0SVwVx/4HvjlIQ5Jh73ylUGxMM3N3B7uOBCxdQotuCGnejA4gvNXfjyfGWKJ3ybwE8stN62aAdKG3h6aZidwJ6x1sjZaM7RrDLPkaF+xbZ+haljOhhskdgGglfWXLmGhiXP5Bs+eDZVxDW8Vj6MlVkaRRzZG7jM/bRzqQFnjSYZM2Fip7aN/HTB5848qKaYN/ZduG4JhsTghFsXObYKizJ7TxCheE+V2jvhLMg40NaMNI8MQhXhKgHVnbvkiXu/h76aYGNir0I0x1gqfzbN/HxdqQ7EGWtpqEjv3As0XRDdCTbTasClA+YdwXdtT4ozOOMpZ4zXv69PWA3oemVm49ZnlrayXxUbDhxP71j2xnMGtzkOXjn6J1qdUKus58ppWJHqqAIuerKMLNfCuhtNE6VMvDyPnF88sWYOcaBY5GFZuIRMdRVN/WSr+8JdCExz6K6LMKGuB+7ERZo6rjUTXBchNfNorWx5x7nAplBaJR3GXn0NRtHGdjiQ8kpVYQ0N9Z5flYF0FloY+Gg3/IsJ5j2ewiHAUZR7P1IarEB2ieoDdVk5joFRhBIntGnPlVH697h1EdRxGKHtvSrpI7XUnu9ogQ1jFU9znmwFSuOI0mxjFeNJu1CvqfQGiRcQ4WKe6geEAW3GddmYhy6xC8FTCZAmgjamJiTxDMEzWMRbAy0ECoMTirROmrSKDImlGV+qIRTeDgNTo4e0y8IxjFwsYk26mIuJJoVZA8dcCG3jFCYWG8EliiWCDTjxJATMXtsgf8JhoLOYBdsFcQOiG94gxkBzjmxKaDu5TeSm3IjgrFBVyQirGyluprQrEhUvvR4Yx7e4oKhVyrZQrXKwe7Y1k7mwx4WQXvB8z+4mFipLKdyGwuA3alnAZVo1GoXcrqQp4BwMMRFDxNqGtIx3cJDKX311Ij3MfFx3Zg+/CIU3l0/8X993K+I/XDeyJXzM/OtfLPzlu79jaH+H1Q3VBZwhaeSv/83GfyzP/L/++IUcJkK84SYa//DlxJcvjmcXyXFAJeDU8cffP/PlxuO+HviKnfPZeir2eEvhhq19YQifSHLFScfmVipIQyWR3Ing+7+r7JGsB5obCfKGFA784tuEPSw87cKFhPORhzYyhcboAkqhtohPnlB2rER+fP6a3/7NkbvDB072E//dX83MXNk35aW84dPzWxbec673LE/P/OX7nSk6NO7EsHC6KcToCGliF8XbyLwpb4+ZfzxnMoFclaKRYJljEj5Eow5wrcYuPaAWXcZPif0itBooTVEJPP2YsH+sfP0Lhw8r01FxbeNdDJx0JDxmrJ0o4cD2OEMbqS6TnCfKC2/GJ+6PS8eHiqdRoCmC9BO2E6IJ3nmi60pTkY7SDT5SbMNFhxcjhcjgZx4unrwknJ+oMZCTZ+staVpRrFTM7fi486v7zLcKw0PiD9vMgvBolVKV/y9r/9GsWZadaWLP2uqIT1zlKiJSIgEUSsCq2tqaLBqnNOOMxv/MCSdl1dWkoRqVQCIzQ3mEu1/xiSO2WIuDc8FxD3KaA8+IuNfP3nut930eh2A28OfPF1I/ENhxubzhnz/1ZJ/QuIVk1RzFEpodb8fIrz90PAzKw66RQuCHZ8cfXxZaZ/z29pmxP4E0XBRQsJaRAC44QvB8PiUubcThEDOaaxszwB9YrEN0hyfQDLR6wFPV0bzjWhQnO5zt2HlD91CjoHOP6EpTT8ZIPhAlUeaJseto2tj7wpu4TWRaN1LqTLENPe2dkJNn9RXnO3K7kC8Nu1RiCDRp1OgpZiyLpws7Vn1FfGt81QQvQE9MjosTWhiZLdNCpbYrKY6o2tavd8q6YZ3wmjlGI8WEn5WpRZaU0NbIDrTAgx+5tAsrgsa0VQbZLt1+2OqDNhcG8TyrR4PHO8PpgpZG8xGzkdI2aqT3iVYc+TrTpiuHIfLJV1Z1OB8oVSlZWRVe4i2/18CH8cyoK60VpsVwKLZksh/5QsfnVXlZthwKAjSwLOxe6aquXRBr4BvPQZmDbVY9NZI2HiTSi6eGHS0FZhydSxs1MghWA0teoPe0qphASo5KxXm3GRDN6GNHVbiWzMca6O/vWdeJkjyuJPzFEWpAB8eZxEUz2Qne7YBGcI0+9kRxtGpMRObqcPT4bvOOLBWmZkx12+87hSSeVgNtA1IQgpHb6+hd2fJn3m/GQ7eFBqM5jmuDJRNc5Un7jTcggUpgFSjiWSxyWY1KAzy9T0Qag3OMvifUbdISQ+BQPaKeVjI+rJguJBc5+EKRwklXFmk4B00rTRyTj/zUhHU1hiAEhF6vuJYQv5khnVV8UFbvKaaUZjQvWKhoW8ilY609XrvtgpdeLaT+L3wZyDSyZsaypW4jHSkMUIUswlIW0hqhJrw39l3C8YV1nTkHx4vffrFDhFRPXN0XSnjHMe0IBGLcgwVK2UIbLg4I24vuOmf0MvP5rJyKos54f5g4pj9T2k8s+ZGmMxIq3jVUKyF2G81KoVXBNi8Xqp6urnwdgeJwvTLYipsnbkz5t74nDnt+mp95c2z8+/s/4su3QMZCANljwSMGffff+X/8T8bf//qX/Jc/Kr//MsEamcKRcLjFV8XHgBdDy4w7CP/9pRGt5z/c3/HjuW7cbxLjMHK/y7zZNbw6Solca6XqKxaYwBh7OkmgiaUNuLzn1AKL9ry9ET4cFt7sB04l8d9/Xvjz1fH73PPHqoxhG/27akR7zTAcB655x5/PIz+7yEOG/3n9hFrgfM48tSOT3HJW4cQOkch/f/wX/s1fGW9vL7yLF/rukbUWvN8OQmkz+34Pb5TdY2GpkdTttp+nzLR8wTOg4hljJPmMiOFb5Uk9abgnzxFaZq6OqYx8+lPm8HbPbvgZH6/sunnDWU97Jr0hhCPruCef4d0b4fj2zM3+hW680vwXnH1GxRDnEd2S4jh5HfUFPB0qO8x6JNYtBR82oIzzDidKionB74kMWA2k/oC3cbvMpIow04UNcPLBXwnpwt1u4df9zO1U+Y/v3nL/Ynz2B76znvLoqC2COP75c8+6+w2Xx5U83fPT+UAToxsrkieoPWoBq45f3Ar//n1hSIow0XTm67vK737lqVwI9kei+7jV2kIltEJkw5c6cQgDP3/ytDViHkyVbEoLEPeBLnWE1r26OBQXOpwFaErFs+qIlEDLC9EyO99hY+RSPcsSMJc4pgPBOqZiQAdFtnF4t3LfX9B1RZPDqmwhT7cSO8+kRm0Oz5YOn9XRRrdZ16LnIpnSKm9DRy9Xnt3AipA1QTiSBpDaCLb5SLbDPrJqoVnBD4XutRXgpCIUolvwdUFVKF6Qo3ItC2cLWIiU2mMKO+e4YSRi1LWhbkffBT7bBTd05JapnHn/sOPL+cRcItYyQ1qwMnNxt6gb0FcATG2bKjkvhfrlzOqEfrzhfK1UNn+LNaN3yth7Fh34tjXu9qDTF36clL+KCcsFOsfzKfFxDawtImrEqAiFNm4H+DBPWM2oa4hvVAergbJNN4I3DhIY1HMRzyw7XooSw47BJdQyPvQIiXNZUBytFroklOVM6HpE2V7q64RJIvvEt6cT779JNB82ouMEdTacNmrNHDrHrUWutv1vGwKkEbqAr3BoCZXIXEB8h7mNASiiJBp7hKSJVt2Gl8ZgyXQpUUzoMVIyVuc4zSBsul8VCL7D54xfV9xaEbe1WbwLm47aG80FrkUor+H4Td4V6HHcRM/bw4BUxTfFdAui2hg3j4p6WlN20RBTgiuQV7xsfhPTRjBPq9t/txY6ZvHkKgQ3EdjWUK1VvDSi3yBMWfrtIu6MBWXVSrSeddncEk0iRXpWPOrB/F/YTdBs+2F2m1eWFB3eElW2KsiUe0a7IbrAIQbEKlWeqO6McsU5D8zU9YSVlZICTxmqe8P7IZCLkTVQdPNLp26k1R1KxRho6jDnMFFiaPz2w8w+/Qu5/ESKBbq4kQmBnDMmlVUDL9MttXbgM4pjLfesOdDpxIfjgZZA5IlVjdZmPnQn4s747YNx9zbz2w8Q5BZdN4NWMyWEHmeOdX1mx//Kf3j7mV/t/5bPLyN5yXz3w0d+8jec2pFWdiTfMwRHDStXq/zDjyPneUdpnqtLPP/YsRs9l27iP31jHJIj0JEk0nmYSyYvGbUdRXvqeuDyGHC2J/hbpmK0nSJuJVhmF+Dvvt4zniPffZmYPDyuuvWBdWCaB1IJ3B0C1xb4vuxYrsKQbhjaI3GCY9tqjiUbWjJG4eZuR+SWhT+x67/H+TPV5teEbUOaI1jCtYazJ97shPkiPOw9XZsQrZRZebwKXzQgI7z/OtHpQp4zpzOseaBPd+zLhLXG0oTnKXKaA3eHBXimtIrJPeXUI9MNYp7UGe8/CIdvKrvuJ4J9T7WZlQK64tJmkQNoRcB7JIw43lLzA9UfCPFIjF8Qv+lR+35g0Zlu6IjFEfOA1Ij3guw3otybw4mHm8/04RPRvTD2KyFsv/MpOLrFkeKeb956DmPlLJ5f2xv+gXs+ho7LHHm8Rp7/caLqDqsjRk+nAR89YQwE6cknIze2qc3TF7r7B8bDjr5bSfEZ8R9p9kQIT7hUwCJa8kZ6C4EqjtZ61mlPXnryRTenvYPZPKVLqAv4taLFM617UjxsFVvdVijGkadzh3fQy0wKCzXPVOlxUVhzRFzgaJ5OG5aN1TwlO7xN3N9AtMsWHA0R7QWrBbONPqctkBW8RVqDSzDy3m8npyvUYNicOTRjNMUaXC3RtONkt+wGYagLKa94YFXF9wN1LZsxsGXSEGi5UGth540bKaSaiS6ylgtFjOgDSQOqWwA4WSBY42CRXq6kaLhd2vbEsefqZixn+lD56r3wT3JmffGMZhzqC8lnWvM8WmIJCdqGCdaw4Y/99UqsGX9MuBhQL1TdJiKdGEnyplg3x1krYej53Iy5KmOtaHMEC4SaQBK76LH5ESeZc3AsXeTJIn23kSAnNRyewSeKbMyS0XfscPhSMRUmiZy9IzqQ2MDDXArVOayFDVjUOvR84jgkHBXFQ+xo1YFPNAssjDxdJ4Zxa5GoKqUA5lAMVzKH4EkY0RqDeXZdotaVoBtiPIaeHd02jWkr+fyIw3HXDXS54r1hZogLrObp14G0KF1snMQRUs+FzMV7vNs4AU485vyGv7ZKbY05g0VwPuJ0Ww+osUGo2LgX4hIQNkJqVnRZ2fvXtL6LZINsr/VFSaSmSD6ROk9pCRgJohxDpK5lg3K5Dtx2vkZxpGb4OhGdUNtKs0KQRCtKDT20hisTRuOC5xHIWhmjg96Rl0ZBttWCa/jW/rKXAadAyTRVXlZhuUz4mFC2nUf0ieI9Uh1Kz8t5ZTw80N88cgwnbn2jtpXcrmTnqFGYV9vG7u3Cup44FeGUN6KG90LJlRpsu4C0RvOe0PfsLHC7n3D2iLUXGivmAi7E124rWIucrj1PLztavaO5K9k18uy4aY4b74i+4IXtMBv2tFWJXtnxQvTCTYrUIlzWhHcDZlfMF5q7bCAHJxgF41tud473+wd8ufL3X1cWfeanqeMPnxofnxNzueeSj5x8x8cwcnox3g7C/b7R+SsSA8gNHx8f2L/bg05EHL1zm2e8DbT5li8/RabLHuUOFcdcwNLAd+FKaI3gAruDo/dnft2f+MWbgrVGma88ny/8748f+K8v72j+A48N1qY8+4Fmhf0wkkpGlszYj+SnzPPLCYLSdZHkAqmXzSkgVyRkFl1xKZDCBtNxOaJzxdkF174QfONN79mpYp0wMBLKSJsdl9m4aZmeFVcbu6SUy2cG6XinUItxnhXbRd7vMkOY0FZRhdwiQd8Q9J7oM96d2Hee5DLSnjDbJEhB3CsFcXthbhJLDxZoLVGW98xf/o7mIodbZRwulFwwqQSMIUVS7PElYqdIq4mugzcfLhy947j7np3/B8R9RNyCkwV8Qb3iZcTcDc11pGFhj+EWxVZlJz1pNyCjUVfB1R3qeiQOm3dAHTY7ytpY9EJWj6RGXmdero7Pc6MfJ/72Vwu7/ozygotP+HCiUjfYTZKNG+S2j1i+7CnXkX1KvOSCOKUFhycwT4Hz0gj7QNSBP/zxwO3te3Y3mw99a2nfUsuBOFYiVxLrRvrTiMaADCNDHvGl4mshlEAQpRBYW8fjy0K+j3S+Yq7hQ4dXJbQES48uPSaOKgNLi0x4VlNUCt5XglNu+0AqhsuFrikSKibKKj1VF/rRcRDl0QorW9jOfHr1HRTWlvEBog/4dWUIsr0Usc0roEqLHrVKjIkklVEdXS0EFG0Vj2coHrQyiGPJGZ83jkNzhW4orNdMaoGoKzs/83Xn+FICz3PA4UkpbE0NPGJGqpmwXrEwYPTbhbQ1eu8QrZt7oSpelZiEDDy1imOlLZ7OKW9CpLiAbxnXKjEqVWcm52ip49LA+8Q0X/Etb6sNQEOgq55gipUrFEOTbis073B1I1fiYG0Z3LBd4nziuiSG5Dl0bQP+xI5IYFWhSQA/0lql5E1bTnKsrqHZEDzNszkcQkCaEWIg9QNLzQSt2wNvLUQ/Yq3hJLG2hKdtXANzvB0aOw95cVxy4uACVi/EKBADbtzz+eXEZIJztsnoZKsTm3NU3/NcEp80chLH4hRnnr0PSN2CfpXGxTyq4F3Eva4Ep8vEw4PHktuw29XQEFmVTSWuHpFEU2VqPcUivhRuksf5LYCoLqAIrS48iPHGZqTWzczaDKkRcVvOoXpPXbegdF5Xru7AowNU2Zti0uj6DlsVa0oSI6n9ZS8DQWd6E1pufD4rZ1NcaTgPrTaGIdCCx5pwuiinyfjeef7jb97yls/E/EyvlRYiOUZalxjkDR5j738kyTM+OGKKTAWmciarQyzS6pm6XqltpHrHroeu0w3QUWyzeukMYSZ2AdRBvYV2oKaOtTYkdmQqc4w0zdz2Ax2FeV74Yb1ALaTuhuYStc3kNbNMlZdzZVoHJF55cxtRTlQ7IRJwXY9oIoUDUvfUqeLDiSH9wGBn7g4r/+ZXI7W+58c/f8U//vA1Pvya/1Ijbrwl3Q58dbxw1zeuNvJ4NX563vPX744bHtdt7IZAwpWBtnry4z2P53uWMHDNKz4MxKHn03NDwg2myu3Nym9/HRmGM1LPSDvTdxO/fBDev3X808cTf6xfwZRQJ7hxpOcTH4aKK0AVylT59HNj5kCXOgqO7z9njg+wP3piaDjZAE+qDVVHCLLpaa0nFM/dg3J6XglNNoRy9Ftdp84b0U07rhO8u4+kTsnzxP3gGWclzI18asRy5L7reDhkrM5QFMmOUlb6aAwdyGpUXUipoy1GGTo6OSARXDhvClQtmDNU24azFUPIiAnXxxvKvGf9/Ej/N3tSSpsUxQmEbVweiLD2SAvEeOI3Hz4x1ysufAR7pnLFO0Vlu/SGGFHLxKSIBmpu4DOx38KVhJUiW8XLihD8jpAG1Ccke6SAlU04pOpw1uiDcr0s5HVl8YXduhn2fiOBbjCO/QbuKpbBZqKvuC7gpVF1oOY7wuS48Vfq0LNWQWXTMdtqqOt5vkKSyKefbvn4/ZWvficcDisSlUbEOxjCisgT4hvqhHkR5rYD3xHDJptZc0M1EP3K0qAR+e4x8du3R94NG1wFHCn0xBJxLeKWSB861gYpwj2K1fn1g+zIqrjWsJQoawHZ2BDqtp2ztQ1JfHfrebwa5wpVAy72m5GwQauGua2yOA4DvWxY6WvbUt2rU6ZmLK3S6kLfjYSpENmCZeoGqlNUK4dh4MUyMfR0ThG/UTX3baGFEWsd1RJaLoy7lVu98MNygHTYVn/OoRZZmhFcIplHtDKvFyT57TVoGcHIRREtPHSeEB1MC49N6WNCW2RUobUFc0pziiXDu8ohOhpK1h1PWVFdGWJk77epUKuB4gIWHHO5EsRvawbL7MMep5Wdl1dOwna4NStECqjSjx9QWbAo5FaY1wpeiG6r642tEEOktEwWWL0gByU/X4hhJANFNj+J+EARuNqKuUrXQ+cdaVZqzhSBz9lx6d+BRUpLtHyl6pm/va30zlBgaTNuMKzzjLuBz4vw09pTnMeMbbL3em45IsUiX1R59JHFOVZd8LpwwDOK4spMCw3iyKSRWo0QQFwkl22KJbKtr10MFJTitsxBRQllR2oZZQsiO3qiD3RRqGYU5xEcg628kcJRFHv1RgTzRDaEs5GpteJcYq2gVahdAwfNbaKsUFawFQ2R7BUXHVL+wtAh2gVv2wfkZTLW5DchiE103lENLASsKVJ0g6NMA3/+GPnqVyNheUH0tXJhkdOl4LuNZIee8PLMYXQcgkdFOF8zRreBFPKeuTTCJOycB1dxviEuEHy3tQQcZM1Yy/RhT80JnV4T01SWcmbrGwg5QvUw5kxdV05r4PSa0PfDq1LTbSuRL/OGee27AXOG2YzpdauXOMPJiEtvMPeBvCrIC0MoWLuCXpH5Z2L3zK/eO278DfOfP/NP8cBTbvxwWoglcxkUPxir81y4obkjUn+kWcOkoy5Ccj3SIMuRn+oNH5fIUkb60OOro+tvCcUxXxd+mM5095Xb+IguHrEDgcaxV/bdwv/yt4kf/uHCeeo53HrejBVXfmB0p9f9dGZpA7O7YwoD6g48rhWZEw+Xjr/+uifohoN1zqPeU5uhOvD0uKeub5nme64rHDvB+X5zinuHHo8c94nlZeE6eZ6mwtd3iRSF5DpqUdxF0OXCmjc2+vW6oNYw2y56vfdEF4jlBN7Ahw2yEgPPbSAvb+nawH4odL1iztHMQI0QPLo2xBZCUHy30nV7zqc7Lmti+cPKv/m7XxA4YbZgtaFsZjLxARSSn/HhM2l4Q26GtgFsRNxE8AFxEWeQokdYaPGCGw60NSIhMHQrfxW+EE4rP67v+CF3VDcSpAO2KqELDtaGUzZLnxl+FU6zZx+FicLt0NP5PY+PK8eHkcPxE0pGbUVcBcrrR8phdUTWHYe8cO9eKPc7vv8Cq8TN8a4Nx8C0hm203ByaD/zwufFVcqQ24bwQDNayIs5zdpEva2NeoYlndhkLiiWo69Zmia4RrIB2TG3kH7+9cvjrnhjyFtIU4RAjljr63lFZSMnTy8xOLuz9Qpsj//xx5Ns28ozwaJk0bNuD0PXMdUPcNhPElL1r3O32fDr5LWDrN0Z7Wx8RL1toqxVWqVi3oYJLc8wWyLpgTTHLrytKiHH3ikn2ZBW8dyRZia7RqhK6TTbTJCM+sHdK7TIZY3KOviakRWJd2Tnl+lorU62IOIoqqzjmodsaL/UKrRHTSPADuVSKzuy98YsuMl2+ILoJgR5bZOwDIRtDrVRRrq6xJEVMwWXeBOG6Zk6WaOIJVrnzGyv/Kj2zdDw7JWnC9YJIIzolSSWEzWpIFbwP7PvIrCvaVmLv2Q+JsgpTXglxYIyONHi8ZfY1M4YNt9v8SG6NiyjXHsKNw6/XbbqcPJ3U7XvnA1PLTHnGyspROt6P7+mI+NVT58B58TQGih9QRsYmZB7BJnToyHkmJMMPHWud+HTaRFx9ShiFopWimYRyqw1qIzOyOkdznhj31CwsWrlxlSSFZg5rE14S1Q84SXReCDVScyZSaKo0t1lCVTxZG06EXhKpGs7bayPGs5RM7PyWyxGPykpHo9ta2pj9qy1ypvkrL87Yx0KojRdVWjQCEaMnrp6xDwx5pdcrqV2I1tOIzF0i/sUnA9JQg6qJU9mgIcENBDGSN3JdWZuwnk+s2bGUgdUP/PgY0fcjoQScdVRNXMqR2b5ieqkMHzJFFsQZphlnDe8a+74QvN/Y0+sdP9uVSwULkcxCW2eiU2LUjRplmRQV8eCdIuGG2o6UxeHTZlVsdaFIh0TILbMrhYCQuoHkPa0LvOiFa3mhG8GVxrouhDhz6DKtXlEK1gzzW9fWZOD8tOc0eZSBd2PP4Bum2yHiXIdTg1q56Tr+r1+P1HLi96swmTCM0HSh1RUXPJdceFk6ehJhUEIYqQi5vmUKb/jn6YE/53c8roYP3StQ5ELqKu+PIyoDzpQoK7fjkdM0cy5QbM+P103+koaO//SLzBgmvto9s/OP4Gfu1srLeuB57vhUB67+nloDLxMUtmDXMi1bp7zJFsrzitSCY8fL6Y5/+uG3fP/5nrnucP3I3WHkqoGPpz3fPy38YII/eIbdPbN4vI78+XHaqjcx4oPh5ULfruBArTFdKstlYegMzdtHMPiNMjf5kasEisLLKfCTBiYGdsM7dvPKXXN89fUzjglV20aDZjgM5yrmJhadeVze0uQNTz8aPl34u999QuzHTaqy6Yk2UVQFyQV1C33fUDWadwTfU3XGpDCksB3CbcFHML3AkJF8jzXj0B45Hn7kTt4hH3u+8Ctmd6RZwL1WiLZDKAAVHyCoo02Rnx494dhTqNzsGvu0oCWhZcB7tkS9bWSzJRfEVWK/CX/6LhKYOcaez1KxsaPOkeQaQQqiQpUd11x4qUKKkWVJ6JfCcX+k2pEp3/PxOWA58fnZuCrcdXCMgSSRU1kpQQmHRCjQFiVFI9XNynm+Bn74ZMRfeLq0WeE6K9yPEzdDZGoTKSgPyTNMP+PLmUUO/Bx+RV0CJXZcyRub32SjF8SO2bZUfEfAUWiyMiTPtSjejjg1qnly23gHzhw1LzB6YuywOSCrcGyVZJUaE1dtGEKrmd3QgcF1KXhxqHMUWxG/jftb6si2Ufje9sq9X/gyRuZlJatnPQWcLRzDslEXTMAquJXWYKrGY8m0mx5ZK04aVV+4lAt92vFw23HIE8fyRMwvxBQ3rgtK6zJjiqCBft/xNE2sIRJE8MsTN67QB/gkBwrddqFtlYHCSOFJei7q6dzI4GeOOyWIxztoKJWCyTayxq2vY32l0riWJ5zrXu2jidiEh3Sg44WoK0EyakJFUPOsJlytQhLGlIivEw0RBYwlK67rKMbGDrDMsXzmfdfo8siXZUeWDjcM9HHkOl02pH3LCA1CwltC2wrLSpsc/bljlyJpzhCEBccsEGi8s0IoSnK7/78q3EtA4oBJoemKOEeIjtrWDWvvdpTWuB+gQ/DztDUXaFjaOBXCRsINonR1pjcIPmJRWGrdvJGqJB/xYbsE9kBwHnygLTMR5Rhmer8wW6C1dbMoxsTFKUkGrO5JCGNUelk2IqYuJF3wrpHZaIt/0ctA6gslZ9YSWJeAtYWvbhy3LnD0Gzf7dJrYWWAthpeOlcBkI+frwF5HvAVMej5fBn68DmTr6NOVXzwMxJgRzVt3VAXBY7VBWPG6UucJWtwqIU25nCtut4LMiH8FRbitrRBcxPQNznY4KYid2QlY3JzsyjaCLcWobB3XNCT6o+NtGmlSMR5J3Zmhn+i6MzfDCy2fMCuId4gPeH/D6emOn37uuThBpfHmZoQuwKs3waIH11HmDjsH9iL83/8q8p/bQnVnQnjC3IRLA1ONnMsTz+sTQyr0WshrI7c3PD4bqwb++bTnoj2ur9SWmVtg8iPSMp2NFDP8WsnnF4iZ43hLmYR5qTyr45oHtDjepO/5+w/f8pv9TyRWFkDbgfr8lsfrDf986nh+Favs9/Chn3lz64hhYRfByw6ThnOKdwlnA9fzjmX9hnk5MtVIXqBq45c7I+N49Ef+kAcePxvHPDCkjjvX8SX3rKXCohxS42EnzPIEzogqkCt1Frq9YbVCDWhuXNY9v3/8io/P8PbDHh0a57yyqjDlyIkjL8+Vw/Ej+xFEPKa2kRm3v6003SpB56VRfATe84//dOKX77+wH66gE1420I+6heAEqkNXqOtKl6CxZUeCCKUJHsMJqIcQFG0ZmIk7KCcl2AUvz9yOjuPwS7rgubRtlOcNuhZQkw38gmCWuen3TIvjMgVOcw9hIshMGlasRpx0Gw48RFqpqIGKkMvGYY9yoQuf6faGP+9p2Wh+E94EIIaIZk8VT9aBx/UtfuhpBT5+Gjl/B1M9kpcR6Q94B+oiVa/8Td/40C1QlGlVLlRacow7GLLjVo03y4mYles68+lpxFzlm689Q3/BpwLtESxs+HAJHL1n8CPrXMht4Cwe0oBppIhQ6kywzG3XE50RVLgslak6Tk2YyagPJGl4faSTbTdvtTF0nl00blRwZaK76UjV010q92F72fsu8dwcP6+OEpVieatLxm3tMqRXIqVWDEd1kbkatl7Z2cxN9Bxq4YdBOJ88be0JgyfVmSFtYDRx23TgLIniHWVRVl82x4JOpL5jzRd0mbkd9rzf98TlZ2gZJ5GLea7ekRFK9JSSiar4XY/UhrOJLhh9XdlJ4z4knpeA+Yg22PnGoAVfobjAIsrqItVt+38DqoA6wbtGJ8K0TGQyWWxDG+cV8JhFXBxYlspduPLNQ4cWRYaN4Ke1oRY2nTKGCZydMLxeanK9Upd5M1YW2cBcPuIcOM0M+omb4Z7LKfHs9lvQtBpOEvO8yduCZUITalGmxaBLlMXRWyIViG1BTfF9JIjDF2EXAt5AWsG5jq7bGgNFleADJiOlbd97RIi58tCnrSpaf0TqE04XxIwQE2LgRHBAqyvSCn0QOjWcGHiPlwDo9vD1CSlKckIUAWdUJzQfEAqDyxzYQrau6uZH6TzZHM31pBSJFaQJQdiaN63RuUrvKqJs/+x/yctAFyCnLXRk5jcvPQt7EW7UQxqZ85nZrbh9Yqxlcw7UwvOl8kY6nC343ihnx/USyW7gj39a6Nwtb99mgteNECfbPtE5wUxAA12IJIssEmgamGrHzoRmhVIqugVZtx8GgbJ4emuMfcPFikhg1IHqBubJwVIpayU7WJ1sGGRWHsYz43gi9Wdwj4g7E+IMdqJEh2NDZsYwINxwft5zXT153EhTpWSggN+oXRZ6cjlwnu+QfECqh/qZu0OhP5xo9ic0nHHJ4YaR5irVn8n5mdYmxDy/eBDi/Q3ztfLy+TOfrp/5cHugWM8/fBHW+JarJl60o9XAQQp9OlDmlWX1+PHImITHy7xx++dv+XfvzvzV8ZlUnxGgQ0GekMMzv4ofaMPIpCfUGzdd43ZfSQNcs9FZZS3bhz20ytB1eA7U64E+B75OhcUyF0vkKRNSTzw23qXEp58zS3ZcX54Yb3dEv5KCsR8SaCGGxskK3Zs7+HTGrgrmePlSudlvFywfd5T1nh8/vuH/9f8x5Gbk3W+V28MLuhTO18hPpysXBe1mpqkwdoa6rX4tZsQIIUWKDlg2lEZuFVxPqx/4/ofv+De/u8ex2cSQAH6rm1kTPHvOL4Hxrcdv1ILtJRAjUMHKlscQI3SGi8I8TWx4khXvVpI/0Q0XvM7s/IiFDZSSGCkaWF/fZdEKvlwZnaMuwnwKEATTjPoJcTu8j8wTmA/EeKS2TK4TrTlq29C0wX0iPYBcH1h/NLSG19fpTD/usBXUthH2Ot1zXTq+i5Vzvud8haiOvzq8I1wdM401DRR2nOczbpyRNdNLB3ScqzAm43eHlV+q0coXnrXwL9Lz7O/546fCbDP/9m8MzzOVioVIpUNbo0/QW6SUG07ryFOraL2wE2EfrjzcGA+95+AyZZppBu0AH58y/2Pumdqe4Leph3cLt53H9Y4YIlEzOw+pCL411vmMVCGy1ca8VGJQDn3P4wxooDZFROiiY3ydPog49gFCaxyao6+BuYTtz9DCrgnRAp8s4ccDKa6oeRKNYo3iIKfIix231YX3hEXYhR2LNc7z6fVi4JBrZZYDPibMeZoE5tTzOUNfPWeEUoz5+Qs3xx33xyMDHTt3z4N6fHviIQr/MkVW71kkc3SNVLfvlIpnxTG7xiFEmkSagjqj4Wivq4HjPtLmxqXlrZroNhstGDMLXeg4vZwY372hxY6JxlorGc+cC6UqISZUA7hEcwXfdczPZ2xdSQF8n6gIzgu9Ew7m2ecL+/DMuR/5XG7IGtEiOBVc2ep/+84xF2W1xmOtnLMS0ogbKsEqpV0hOKo3mtTtwG6botm5reZXsuACdE7YdT11VSyCOYW2chONd/GRT9PA2SJCT8QTtG3TQDOWZaW4hrNKkIbzgncbNt3HSFBQLWgzsinxX3M7IbBaoQ9uWyf2/SacahUTQS0RSmHnKxc/kLoBXTeQVTOHRIe6LeRoNII3vG+0uv5lLwPja2dx5z2rNYI0ksv0XdheWtHw3YUhXcAq78UxSoH1Ew9tJUwnXPuEv8JQ9yzXRk5gZcc//fOOxg1v3xqRK+GVX27iEddhmkhDIIq92pky5hpN5LVX6dDWkAriIRTPejZcueJtIviMhJEkHUfd42xHlhXvN7FItg5RIS+Vl1NBdeLWncFdMCbQjJOK2eaxjnFPjAfKuiOkHWlvkFYQT/Irui6UBpXtdqjVM7WRHQl3vSLrhctVqTGSjgnlddqQLng/U/OPePtCjMIYEq7+QKJg3cL/83/6HUZP0ImXKfGmj/y/Pwf+6aXjqQacrby7Kzg38nIZOBVhroVrVTJw2135zbsn/mr3jG+V2m5wEbxcqfkLYk/cdD/yn74ewB2Y18hqggaPyoEgHS4E+nQAg2yVOUeeXt7ycn6gTJljK+yIiCU+rxUnhUOa2ZeJm7vE5T4wi1EpHMYBfV6Jdabolavt+P9+5zmWnl90gVoLzgqn9R3fnlaW0vHyfMs6PfB8+oY1jHg9cbv/yPv733M7zTwJvNt19IeeISh7X3CySbWcNMRvL3dnCrXitTAkoVahWgYX+fx44K9/dYOGDKqsq0B7w7neodaxXCJFEmH+RIxKSm5DdL8Kp8QHmlW8d4jf0aa44XHDgpctb+FZSGkljcJSt92AF6PaQmmJ2rYqmKPQ9EKTzUl1/qz4XUFbwWQhdh1mMC8DKgMxGcF3OHW08oKZQ0056xM1VvTt15QfGqIJYUa1EPuCd4VUJg5R8HOjSuITHZ+LxyRxlMYhdPRzZSZxagO5bfveZI6mW5tIQseiDqTjvX/kfX6h6oIi9N6hTqjplh8+C+/eG29vLuCV6rZ1VFTFuwsWFHYd+y7w7/aVtX6kizNDutC5mU52uOmGLCBB8Z3jVzcD9V8C//jF4ccBQqPzxjE5tFS0FTxGcBti2mHUsmKy0O/2mEZqcDhXSD7z4eApQWmtp7ieJgVpBXOR6pSx87wVx+1lwTf4UgKtS6hsCFxbHHk3oi5wxVBTBtHNGeJ7ZgJLGHCWKVoJ4mlz5qbryMuJPDiKN6o4fnx55pdvdkTpWM8Lc6ys6cDnuUJUri8LVpVw+cxXj8/83cOOb/YDO/X40DF6wUvHwshFGkerdFHocmMSx+o9V4mcreCcp5pS24bADk5QCj7AcXA8z5HruqIOQggEH7nmCXNGUbBlxrLwNGUWz1YLTInODBcPZHWczplzqZgVOtfRW3jl8zt8iGQxbrvEfVV2RRjalffjhR+WlcdWwe1YmxJxZDXO64VijcvamHziSaDYSgorkxbuhoDpunEcxFOcMBuMMeI7j68rao7kYOgDUJAI+74Dy4DjXgtD+YjUG1Y7EDjg4kpwV5TNylsFaq0k74jNvTJuImYB3xKxVVpphNATm5FzpiXBXMeqnjMzYx/okqeUysEnvpwal5Do2spYPDe2YyozYRigBWgThIo6j8SR1jZ/hHMQ4l94TSBW6XwixsDqBJGC1nkLuHkYk7IfLiT5gRgmRs7s7AvmFsZl3F79+ZHkA46ZtRhLMzQNyPUDf/o+Mh46Eo+AbarIEOi7jsCR7gaOwwVtK6H+yC5+grYQHGhVfNjkTN47XDuwzh5xYG5FfaVJ5MfnyPdnh7qRz80xhC3N23C8GRzNFWZ3S+cgz4/U5jHf40sgpQVDUBcJ6ciS77icO/xYeHOzbFaqa6XjiTpdmW2hhYb0/baGiJFdBvJMzYXmD/z528r+q1v2x5lDZzSZWOZnzBVIHmUTaQTnoTmkQKRCvuKacC/wn7+BX/3yzH/7+MQPn08chsbfvFOGdibbGRfBtLHv4bg39umZuxip0w2z7vHOuBkyrp5xq6CvWmRt6+Z5tyOh7Zk0kYHoetbFM9cdxUdcnyhLz3ffHXmadxv5r1ZEFoKPaHE8Pk68eQth8IgLMEO5TFgU+p1wdz8SrpnLvPK8jHybH1gX+E6+8M1QWdYLP55u+d9e9nyabnl6udkaHGukZM/ABV8f2aXfM8qV2x40COYd4t7Q1j0xHKE9b8FP0S0FYOBqofOFoBM3XY+2st2kdUctd2CNsirowLz+ij//PHCZAr7vOIyZB7nizBNMNhCLbOli04BiiNsR3NdQ98SuUrVgRfEu4K0HemQMLFMDepwqLlQ8Cm2iCzBYoe8cFxQXI7U6WBU008oLpSkuKqEbaXpFaUjn8DQ8M7alW14reBOZy6ZntbZ9wNiqdCEZ+7xy0ATtAhII6QFxiWqG6oUdM4Nl1BLSdkg1bpIS6ozzHUNK0AJVPYcUOVpB5oXkHbt9JGVoqlQRcDf89OnKm2OHakXFAz1Dp4T0QrOJamci3/HN7RYS1jqjbUIISG7Ua09se8r8DNq4Gx2/e4g8Pm7s+eJhDImuVpx65gotOM4lk1yilxWs4iXTjcZt3BM0IZwY/Qu/6FdK7vjDxxs+l47iHU4iIlvCXoBjU3rdgDJdDFxCoLSOsmwHSOyM1QqoYNqQ0NhHKMu0FdvjhkUO0uicEv2C1EwU41pXJPXklnnWyh+/vPA3vSdpo2tgq/IscC0rOIeFiK6N+brSt5VfpZ6gBRcCwba1RrNI8TsWW3F+5S5sNFDz3SuYbcLVidQc5gDZCIsuRdprxbKjoa2xNsP7HkEQB2tbt3XcdKFmOE3Gy2uKPlLBIl2X8LFn8gvLmlhrQfXEN7fvSQ6qc6w+4HHsXeRt39FNGWcLnb5wF+7o5YaXlkEczhmtzjS90lzBhZFWoXjHjGM2pTiP1sZD3+M14/FbnTEGnmtlPQi9N8ZXXbcLUNXIa6VzAtoIyRPngixnkibE7XHOM8TAwXlObd1gQgSoGzehD4l1bjjfbfXm1eFcBFlI2gh1IfWeOVe0QnWw9JupswsDbV046IWRkR/oGWLPUZShNYzIdelgVXY2E0ZPdZ4adxS9oa0j25fkL7wmgEaXEj4q2hlBXoM6zsixsAuVTmaavqB6QeQRJ48bp91GoIcgOM9WR3MRxVMUstvx8/PMN2fPzoR5DTzPieu6/YK9PQ58825hv3tklM9UvnDcnTCNNBLJZXxUfDDQgK4DTXqyB/xAa4mn8xv+6dORH5c9sxq9P9CFhBPofaFlpUsJVz2i9+zsFi1QfKS1wk1yeLdu9K62Z3oZeDmBxs94dyJ0d/QBkt8+VI5Ele1z0XeBvmvEz58R3brMjyj/9PSGl9OR91/d82/T9+zHR0QyzbZVQxcDqEdsD+Udtr7jMh2ZJmF8TfzmeOaXH37kl++eKfURb4/08oJI3Xq9r5UnvGI2s1SPTX9LqXdMOhD2KzdpBj0hfusTGwMSDpjr0bZH/A3iIY09kYHzeeEyHcip0h2E0I54/7DdqB04F8mlsoTImcifv0Cnjp4B73aINLpjR0b5cjaKXPnN7sBNgPoRrr7n0t2x1CNXVxC+UF8an18Gvn32uMN7QmvcS4eThUhHXlacnYFlo7cVpZZu8yG0PWp5+91gA42obt15p8q+2zj/5iZ2rlD0SqiGLXfEnWLtDLVnXfecX3YUSUiJtOuVpj3OeYyR2gqNW9a6p7Q9a0kbvMjHrVWxJVSgD9urvN1T7EhkI/895gXnrrwLmfeD0t9kRrcnXztONXCeL/xrGEGbYK3SygQk8hLo9nvMXTAplFZxPhJCh+Zlo/IhhNCRc8TmSrIJHCzWkWsFqRw6R7RMcEoQ0FZJcaTm7RDYyUzSidWOSFtwZI5pwemMYHTdQMgJX427fku4x9Shzogi7CQRl8BZbQvf6Q5sIqZEM49Lxr73SBDKklnqFaufEQzv7hDZgg6eHdpusDqSl4aa0FZDmRnDxDcPD5xMqC1wdMYDWxPkpQmfm3LpE857bqOR7ITXSheu3IwwCnidGPnCPpwpacfP/ch3jwpjoknDebf9HSsFaiPUhtOV0ZTaYJXEYgeKBnoveJ3xGGu9omrgI6lFStbNeLmD0uaNEtoK2RtZtped1YyEwCyZLxl+mQYG5+jwuOJoavS3I9P5vCmnQ89ZPd9OM0tWlExtxtAlen+hmOBsodUL2go3XWIRw1zihsY3fsXsyjVFPtftIBW3BQBLy0TXGB28C5G5CvvYE6KjuW1SuHPCzhnTfGKeA/PrjnzY7dmHHk9kJhGkbYenDLTV8dIu/PrNSFbIFfJSGf2OUSq4cWsDOUh6og/3XGxPmQud3/DixbaXdGMlxA5Xt3VzbcIFAVMOQfh6f8uyrBibZrkEcClyiB6nK3ioPtKy0qeeXRAsb5dvL2D/WrNMEWdKJ8qoFW3Q1YYwQl3oyZTW6Mc9l3Uisj1Wm1PwM6VlxiHgWyVMmT51YJXred2AUDUy5IVmCzvfsQIWRlyZ2PmKax7VrR0y0FOWwITg/UTxe6p1mCTqX1pUJGZ0nXC3i1zbC+41/KIGmCJcGMJl++i6C+IypYDJJvyxtH2MzHuKbHCKSqRLDs+EmCNPkUPcQe2gJWqGpQUuF5iWE3/311ec/EwnJ2peMHPbHsZ7vAjBR5x/y2X+mhruWPuBuVVq3vH5yRNiz3t/w1pGEE+xRtWZLlTaknmcJlaf6U34MN5Qs7I4x0ygSuPdfcW5yjoFWnHshoDGgnUFcRdiBOcNrN+AF74iaaPXhTLh8jYKtrAnM/CsO/7wcsefrj03Nz/xn95ExDzFOpoD8UIIN8h0B+stbdnz+Vq38bR3NIRL65DyA4nfE/mMc894f8GoBCeI38I6r9ordv4e7SJPzVgKuHahtk+IvUBqiO8p1x11vqdqj9vfUaVDfUNISNsznYzQ3eL6GR9XnEZsXvBXpYs7JhWe/IGPdcfVjGvbEv6npRFSYAiJ3AqnWjAVcJXVwf525KF4bubCvCjVHfluXelYebtceXtVurRnToHQNQ5kPpVCqcb1sh0u1jyYgHWUPJJXT9WMieBdYEsNAM5t1VQn9EEQrbTW8Kz4eqJdZ778SZDBePtmxNNzea5oCxhhcy60gZfLERve4+qM9zfktec633E67ZnqnqrGw3Fh/4uP4GecU7RWxCfaOmLAsX9m8BPvyszgMh9YuVlmnBa83HLqH3g5J2pRxLlNoKSJl1Pj/iAkF7jmkfF6phtBLODcFiSMyKvOGkJwNHqupw6KMdhEv9uxLp6iERVHGhxSFRWHCATd1hTBGa42huDxVhmC0TlhrpkxGq56TMImzUlCL8IuGj2KD4qPns4Zx9Gzewm8XBuoMvY71vkzKTiGMSGtMXQXxE9IWBE3I36z6DVmnN/j5ICWB8r6hqoDJvMWuKRnvjqsZnZjZimJrkX2AWLZRFVRhehHzuZpzdj1Sh8roSz0fuKmv3AImaALLp9IbkbEE1yjFCG5xE0Pd6GwFjhnhaq03IitcAyK08i3zXPmjizCoQrRVvZxm768LDMzCy+1p7lbTAOn6cpwGEg7o0zXzbQnhmuVYhmnDXONBWPRyiDQB09SCBhlnnFaabWBeDKeNQ1c80zzK9UUSQmv2+VFfaW0iWAr2gpjhM7t+eCMY3mhtglzA7su8VNeqBnUMrvot7VB9NzgeJt6hnHHtMxcNWM0boaBPiRAX22xW6ZgmZ7ZHw5YDZQCwYwhgBaDOHJeZ7I2zAvOOSQbx12kr7CGYZthBWX0Db9OaJ1xODqvVP3XSp+xlkqTSh97ct6CgIqQm+PlMvE3B8/oek5toVllh/HeR67Vs1ijCOAdy3QhqjAeHzBRilZwgUUrK24D3An4mvF1oqtn9n5gdBFHYQiByzlz1oyvlZzP7A89awsQN4nQsl7wJrghUHVBJJPTyrUVjtqxy3nLLbmZ2L+lrntqG2mdYV2mX8tWOVwDU92xtgGJibUIihKDEttfeDLgzJOi4+2xQxs4mXl3UO4OK9iJY/yZzn+7UQHdjLeFxkToO5wLmCgWHFoStQ0bXSkEQmr0daJp5uVUeXiIxBYYLVDyCllZ2fHdtz23h46vPyQ8jlYF5wNi225aGyz1npfTr/np+w+8zB3ZRYqA5itVFMqJnXke3FeA0EKlH5SvjkpvkOtAlpHerUTuEBc415lLDkztHeMQqPVKK57j0AixsbaVTEG04VoDyRsbupUNFsN20Hiu4CPsB+qcCTIw+57POfCr25FvPox0qbJOlaqOgiBhC5+4FaQEahjZH0e0zoR2oVVPaXWzNDqjvnIXqk0oAWvbbdUF2UA71oMesOaItx196ZEIjZlsCxKMkHp0HViWN7xcO8rUcfxwxLuGlcCnH5XpNOAOgd3+iNUJFwbeftixS9vuOPTwL48zX3Jktcja1o3b0FeMmdUii3kyggZ4qYE/vkBcKmvs+fBmx/WxUFsgT8o6rySFrjRO14X1utC/c6TdytCtOHNMp4ZYwmQihA5vI1kP1OWGnPfMfeNw7Aluj7iMSEHEb6AOAVQhBYrB3g9Mq+e7F8/0peNcV97eCOSOXfBUPBcVWnNcrwEtNyxr4v7mgW7O9JeATAPoDT/NgT88PfHuPjKOM2YZ8wUNQpTMLz/8zLt3H9n3hZ13tIsy/Rg4fYlAoMaXLYgY3+FkQEOHd0KzkS+nB3Ar+8MRk4FjeSFqwkzQAp6yXQwsbGu0AHl9y9PjkVqEaAvSwHdHrrXjYh6j0kljddBioluN2CrZtpyFyJYab2ySp+A8DUHSyFoSszpMGmMSgiz4riA5Y04Y9wduas/hqXGnBekcoQrnZ09YHccPcBwbY5xBZryvjL0DF6niEBepLXKZbsjnW3QeiZII0XCSKc5Rck9tAUGRVEAnQvC0VlAVmoBKobRGtQ3RbGElxEIXVnbhgtcTohuzANuQtCbQcNgrIjyUlw1BnEHU0Ff4VtTMPgGT54WEucRBG70VuuXK7THy7mbPH05nrIcVQ5rDk5DaeDMMCI20zPRVXkN8igHBA9GzoFQnFBRJHnue8GYMYszotvcOkZ13mFzIfqWpUVlpJBYyPdBC2taWdabZEz4mdjJg1eMlEkzwIkiI5LY5RC55JToB74BGW2ckCKEtxJaRJuyCsdZt0jCkSLaFLI1VjfMC3bDZMqVUAhCcUMoWTPcaaVRoxn3fcegaTRTb9bRpAikMQdlZRdYTnUHnJ7JBsYR5oagxv1549sNAK0ZGyOK5LjNeG94qwSVm57ltMw/TmbF0PGnjSRZaWkgF6jwj6YhM4Bdlbit5rYSxMupGIY3thC6f0XxhGG54ONzwVEeWqWwtEo1gDl8SthS6ceBlnmlWCUGZbaXJNjXQVlDLDNa4rAu3ZiQCA4pbI2vryS6wRCjuzN4p9zmjVfhixsWPG3QvGr5FXBX8X3oykF2BZcVy5euHC7v0xJvdQopXvJvZ9xc6+xnHM95twohJZpqsOHfEfI+uRpNxSz+2FT8O3IxCd1Yml3ieDH0n7HqDecXKzK46JoUXHfj9/x64GQ8cDwvOhBA8zoNKoJGYlvd8enzPJe9wXeMYnzjEib3M9AT+5ccdHz8vhDbhRZh1JQQ4RvD5ShKlqAMNlNRhAWodWArgBr7/BJkdWjN/f1jx9hnqhFFxGMmDBEGXV/GEeVCh5RmYwe9ADqR05KfvM5/nwkU9L3PjmkdO8y3LulD9hDrFu0ZQJeEoOfHPPxU+n0+8uev5xf09zDOaZ8RtEwgJd7SWEQvARodbcsHjaXVH6L5hXR6o+sAaN1HN3mV8qtAKEt0G7/GNx2nk+XJDWSPL2PHhbaBcCvPltHG6s+NySeA2/HDUStf3+JK52RX+vh/or5mlXfjlzYm/vn9hjBVxHXMbOM0jpzmw5MaXVfmvjwMv646zCyxsieNd53mQHbrC2K745PAcOc0dLz+f+L/9n3t+505cHp940wewbSxvJthSSSZYjJRyoM2F/n63XUrxr1CPHpOOEITbfWNxlciFsFREE4+64xre8/Gfn3jzUHi399zvNnJZasKiM0mvNK1YPKCrMUyV7iR0tWA2ceKW83rL+fKZ6DMurRAbVVeEH/nq9gutXHAsDGFExltOu6+5yMjawgbT9Bk1xYiIM5JzrM1znd9wXTzpItzcKO/KgFwHfEzE4NnHRHIr3irONdaS+PHH9zyf3yLxgGglFyV3jkcCc9wIg0UaizjEbSPQPjeWpnjZKnfxINQcmGvGxJha5BoHFq0UObPzm0zoMChEaBjNRZYQef4ykVbhmwTVFq4fG59qhl543zK//eWZKjOr6DYWBoZ+42mc5j0ffzjy+GlP9D13h4599OAdxfesLZItYRLw4rE6Yf5MiwuyE6b5iHX3LFmpTbaa4rJjzoEYGv1DI0rZtK/qMfMgG8rYzOj6jlahtgq62SnJtr1Kg9t29jrjVdiPO7z3OBnoXSHmmdg6WtvyHw/Dga40JGaaj3jrNu5BdRxiRGohAsnDy1JoGDFs1cYpRp5UeMnwqJWQHEbZeBje8EDvhdug+FTJqeBD3HbSsOmbW6FzjdFHVjUmq+jlM/rmK3y6Q1umsoW0kbz92aqUuqJOoG0cFTNHXhvUlaCFQAdknsp2OYvBEdVoGCJ+290jVN2mPd45Ig5Lnm7tYTLuDj17Al0SRq8sZEoqiEI08LZy6OBwOeN0IforVzWK7DEzsijNQ84rwbb1kPdx+//1HV4rQQvFb7bATitumXFlJUoE4PH0zOocosoPP3/LPiTaupCiocHw9YmjnRnVIe0K7WVzsmjZ/l6dQUti7xMu62ZVbAMyzbSQqVq5thkvmeIL1/VMjJHg0lYLNuHZjHuneGd0osRiVO/x3misuGB0cSWeZyhG9D0/S0ehY+9tazlYw/EXvgz4sGPKQm+F+/FCkicsn1jLROo8VzPG4xGpV6IreG2MUbi2jIsL5josBJprjHeF27zSHx1fJce8BuaqFCLzdGbUZ/ZZSGuhqZBk68E/Lnd898PEb38Fqo16mjnc3NP3M00DVm646Tref30i+me68EKSE74VJHc8Df+Bn1ukaKGqgSTOp4X8PtCRKfPMtDhWP/Bl8RCMGjpq3Ix3k96yyorIitkXkmTMQ98NuODxNdPy9lLt/EDnHWKN4BzSAzc9tB29jfz2Fweuk2O8mZH8hefHT6heiNETkm4ELWd4BCc9T089v/9hx3Pb8f0FPp5nvr7r0Nhh4lmLgTY8DfFKqxWfPETHaoEi9zx+GjgtA+YjISoPcWXsroib6Ue/wVhkJe6Ma/Oc2kB1xuk501wHC5Q6kuJrUIme5VQJa6NpYbYLhy7hGxz8zP/p68DtwwtD/SN+PdEIIIF96ngX72j+ljwrP6njvz4KX+w9P04jk+voi/EhwWDGIQ0EZnxn0DyzCzSUb+5/4t/f/sjn4YrvDV3vt2qcAdlDFazNRA/NeoJ1VJ1pTWkEqt7g6j1Dr/z2m880NxGbUZ5B6pEfl0zWTLEjf/yiLJr5d4cLLhdu+lt2vuDdShXBe08iE8sKqjgTBif0bKz5UjucHza2u2aQ7eXunUdkRWqjViPJnpgacex5miKWAnNulFIILuGdYZaBuqWU7cB1qaRhZlk969rRGFHXM7rAX79XOu+hFdZ5x+fPd6x2oN/dss6FqcFl3dMY6MKAcyeGwUCMUo3eVY7eSGPkQQLv3lTug6d8Wnk+XVCbMV9Z5CPH48Q3u5mh29gAgR3PRGYXebkIFc8yOw6x0cpMpuLHPfgbzrWxzJ/pTZmuZzKNQYTkhOAN13b84U/3/I//ccR1tzy82dE7I6RNz5zTDeWq0Okm3kmeN7Yg/oVd+Iy4I1e353RR5sXwNSGaONXEs0ZMZw4hwzFutEnZ0LIm28rJ2cb/r2ScU5xz1HOlZcMMrpK3tWFxNKt02njXjUyWqJbxUUi7HU1nXqaF0Xf8zWGgNcfP1TC2Q2gtjf3OU+rWTBnEOFhknjbctvmRp+p5qsJpdSwSaSZ4rUSn3N4faVVIrfD+EHHHPbMzaq48rx5TT+8CiQam1H7g3DyfqtBl5Yenn/mQIv3wllo6Cg3PRHoNx1W6TQrkE8VAVGmlYHWDXQUnzHVmNbgsjbp5ehEE7zsk7ljMqDrhNCC2gXZ8isSiTHNmt0+UZeVhNyKmNOdpPiPBcCJEccTrhaFURnEMyXEpgSzdVgG1rUEimhEFF7ptrSe2jcekgdu+3ckJgzi0KeI9ra1UVbI5nkpGUIoqc6eU3nBa2DvlVicGFWhQw/azchSsNMqLh2mbsI2vbYsyFwZrkI3rvLC4iWdWWl1JrlGcspaZQCZ2O1TgHByXEPBq27+7XenkdqusLgsqM91eCGdPbIIvleoKL9Zv0ijv6EX4Pygt/D9+GbA2UJpyG5TBL4guWCtbP3VNTBl2O+g4EexCsHlzMIujWt7AQH7Awsw3v1spo9C7QF9hSkd0Lky2bujEdcJLpE9Ql2kzeLkODUeer2/57tuR6+K4FIjDhX//u8y4K/iQ2A2NQ/wZpz/R9EzT9dVIcUB9j7odFzWylo0CVh0/nysfuoKpYpaYs2PSgJIJPBNCxmtjnVfWoETfKHWFWIkxotKhuqVIczMsb8IIHz1RHNY2YllLHeucWFQYDhP/+X3l/+JnfPqZ/eFPLO17NBnEjZ++WqHUguUKa2DSRJMeJXJunsmtrPrCfvGEbqR3Qkhb0jl6A1cZfcDsSK63lO5IA+Z2Ys0FHxOiE+uyVW1wm4RD4sBXv7khPkXWfELiwrpOrCVxroXgemwpWJ0ZpOc2RHoCaymYD1QLtFI5PS0Mx8Ye8CUgJJozLApmgVQbIxO3t3tuYuBuupI/Lfxvnw3Z3XKyG/Ky8kEbT3MjFFDXbcAglzkOJ6L8mX3nyGnHx89fc14GUhQOQyRQ8eK2ZL46tHS02rFmTyFS5Q1Sb3nwmX3/jMkLUgR/dyDJQLgE/nCe+FZ3PC49n04Ff3QE57CohEPDYsYFBSqxLjSbUB9p0WFuh66GuIZzhRDaBqhRodim2u58t0mFEKQ2YKILF/bHzE/thin0LNcLWhWRlei2MbzJloNQbKO/mYPq8H5E/IE5R17U83hq3PYZLFOXEWl7nOtowTEtiRMDzzrQrKe3SGgd3VBxsRClkVRIKWIp8VWEr+9+5kP/hYebzN+uKyYv3I6fOY6P3OwK0VXQjsfLG/L6NX9+TDw+zsThgLiVSIezhjlw4rfdaT/QloYXTwevNSzP2jLOb69HXY/8+Y83nPNbUoi8XCrjCDG+rh+90MKV40Pj6AH3M1/LCafPWJs5FeGnXFkfjW5JDBIwCcx0nEtiXir/8t2Jv/uqw5uidaISaX6T7Sie9CqEWfIEd0I9G3kSRDqeW2McGuM+sZZGnpXkC0UcJULrOiZXuV49ThxHa7zxjg+1McWB3DZG/zUbY+dpLiGhsvfGm7EnHTwaev7lwoavtq3FhQWiOkKrJJ2597KxD1TpUuS//aSUxWNViH1PjY5UlYMXxgbXrJylZ5JI9cZLnujXM3ceon8gZmFvhaoVXidJ6j3rJhQAKq9kG1x0mA/8vCxcl0ItjSDQxYC3TVddzLFoRs0YpEdqQcpMeV0RXBuEKlRr3GtDnFBlcy1U8RQ1miRcWelLoWtCuy7sDgNu7ait0Ukl5hX1Dhl6siVOJqxrIarnpUGKPefqEBNMjEKDmNCSKZcLFgc0eELLiK9cc2GWhPqOc1NaWbnpPV4L5h2t8ziEuvSIT4wGzirVFnRMmM+EktHWqLWydpUaHLPB2hpehOagqm4AOUlocCw4OuepXkhugfWFofTsBbQZEj2tD6SshFLo+sZnjGL9JmsOinj3l70MtBBx6oFtZ64EUjzQat7+mCZ8eS68v39Pa4rWFzDBy9ZXLbrQGmTzWHjmeFjo54xiDGPgw/HAtDyz908k9SxlId52hCyobrurWMDVHrk4pEZm7Xlpxrd//onf/GJmuOvBf8HJBLKpOFX9ltztH+jjgePtzNe9wxfHv3zJPK87Pn6+8Mu/3hP9SlgdvUuUoafZidv0I1080bRwXpXW9TifcG6mlMpqgWtTnueOWvcM4UiSiSANW5ToE5GBernj5eWeKfdoiowy8a7/wv0wYfEHSn1CXaXURm1G6CPB35L8HeJHSg1o3ehn4y7zyzeVdzt4Ojtyifi0w4mj2QHTE04U0UZyEapjdCtpd+G2h6UGUvjATgNujSx6IMuCBEduO3zbcXvzzO3+RPBXmr5wNc9URtbSMetMscB0LXS5J+aRoiNFEi+50Ro0jcwlUD8Kh3ffEHlBnVJ95LQEtN1wV42xgm9XHoLyHx9mvvrqPX/9XeXbL39g8bfsdx+4L0dShWm9UGNk3I/U5ql5ZnUrPh0p1nE6dzyde9YCKcGv38NoZ5xUfOqY13uwfhvn0TPVA2I9b0PDlYzqsgE7LLBzF/5qHNhj6JxZ+68wayxXWJ8apx+vhDcrv/xdQGyBVpmLMrs9kz/wMg0stUOPOyKZPk54m6mat707DmUht5Uu7JA5k2yD1YTwM/ubkdHd82V2TNpRvBL2DbMrIpV9agydY1426MvgtlGkF4gYIpG1Bay+7nWtI7SOfeyQVphrYXWBs49MzRErhFKgGrU17n/hKV6ZqzD0OywmRssEd2HX/RP7/Sc++Gd8fETkBSPjXEB0Twi/JLSAcx5nxt4LwT0TuoXkv+KyekQ6fFTwjuyMWSrQcG0LAq5lwTCqM0RGnh4D55eBGA6EotSr8OmniWE/EOPGY+jTwu14JclCs0csfyb4FYuBpb7l/LkyPTX6JHSdUF0hayKroX7Pl5dtdXXoPOqEsgYkR4pWMpVhjGg2rrXyp+frJknzHd6NTOpoS+F427EscMk9uIJ1AsImhXKeWXp8NDoaXgs3wXjjAm6/Z742bnyjXwsjRqvGEbjBkXxgcYJpx1o7XL/H14w0tiqbBlqt9A2+vgmQM9+dJ37/KCwcthR+27gInS8MNA4YVpUUhOgSuEa1wJxBLs8Me8dN2zHnxqKOJfwrg0Eo5lir0flAtYZK5NKMKSu5NcSU5KAXRa2QXCQ4YZouFO+3wKhA3wqtFK6xhziiPnAuhrmex2nlw94ztytf5sbTS2OaC4exw6rS1wVKZX65Esl0wZPKtvK19QUNkMUx+56LdahFlI5vL4XLUjm5RB8jvRX2GuiDspaVJMbgG/s+wdQYqTQLTGo0ecUZ+46yS7h6QYPnxW+T3mu85SVuimyThdYtCJV975ilUKrw7IWzQK2KC56iijjdsmU+bJRbPIJH0zZZm1UIXcP7hXRp26U5JeZqMHrclBkUjsH4scJsPaMFejN8+AuvCfBbLWnJKws3eG+bvtZlxBVChDX3fH4yfn17u1ULPSSx7cZjkdV6iu9ZgLXNDHXGSYTg8Ul5v98zrmB5waUGZqQOst8Qou9Pnv7pwpAXIOG6hOt32HLL5UfH5blw/8uemiqibfPX+7gZnDQwxIW/3v/Em7jSxHNKH3gsB85rx58+GQ/B0eYZghJcY0hnDu57onymxUI/OFqMzEtg6HaMbYdUobjIPPe81LekFrhJQPX8/JL4+BRY10hPz1f7W359v3JzOHPXNVJymHlyO9DSHRLBu8ZUVh6fV2orPKRGLDs+nxOtRboRfvPLxq8fTiQ9UZYXQpg4jIGSHc9PjsoNw37HOADq8S1ga0XtmeCV235P1zl4EcTvCMEoYWTKRtUjqdzg8wn1E2s7b/kFf+CwU3pWDnKl2hm9aTz4byhPHd8+RR4vkVIKiJFVWBGSDlxXh9jIda2cZuFaB8RF/AHwkdAyay3U9swv0sSv/3bk40vhv3xR/ts/NmL8asPs9gOTCKVCrYlPl573hx0hOLzOHB10x8TL/MCnL40v8cJ4t01p5gKnT3uS7LgfAjCwZkcMgtCQ4hGLmDRKayQaXs98NXjaO8fLuXGZOnIulEWZF895dTz8YqQbZ14uysv6jpfLnqfphuf5QG3Cbmn8+t1MFxu08gpv8TjxiN+UqlZgdB3BDPErTVb6m4m8Vh6/RH66DmTpoW/su4mdq/SqJBpduOFaC0ErNRupM5xmYolIG/Ctx6zDiScwsHeesVWWrBQE5xvvnGHtQluNOcOyKu02cPP2gMuVXXTsO9i1QKmN1F9o8jMqzxR9Bskb7Ms7ghOc9RzTHfPiGLxtqzI9ofqCxB7Xv2OeQb1Dk2fWShMIrsNpotZMK4amRMkOKwM/PA1Ud0Ss2yilNdHmjh//eebuvSMdCu8ePLtugTpRr1dUGyaKS0aMkfmycQx86F69EY5aHUaHEslt4HTZMgq+39NqRJYV1UaIC2YnmkTm4rlePJ147vYDrYL5ns9uzw9noc2egYHRe8Q5tCqlQXaKpYhSWDVsY2yvPCRD5cy//cWBr7nQPX9BykQL2+GQ1VF85GkyTqvD9XtEIuIa1io4IYdA0h0U5eBB/YZ0n2tGU0C8Ur1RnBDV6L1jr4aY8axwRZCY0LrQYmKqjZCvDGJEF1ECL84xdYlUt0OzuEZoQjSjNFjEMb3qnaKDjZSyfQcwGB1bRsEcici7MNKvz3jvefSw+HXLB1QBnyiux4Iw4/j9D2dmDbiwoxRh0IWh37JajY7ahLnP5NbY+4XoC8TIWZTzq+fAu0CSig8JDZ5qRy4a+LPukfWZrp14Ox5pKeNly0K43Oj6AG5g0o0lghOkVyYtyJiYFuOlRi59xHeRVRZMM5NONFlJmtk5I3ZwMc+z2zEVxWSr0vqmBFNS2OyFXrb2njhDup4Xep4zmAjBTYw3A3F2XGpHMWWQC95l9sFtwUpJZNdtE6O8WST/opcBcQUkMufAy3zDMHo6v+LcivdXPBea73lZjfNVuE9h68iqginKQIh7mhxY8hGRHi8rkow2jKxUumSMwVE1YPMKr9anFD3H4f/H2n8027Jl6ZXYmEu52OqIq56MeBGRCCChiUKRRfbYYIf/l8YWy2gkDYShrAqiCGRkRmQ8eeWRW7j7kpMNv6g2GvEHrp27z9nua835fWPAbRewtdGdC07d2uvXhc3xRLrLHF3Hcan87h9ssfYOUEorOOuoLTMMT+xbYaQS64AxhblMVNv4033CXFmusmc6K9PzTLdNXH279okxE00yl3kmY2ntDVIOmKwYa/Fuw7kYOrOBJfD8DG+f9ryNB4px3A6GjZy5GRcO9QkbG7UJicxSFfUDrsuIWxikEDzUUtj5hskWrYlQI1oD33/K/Hw3cdPB1l3x6rpH0wPpOCPLNW5zRWyZeLlwCEKfoOYZTKOWypQc5w+GVg70NrAdO6xmujDSzw5fLeKUZKCoIRrPeQmUFkhxwdgzXX+mMwNGHYMPXF/1lAGCH3EmMVfllA1BL6hpxAskCajrsLbnNE38/ZOyFYcTg9pAZ0Ev7xm7xJcvr3lbr/h3fss71+G6A8UtPJ2V5yw46fnx6Li+hsNuppaKViHXKz69T6S653jMLDdQNHJeAku7ITXFtkznDaY0BhPJ9YSxF1QrVRtopuYZ0wJC4rAbGWPldB64zDNBV3HXMgmf7oU32xGGkfd3e378tKOal8S8I8fMaYrkaeL3Nxu2wYOUtSuPrBMUMgazcti1oLZQ/RVvP37Bf/7hBX98eslZ9zRZ4UqjL1xvEt28INnTTKP6gYdZ6GPDdxbaBXLCSk9tgURP5yote3SKHGzPDFykI1Rlz0zNE6fiOOaeSXf89MdH/no78MWmZ0Nkmy9oqZQGbjTkOJM/j3sLa2FzEyrWrOyGbrxQ7cI49wgNtFA0kfWZs1xzcSNFC6kpS1vZGl1YrZRahPNkyGVLLh33D54/vTvw2BzOwlgMvQMTK1wCD6eGHBLfvXQ4W2kVtBnQAHbd/1tnqcXSdF0PqFViSVymFajE5173FEeW1IM4YvYM2lbgWg/n5xOflp5qPM2OGGk0U3EBZmk8N0PRLbYTQrPgBBkME5ms4GQNgLpmqVQqMJXKVDO5NVLtuO7B6xkjFxgCzfVcjoJOZrUKZouyvuw6Z0gVZl3tiNl4lhghZiQXTGn0vpHsshpGnaPKf+3HF7woW2fpK3Sdo6IkF7iYhquZUYXQJgwW5zzqHbm1dT1Vlc5aNkQ2tnEUg2luraDaiikJx2rb9BLAOM6x0DQShg1qHUsF2xTvlF1taMlkZ4GOpo1aHcsMv3z/9JmjMSKmrRwWk3EhE0TIpbCkxJ2r3DXlhYXXhw25CUU9VT/XIVXYGGEnKx48RlbyZ/CkPHBJiegsQSDVRi+GZjpoDWeFPnRcVChZubTGuzgji+O5GRYBu1kJul4axhRCzVQFmwVyxAah94YaMwlFDNjW6Kvl2q0yveICpTnEBI7RMJSV35DNWs01ZcbYTL/ZcjkPtCXiKXipVFOxLTJ2hqeaqcZjpJHT8pc9DHjTKFLI6nj/bNgm6J2y6QMvNpXgEllnCIalNs4x03eWzo20lNeQi7VgNlymgRsZ8bZRXaQYg8iAsWeMuWA2mWYjVj3SLNiK8xnXLnS7xugCKQe8NWynif18xAILO+7veq5fT1zv77D2ghhLKQWpM94+YXymFIFNTzdUfm1mvrkduPKBennidDHkAuID87wFc8D1H2ltwbmZbVfwfc9GGkY7urLWG29FmUMhlTN1rmutI1su5/UFeNKFdqvUlsm5cTER23myzCxcyGnG2gU1T3h7xJkzfbD05ozRd/z+VzcMu8r3U+b/+VPHw/wlOyO83sL/wZ/49saRZo+6hHOFqjNqPCnP9AhG8noYkCsej9/xb/7jG94vNwyDYedO7PtGb4XXQ8/t2NN1Zw63HUMQWjrTEWkpEroL1kxYa0nLFae8pZ2FczuhQaBA0xPWXrjeddwMhp1OmL2hiuWSLjwcCw1lmpWiDust/Wjow4kcT8SSqEsjthcs43fcnwxm2DCOW+LlgvOV/TBge0fpR47yQLUdU7nlh7cbJnmFsZZSDNtcePfhiUVGNAwchpEYn9nqM5gKKhyT0vmV+tWa4M2A9DeU1IMRJnF0o2W6z9zNnr3ZMwfPWYW/v1t4tg5vlDcvr+iCQyu8+3jPORpi6Xk47fi3/+ULxvH3XO/e4VjzKTFBrusDuO+2VKcsteOHD1/x//6P3/HH8zfM3UuaBmJm7dSfFt6EC9c2QZm4kJjlik/LNe/fjXx5dvzVzZGxJeY0cvcx0Y2v6YnsO4vzhl2vXFrmril2WejrhLWVWToigagblovy458XvnkD11eFXiqpOjAetx2wGFIypKw0B9aCmvr5XtgoHBEfoBWCcYgVMBBpnKk8NUNpZt0wtkxnGmOIK33TGZ6mDX/3Nz3nZSRxxcNlIMpAM44+rKTIngs2X4jLwHmGy1k49AVtcf15xGBEEGtYFl3pegHOLWFFmc8CNWCNUJoStefuvCOYD6T0TJMDr3YBjGEqhql4su3Wf9eA1oTt1s9gNp70mfKIFiqJzIodDgaC80gz1AaYjtiOZKNEwxoeA47LhUzGpESrCdM7WhCi9zxNgWR6vhl3jKXQ6oJLZxThzux5soHZKceUWJ4jJk1YNWxMxTtWy6pUsirNrWsFa+xnc94qxDIhoNZxzhaTzlwFKDWStTK1RjN2RZC3hnUW0/JqjlXwOJwNWG10Ymkt4VQIzuGaQTE0NcSmlLKGfKsRrrZ79vXIRjJdWUmZ9fNUV6YTl5IxM9yEkaqK0Eh5ppHWRL01rJL6hjRDEsNza+xNR99tkOzp3SpDChb6DF3LeAd0HfdzWw1/UlHtuDsXurFHy8x2sNRosNqgVKyzlAJWLaOJOBXul8pDcCytsUEJbQWXhdZwInjjPtNjKzkW+s6xl8ZTXSdjOwb2duC6PWOC8t6MnGrg3DoQyxwtQQu73mJoeMp6wPIF0zU0FoJ1SJmxfr34hAZjyRyGHpcWvP63JQj/29sEojjbOLVGiYW+d4jbMpdVWnMdzGofszOhzWTNpDhxhcPLBmfWX1QtPWkeGHyHC5FkFrJeENeBPJD4hWg/0jYNzwGrI6YqTjOGiO8N1kBXOuy5MNSCVV0/JAqt3PL9L69Re+Z6/wvIeUWA0sCcqWahesOlOOzuwDdj5p+/OGOXyN1seBcHnlNBqaA9x0vPfl/wdqFxwrmKszNKh4YvsOWaQeGbreWmu1CZ0fxMZ3um2Pgvd5nYYAxPvBgaW3fFFHuyVTa2MfrM1hVkdDTrqHhaEUouODHQKg1IsrAflf1s2W2uuVuUi1p+iguH+1XuEReww5FhvqyJdJ/AFLztPsMyLE+nb/jTz3/F9+evuPc3SPXYJXL85YFgOgbt2AfDTX/P77/+hX/91+8Y/XsOQ6WUGXEVrEH1JenY4R4NLSjSjvRmTd8bSdSWucyGF+MW2vrQWmKklcx+q7z8ApxbqKWSsmGOSk0zRjKdMTixDBxJ85ljGvjx6SM3pqc2RcmoXigtMjehuTf88Zdr/j//ccvb4xu0H3m5q3wRIA2NSa95XCrLKdN5ZRf27PtA5yqdC5weT5i68OUr4fUuEmhM3SvuZ0vKQnWeZoSbrWV6dJzVcKeOZ38gniN7d8vII6+2J95sd1zZRn+GXxblk4Fz2/Ln+1/h/pfE//DXnlc3T5Q2E9OA2kYqhk+yZ4lbPny44m//9Irn9luG7oCzlkhDccw6cj/v+Xh+4Dc3PWTLqQXEjGR6Uur506fG6Dy/Gh+pOvJp3vFpLsRSOJjIf3fj2O4mth3IacEvha4taI2ItWTv1sNwHbn/CC0tME386oWstLzzQm1pTWNLBrfWB9WwWt6ykNoFvJDjDOYWgkNNw/aGzjrk4jg9VvAWlgimMnaFPkxknUl4fnnr+PndS1rYYUOP8xskGWIpcABXKrtcyOVIq1Ca4+HTwsuriMiE76F+9iPUuuX5scf6QK6ZZBsalV47OlEKjQoUM/LpyXPTDWRVFiyD2VJy493ZclJDc4JYS2urwtt3jrlmsto1HFoj1EwxMNeKS4VtF9goaG3MzXMxA9mMnOpMtY7cKlUrKSYW8/nhXQqaZsoQyLst98uWY+oR7XgjHS4mNlrZbS3fk/g4LdTgKc3znDNjdVAzm2AQU2lWsMGxJP2MNwdpBlcTg82IFhKeiMH7jqAF1zVyXJhZCX5RCwbDUhKGxqDKKB2GRmsWY/u1v58TgxnwpiGskCyawcpqpSylUFV4lMiZgS4Z3vTCaBKuRLpFUW/xeSHNmQ1rUyvmRg0Fu/62KKZx0ZVA2PVrFsgaaKKc4oWrw4FxNkTT6I2FnOmswWEwktl3mZgMZolYXRBRLq4jauN6v8GkR8QIIlDban6tGEZr2IngLDw0iFXJwXBOiQ2J3kyAEmxAtaCmw3QDS4600th2jtc2sBS4oaM3BpMqRhPogNYVJO6sQ3wH/7UVAaD1c8MsIlzWg5aa9bDtInkxSC3cBMGXVXksn42of7HDALVgm6NrhpGeECu5FvAOupFDeGbTzaB31HxioRGlcSER2JCKY1ZDygM2bXHbyrB9otU77FDBVXqeqfmeosdVPoRD1YMqO5sZjF1P5KZQngu19pytZT/scbVQmlJUmac9P969Qc0zQ5joOgsouV0QcVTxPHPhzJEXfofJM9Kgyo7Ym/8tVGTF8eGh56s3HdauBqpcIyUps/tEnT+RH2/wXKGuYQL0IdJsorMTm11m4z0pPSEhYblC4q94Mh4NM4NvYE5UHqEslLLQpKw4Y93gNWAJMOwp3TVOPW9ulX8SPnDlLFMWUp4o8zP3jzOH8Zle79nZB7b9M96fGL3i7QZpO54fv+I//e1f8VP6jjpcEVxHpmG18cXVFaeTMNUD0xR4exw5a+Of/u57vH8i5QiiGFZtaE6F9NyQtODdhYOf2JiFoplmYLFbtAW87ajLxLIUFjaU0ujCjOgH0Ge8L1graM3kemTwlr0b6Bj4J4cX/MtXwsts2fYZF++5/zTxeK7cnye+nw/cv4vMdc8ffvqGD9PXLPYKq0q63BFsJrZ7gitcG0dRSyyNjxfDv/teOVeHDVvUfEkQuH6c+ZdfZ/7FzT2DAv2eOEdGa/nd0Ph2SJyv4O2z57gcWC4gUhltZe/XVdrEM1trGbdXcO4RDZiiVFWepu/40w89B/uJKT1zf25sbhOqmTkd+Pj4hu9/2lLdF7zqLK/KmZwXnrXjJwaekiGbgYfUrfY93+jUELqAmwqaQduO908TX23WMXaQDYIjOeWpJk5l4WZsbIeeMTfqMuPKQpVGcUJsupLLtOGkYzoZ/rxMGFG2nYHnSpouGD8TQqQ1pawtJqyxtNao7YiaiWpHtB+JziKuYELFuszGC51ZR84tNdQU1BdyTZwrHJ873t/tiOYW51Z2xHIqlDJSBB7PM9c3FnNecFLwVmkFnp8rSwJDXSmNDNQcePwUmJ56ulCRmkhjh/RKuz/R0a8XF2OpGHJxiNmScmUyjsdZKNnSJGBco9mKmtXqZxUGa7E9EBd6FpyLaIY5W55ni3XKm67RybL2/6WSU+ZkhCojRSNoWnHReSa7C8XEz+uOCNnT7EI1t5ziSNd39K4xyMChLAxl4jdb5W2c+LEGqh143yI7hOgc2NUr4Aw4WSFoNlmkGagWnyvbmtj4Sm55RV1T8WaFKFURLg0mrWQEWxq31tJUGIsylAXbKU4tFEcQy2CEnQjBCVFgqpZcVzgTWjAWcqk0P6wj8HDNj2nh1cbTLY9cy0xoCjXTasUaIWhF6D/7P5RCJXXKg1a8Kj5U5tRW0VdrJBpJF5BGLgvWOnoX1t2872lUcnwgL4GDNkhnNsHzZAPVd5SaMUZo0mjOkatbXRSyaodsnuhsw7JOVUpRjDHEWJFecEaRmuitJ9eE6TqeZOC5WEpq7ILhi+DpjHCsjrJ9gbWVAwY04mxAsIzisZbVl4Al2BEnFSuZjSak7alaiFZpCCH2dCVTTaZYz4KslNe/5GFgVEXFrWn50bAfJ0QXmvOEFrH2jNNEK2eUihf3eQQoJNmQ2aOyZ6nXVOkRnjD2B4J5h3ULlYwI5JbAFrwN2FrJcbV5ZX2gL4KXnqwNzRZvPa33zKOnxIW4VAhP9GOi708IEUtFGxhZYTRSoYnBOL+eyE1dqXAi4Cqt3iGmozSIzfPhruP+dOD1q46cn7Bt7RUXnWhm4hItvm6IZaaakW4ofPnFa6ydEAzm9ISpM/9VOefHxDddQ7sTzifECDEq03lhobCoYkY6E4wAAQAASURBVIyjcxusdYBbH7jdzP4wc6OJ1+UTMytoI8Uz283EYfuMMw+g92AvqKs0gaaWmBa87jidv+HD5Rse2sAsn0EXJrJxkf/+1Rt+9D1/f1KWAlMNxGxxatAykEtHk4wUw3I58O7n15C/pfcjO/9AbytaF8QIKUMKQu+25LjDuw13lycuNhBC5XqjeDEsBcQYSkpgG34AVxdMnqB02PMb7PzIzf6anV0Y7RO/+wcDnz5V/v0vlX9/f8XluGc+b2ntW8bhFSxt7X0vZyYPX1wrt7tGc56l9lwujh8/eD5I5VkDpziQW2a3H1mq5fJ3v/DlP575hh951TduBdL5jGTL9YstX38BLyO0H99xWwy74cj1+EzLE8qIaviszb0h46kCwRdsnsgkSgxsppF3PxV+mT3fHQTDkU/PG356uOFZdnx3u+WmZNxy4ULmwV5xqoH3xnEpFXWBrnNQF1QrnatsiCSUpMKcHHRXtHSid+CzInisGLw0gons8Ly0mcUXui4Rm0FbpYW0BptqxSpoNUh4yd3jkXmn+F7QlhEiK8JGEeGzBRJElFSORAqVPXbYwrZHjVJdwvrCxlZeucqUCwtwwnCJ8OPzSDg6fvlp4G16TfY7LMIpwiVBWSHy5PPE00b5sl9H9o4GWC7Llp8/rXmmEhtiBvIFSHucbNhYwdeEFVnJpN7SUsPbhjMNtAKNVqBpR2seW2BXIhtTGaznralcvOOSK6ODb14kdn6icsaaE6PLTMc9/68/BCa9pTTl+Tzxq19t0MuEtR2+BsISec6VKoHkA7ElQk1k2yh9RNyCtQYlsgkFQyVGsF1H1IQTQzMBbYk+zvyLzRbzmCmuY7GWZCpVlGQd1iihJWosa4bECosRinF4sXiUXhSfVlfAYO36Ms4ZkJX+qJatGg6m0bcKw4bny0JWpUPZGsexBZyz9Go4GCFIZdJGpiOrw/qKqTPSMr01zFrINKo0pup5N2euzQHSxFWoOElMslCNkk2mesG4AGpoFbCNzgAKl7KSX52xSGcoCc5pJmlmaQ3NymHY8XU3UOPETGFRQZvDtEYAjGn01hCdxVsgsbZKZCVXdgJZE14byoK2yiYMDEBq9XP2x9CaQ1gbFFvpyMVR6Tmr51kcWFYOgFZuZTUA+25Lb936t+YdWhvNGLwPTEk/w8ccjUznGz6ndYLZVx4msN0XzM/rZftWDb5WHiSzoHTmL4wjvpRErnAxjvhJQC2h22DHwKEb2XZbfnW97sUQD82uYA4NnNPAMh94nC1PqlQF2oxt68uIOiO2fSbngapgjEVMoJYeYwKekRorOZ2JudKscjo35gonN/PVl8JNuPDNZqHfHCn6CccjziasC2gFwaPaIeqoGMRaii3Y6wukJ6437/nXNx3nM5zjyM8fA0scuT/1fPFlv+YXyIhdHQQqlWAHfO0JznLJcH8P3ajcbjM+ZXQ2IBsyDT8ecIOD5YTKtHLmMRh6aAXTGaQtZJ1wpoKxpAxVL/j+CaOREh8RfaAPM13IyK7Q+YozH2jtDrUFtaBq8K6nVrv2ts0Vw/CSr64X9vqW5jJ9f2TXTVyZjvldpJWvSTkQi1BLpHcTg5toKVLbhqIGY/d8vP+SP/zyDUlfELqAUcPNfuR6t6WkmZwMkZEiniF4RidI9y3HcqbjA0VmPIo13WemuMX6hraCtIQxClJopiKd5+0xMYjlIANX54nffLlFfeH/9tHwc35BcLfs3Q1GLUOw5NxYauAyrcAdg3K6NP72hwsfzz2XvGEML/hqs+E+ZU7LBwIXnD1Q7IHUTvgANj3gUkff4DTDu2VieDXxxTeP/J/3v4A+MnYLOV3IbaSkPa1dYcTwsb3hssz8pjuyCxdsukA985qFEB2XfM2fHkfyz40vXt3yaQo8lmuKd2z8xKHOSK20phjxFHZU61el6ujYb4fVwJcMKV+wrC/irEpTR8oV7zskVnqxuJoZTGNrE64mqJGvrVA7wXkhbeFyXvi4JC5AKJkhNYpajHpa6UlLoe8F7yqIYODzQRu8M1gRVNckeWoV1RnTXVbRjWsYl7BWkRzZq7CxcGqNLMIJy7vzluVUeHy8JXO11lNbYkkgOiDiQTtahrcfP/APfz0i3YAkg3M9UwzcnwIbeabrE61kmslsr6+5MGBywJ88W5TnkmjhCuPWyYmIQ4oFZ3Fuy6YYbIMXFbYxUtNM1wnj7cgvWth4+Ov+mX9x+wumfSCbBbUzQqH5X/MqfMv9FCgmkGIGOnq/SoyCgVfesW3KRQ0PpZC1Il6I48hUB3rTUWOlqqCloiXRKMR6wUojiCeagG89vs4MzLwMAw+ScT6wtJ4mGa0Fa4TeVJaSydWhwEWU5AO+dmugkjXTkU3Bm0LNiWIaVYRq11vxtjVeBMPOKpOxPPmRow1YIh0dXQvE5gnGEETw8YErXUN42diVGYHF1Yldq0SFDzWxGIMaR26GU62Y0CE2Qopse8sijUsrZDNhfSMgdOqwrXBjQZsSnaNkszY7jEG9cKyRopGIAmCz4SsrpFZ5xxqO9lKxZl0x59xwfp0qlFqRBA6/BkgpeDzeCJayToicwSEcgme5VILA6DpG2UCJnJJQZKCvHeI8GwfZQbMOI445Z04GNh24fGGrPV2xbAlEW1iMoxFIWsmsJM/eDQRjsfEEpeGvPJ8mw+V8RZ0dVzR2bUab5UkciqX+tw0G/tsPA28vLzmVaz4cR07ZI3bPi+sBt2QGl/B6y6txz8Z7mjqc20KLBDNQZEeqG0xZ5QxqFG9XPa+tjtwsNqyJ6qpC8B1ie0r2TMmgbsDZPTF4LmkmUckbz3XY0kpmZmIOT9zsPtC5e0y7YLUgWlBVmhasdIhYmjha6ynNgwtUiYj/APnP+BDZe8PV1YDhFV8cvuDPbwvWFbQqtaz7e82fQVZiCL5b2QfW4b0gNfD+XrjuPMQzVteAkLGWOULfV0xayEWIaW074HqKGGpbMK7DtIqTjFRZDzFmoXMT7byw5CNZK7lk8DNOhaUJnTeowhrc1tVpoBVVaGpwpuMwfOL/9I9mnH1E/BHMB4xJ1MvX/JdP/3u8ZKxTLI6mEUcEE3D9Hq09fTdS8pbTwwuepldEv2MpStQXmLmxebpwu514cWXJ1UDxgGLkgqkTWz9xtX2iC+cVw5lmigpzmsmXRw57wXcWTdBKx+kcuNpfc522LHPkuYIxHY+nyM0Q+PbFhvvngJZV12vaAlVpeDKWuQXm1NOWC3/4eeZT/hWzfcFT67mUjtMFprjw1e3A11cL8/zMtBy5TAtl27OkiDGFrhsxzRPnwOmp8M1vn9l0f0A4ojSsc7hZiQkMPWJmrvtf+Ccvn3m1+0TX7pEmUHt4bhh5wVyvOcktf/dR+PunRBRP1sqbLdyEhj0fUTOyiF8d82khNQcS2Q+NvrcsoSPF9SGBGag0mskIQosVKUo9TYhWrjcjN6GycRmJGZZn9mJxLmFqQcNKUjxNiaNCME8MkikSaGaL+AM5ZSwrwMiorEhwA20FFoJxK2lO1gdLRRFTcL5SUZoRGoGclCAZIWNGx4wwAUsJTMmB9uxMIGpBFHJ1iO1AhVoaTdd676Uo280WUEKBHD2//NBhe+F3fz0gGtHgeH++8DAZHuL6Pdi2hMi6cy5VqM3SjEHEYa3DhcAYLK/EcbVMuAKpQqmGtMx8dbsn1cTv9w9s28+kdCFLhC5hA7j+maud4B+UUteLU14Wdl2jlMp0OdNi+izZcQQUWxLZwt+fwXODs2srw9VKKA47etxcUVepVletdAkrLEcjtSTEeYoLZK3Y5vFOKBRMLYiz9CHgp5UaWFqlSKaEwNLWrrsCxgpqZa08Y6jGkD77GRzK6okVprzQGMjqWUrihVunCUu2FB+YW8QUobcG3wrBWVJseBXG6tgukSuzNlM+AJNmjPQ4d82lzMxmwUjh0hYudWIGYlsgN6TBtQ98M4zYeEcnFu9g6TwXbSy5oNJYmtBMI38muVIglIrLjX3Y8j7XddRfC6IVo5lWPSk1iq6ch4MLDN7iLmeMNLAd2Q8swueq5nqQcjhI62djqqMoTM2xWM/GW8iJ0Q50RVk8JAPZdZyzR3XhRciY+IxrA1oEnCEipNYQF8g10YlFGp9x2aA1c1oyl2iozeCMI+QzY5uJpWL9KlVD7F/2MPDHyw2z7HmkR+x+1TnOwiYLfuM4xpfcPY1sXlmkNVqpGO+w0rEkocuePie2vjKbBbGFWiZEKtau4JTWKkHMquGs0Jon1565OKZW+XQfictAzAHweB/wwwZP5GpzJPQFw5oit+IQAxhB8RQdgA71W1Ldc8kHzuoY6hPaPoF9pmoi1whYjDxh+2duD18gvuGr0CNkK9Apahxqd9zPlWmJ1JqJ1nHCY9KB3xbo0xNeLGKUYizWbNCcqa0wF0+ysKgSjcGjmDThiAxeGUXxqWDowLkVTiGVOo4YO1CzoK6nSOUyPXN15QjdllYnjDS8CytwyXic31CXRpk/UfJ7irmjLkfwM/0Y6MLM6xef+B9e3fDP8o4//fDE/3eZ+eV54v/+v94yes+uH9htR9LZckpfEva3+O6KXiv3l8LTonyYHUtM/JMvZqwsbJ3hZnzGc4cLCeMr2IraBDLh7CNbWdhtJqROWC90XGHjyFy/5B3f8ONFeG7rYa7frFSwx3Tk0E385spwv1Qel4VRhB7YHFYd9iyF2CxPl4BfPrMc3I5T9EzVUcyw+s5lYZ6f+KvfeWyG59Oed48PGH8Dc+RghCup9AYSwuOzY5oS/RDRVj8LZHpa2WHiLbntSVZ4daOIHLHLB7ScEDMiZo84RZnpehA/clkckgu7F4HXY+TXwwPXtmG94Sln6iaAJN6YBnki+cbX1x3NWyZbmRRim6CuWmMnhSCJ9PyEKxlLR42OZi19v6JyqwjWKRoTpUz41ghV2NSF10HW/7cmel1YWmNKnugSw2BIpwtaG8aBaYoToarSclvFmDiUinGr/bCUM93oyS2sUKonoeqAN4qtM14q+65jqpaTBoSKz4UxK4MWzrYyh8BMpVePb4aYC6hwjhPjUGhpzTGmmrksljoPPD1mrm8DD49H3n+YmeYV1T0YS14SQResg1gh1CuGHGhkmrb16i4zrhaMnbCjRZJHmlAvmeE68mpz4U14+lyhG7ElU6VQjSK2oCZi+lVH7LWQ5wvVT9RiIDaksoqxTEJTYtdgXixvWVsDYHDWsrOObWtc7Tp2ZuB8iaguVFGyZIorGBVqqWTNRF0tpVYqPcJoPTWldR0gFR8EamRbIwdXmFtjEU+nCa8B2h61QvlcVYzAUiqiyugNVistrWsZLZ99ASXTd09cG2UpI7bbUUWZ7ULzBd95BrW4AmOEbVK2aqk5ctUitut4bI4YLGoqsVRaUoLtmOYz6gxLW1BjqDVhjOWpJPqLsOs6dJlwxuLNSgJVoytVUAuoItbQOYdfCtUmameINlMmi/Nr8NM7iwmJLp+R2BGHjnkYOOVGioW/EsHRmFslqiV3A89SSNlxvkCtgRXQnhnJtBCYikOdp9JI1tLEsKkNmyOdBIzZckG5my05JrZXlhYvGFNXJ0boKQ1sFWzz2BZxZqVAOiIEy3QSDnbD2M14k5DpRK/zKs3ze4pvVPV/2cOAGa94flCm0TP4gavg+fawginqtDB3HZ8eB75+4dcd4mcxRMwwxzPniyC6oc4JtZVLsmy6EWkZNW69tdeKtLra5AjU3LEfrtnKyFws7qXhl4+NT2chaYdmj49wOzYGXymlEpsh+CuEjMiEasHJDefTNXHZEuNIlC0nuyPjwR1BF5QZtYrrPEUa5/hIDYq5MbTkqXi82677eNtIZSTOex5OHTkZXBiYGpyMsHEdMUIlohJXBaodyG1gWVbgzLjb0MqRGh/obCbIE8ITmAVvHVY9GFCtiBaQjBv3tHIh5gsYj0qP8QNt6VhyBjPj/Q7VSmkWkQ3YEfSWmjdoHTD06w5PdtiwJarBZOHKPrLXXwjbHf/0HwX+xWvP35xf8/78kiUE8nNG34HWEdErlsz6Ag4r61uNx1nHzj7wD68X9PgeFypW32FlwjQhJ2WhckwLbgDnTgR/xukFqxVTwYinmWvO0zXfn2+5uwSS7UAMU63cxUde+gXbF74aFf/VwOmmJxWPEWG73fHqFv744Y7zY+J5CXRTYF4ak7uQ2jp9qprJzVPtlku65/L0hOTG4m750L7jP/znBzCB/bChu7zjn71pvNj0tLlRs0EHiyZDNYLB42RHTh0p98R+i9aGVw/lgGq3brXFgXOgyvi57dFZj4uKOR357gvlO/fEYYn0fceu97w5bPkuVppmPi2JR+14fdjycZm504LvlVcNnJtoY0+i0stEKIlWlGgcTXZcjpb3OfL1tmffZcznKZfxnpJmRjX4ZrCuYq1gsyKfE9OmKjU2ztq4WgZaCZig9M5TY1o1KBZaLUgtiK78tM56NBU6uyAE4nnHcj+snAWdQBeoDW9gazYsjwaZhMFUevsICZCORwzieqyYNZtAplEo9cKsjzS3gyIYLN5vWcqO+8cjw8ayTIGygK1KLzMdDnQd+iK6GgxLRMVjiyOePRmhD0dMvKxGRDtQsSzRkJonPZ357W2kaw9QG2jCW0F1VfVWKt6tet6SElYLVQvZOZ5zZcHgvKPRKDkxxkQTQ+p6LlHJ1tH1Hc5akkJzjqlkXEi45USvhc6tZsZVT26wxmGKoWmmqWKtUqTitdE7h2sVYsLR6BSuKPQoswqZyiYYWsmcSqZgSLLyUy6tMq9qL4wYWi603DAm4SUws9rRNU289h2LKse5kfueqiO1TFxte+ZLQl3F+oqbBaOe0pR0mnG1YwgDiyaKKMVmYpqpZBSDbaumu1KhFHDC3BofsWxN4I2v1LyQdf1ZSs1454CGiGCl0ZvM4A3VO6rpqdqxlADiV9Gb9IgWQqt0NnMuSrUdxTsepkR1HX2pq/QNWS25STmfM2pWcJvVxmiEK4SZgOJR9Vgt+GYp1pBtY5QeTfBEptAR5Us+zjO2PfDbnSdgaWagloC3O9AzxoIXR9NKtA21lbluiHTIKAxt4mroiTJhSsK4niyGav1nvfdf8jBgVkJSU1BjEVNJZcEJhG5HbYZTumUph7VJnD2xGApbknkJ2wNPx8yx9pxLx/mt4c3uW0Z7h+jEMAa60CCf1nFg25FiwNjGOCqGEy9DZv9m4Ov9NZ9OPT9Na1qyd47eKCLrSyOmiWoaFgN0iOyR+AppLyhsOCXljKx/GFaoIqhE1EBtjWodVSuVC3QPGDdyrmuoSFsBbaheQQ6MLqKSwPVkZ7C5su8cSE8d3jDFiaV5TseeyA25DuR4JvgLL98c2G3ucPYd2u6prJXMOTnUXXNOI3EZKSngfYeRQMkZTKXWgtie1kaiMegyUxQ0JkQEaYFYOordY/QF5nHPYBy962kE3KA0SSuyuUTEnSjljNQbpCrfbSNfXW35/qK8u8z84jynuOWUO1Q6LtPEeTmv7QzjSCr4UNkNjb5eMLJazWLZsFSPdYLrQeRMNwhJJ7RUnIk0Ks4KUj0pb8jLLegbXnZbfvMCHpcL0xLpfeM3X1tej4WtWcil0OITRgqf0pbsep5PH0ml0rsTbpOYTgXNyiUq9/lICgE1jlYjKo4qjkvquDv3tNnwPlmeeMNT/ZqYLOVc6XjJz99/4F99K7zxZ7StyV3xI6KK5gzlTB92LM1yyg7q+tJBN5hmKGaiuYqpW2iGL646fh0XmBs1TeRa2ZXGwZ0Y0oIrHsHSZY+3AzkvdINy5bd8dT0zH//E19uMdD3eD9jckZ4bT6eFU8nky4b7fMvDsuPD7Cl+y+mpcfOx8MW3kaEpS5ppzuC9R9KCRei7RquCykCxQjGW5gLVdCxNef+057zcMA4OaZkAn7vzghRBkmKtRY2CGlxrVBpaB+bplqX0BM14E2l1Rs0aClyWgadni2qHC5m+y9SWkRJADLYUfDN441hMJWtEQiXV+FnKskE/p6ibeKYl8PCpkJOlsw1tilTFSgY7kGUkhQ2nuKUw0rRHjQd6/vjTO/7xdyPXwwnvA2XxJG+Ym5CaoWnB1TN9Z1i1FGdqWLv36isxfzbl6YxXSzDKJcPjR8fjcUBsh7XQZGKXZoaYSGE9nHhjVrW7tasQSAxiPQWomtj0sKkVj6z/RqtrxXEYCNGz8x1ZMs4KeUlAI4gSfMfShJgbrq1tAVjxupmGaOZqGIiXM81saFY4pcykjWoMjcZxSbwKA3U5Y1UJulL0ZhESwkaPvNhOLBdDbcLSoEM5nZ/WSW3LmN4iC9TkyKWh2mgxEXyHbY2kjdwiaipLXVn+XpTD0OF9o8jKrLmIMGnjz5dG3QzsUNahbUWdIratAXNRoDC2SmfWRkQRRynX5NgjIqizTK5S6kJ0GTts6UpDJa9ZChs4l8iVr7TimJPlNGVacni6tbJKBSpWHPsGuli2bkPMFqlKVxUOgLf0KdFJJrcN99qxaIfYkffR0IU7Xg7KBc9sHc0ozjS8MSSFyVk+iUWBRR3L4IglsdOM04xzidx0peLKio1W+QsHCPN8x66/4W5+5jFHordcjqtBKpjCLjRSecGX018Ry8DTRVhyYGodp2VYx6dFPu9yO86zJbx3vNhe0QchV8VJYd9VNjYRtGHLSFPP9kq47hOSn2jlhOiaIg3WkMWz6zt8zYip9MEj3YCRjKkjkgbS84bpqUJf6UaPM1DOJ6iJfjiRyagRKpFa186vNX5Vbua1zz618DksFfAGTBGcfeDmNtBKpNQ9uVnEbbm2jtOz47nd8P3jnvu6IclIUiEEx2G4xtQj0+Mnfv+m0cmJomntb8vqozvXylL2/PKwp5YDajbsdntGE/C6usSNdsRLI2tFs4CMFFkNa1t7INARc8cpvkDagdbNVOcR12GJiFqkNupnQEatlvNkQLYY2+jKI7/tzvyDnedcOj7kwM+nM7hP+C8h5p6/+wAfFmE39HRuZmvO6OQpy563z8JP84HULM4K+wO8uDkybD/QuU9oFaxfT7tJHdKueb77mvnuJajy1TbRxwceaJTRYfzEX7+B/aYwz42S93TdNQ95z2EKZGl0HNH0DDdnTDrTxWfm2pjsDY9NyHGhE0cuSkZJzeARjrqlH69JsVBKIPgt1g0kV6il8HHy/M8/fuL/+E2/3pOkoWbV2Bo8agvSFdJxYakJxHwemTtQs7oIRDDdiCyGfcv8xl+Yl8giE88Z5sdC92XDSUWQldJWJiQ4So7Y3hPchY39mWH4G9QnXL+BeoWdLHZWvnA9l97xsVPezXveZ89j8QQ3EGvhh4cT//03CnbBDpalVKI0vHWYDHtjsbaS3BbchmgMkwZOtWO2HtNe8HB8wavtgOoMyPq9wGHoMaantojBrN1oCcxlx/3pBR+eX3BOns4K+xDwbs9sR/7+uOeH04ZqB7bWI22BVhl3junS6OsKsLGSoK03vWQgOf859CWIMWRniCWuemKNtFRx4ugCFOegelodubgd78vI06knxh2D2UJzGDxLdfz8KbPdLHz1u0esPGCHM+4idKVnp3uGrmNjF7rBItqwLmFDQ70QZcfj/YasPdgVhY1pfHx2PM0jpr/CGYdKRbD0bSbIQmmVWhODDyumWsEYg7FuvfGKo9TGoLANK8inCZQsOFWMgjWKNwXj1/G3o8cvC52W9blmOpqzCIJrCT6/xLPzKJVWFoa+47g8rRpdEVI1NO+wWM7auM+NF65DmuBKBN+TnWcRj2fB2wdevnLcXSKLO69IXiOruS8rSSDtDOkp0VygINA1CjMijlwSTRpVC9aAeE+MlRgbX+4PVCbmtGAVnsVxsY23yfDKbvl6tyGXBy5GWGrCasN0BtMqBwshLcSl4YLDLImQx5Uqa2aeOJJkxttKUMXZ1eVRSsNqoUkBX0iz4xI92fUYaVhr8FbXDIZZg7T9VJnOjV1vMKbDWUcIUDKoZHqfOdTMWSEXjxGDGlCz4z5f2B2giqFUaHVBNVGxXNTynBOPObM1Br/ryQ0uU2NRED0T7Ew0PWdxRBNALd6Fv+xh4IvR4mi8vnZ8Wp64v1RSPlDsniJCjZWfTlv+y8drzsuFpTjUHzjlwBItuA3qVnlJaUJskbkphD2X54XzstLYtlb53W3ju+0Jw4QzwjItJFdW9G9Tmi50PrKvG7x1uPVPCiMerVCKw9oerwFJWwKv2W423OeMtiM325kXu2dCd+aq+0Btz2AEs/KTP6fZDaodYGidAWaMa6gUpAqkBcMRsSvZatAeUzZcqnLQnqdPnpMZeUoDZw2I62mm0aRyThlsgLmjNQeyVlys3YIRrOtJZUQYqHbPWa4Jds8lW0xnV/KWWRjMxIv+mTFccGZGW4ZkCCGsk5LmyOaWHx4jdTxy2M9UTev01XgMHarrl112nporsyrNw2YwHGSHnyPt8YnlaPiUA2Eb+fb6E9v+RFHLb2+2PF8KuHU15Ocea69pfmbjDMOyMJVrfjmOxI/Kbn/iN98oX718iw/KnB3LvBC8I122lPI1ZvctxnRoa+xt5hThnARplZ2xnO5m/u44cC6WYhrNJJSFrY8EjvRhYusiG2mE7Pn0tPA/p8jDtMMINBsZtwOjCvOiDP2GU5641AuYjq2zZDHMKp/5E+D6ax7Pmf/w57f8q18f2O86TDtjracWDzTUJcbdHvccqHWBNkPNLDFyyT2XOPDq5TV9ikhOvLLC0geSgI0dl1lhvKIkSNXw/gmetNJfK9f7EWVmsBcM92iX1gCYRIIEzFkw04K6SAiNF9eGL9PA92dPckotC60oKRv6viOkxHzJXELPxxj4/v6MtwbZDmxfOZZsqKbnnC2naniujmIFZzqengfqyyuaJqqk9VaULS6MiAqilpQNSa/5dP6K7z98wd9/fMlRr2je4xSuZCDUPc8Xy71eUV2PDYpJib4KoQnVJsyVwV8MoTpqnTAGQrNE4znlhc57ls8O+2jXYJYhY7pMNRWDxYaO2gKmbHi+HHif93yqV8y6BVF240jIQs0NgyB5x7u3oL+3mM0TAIe9oa8FCVsO7oYDdxj9QHVnWkg0EcRd8fxwzceHAxM9cymU1jElx6UCdo/RAVOV1hTrB6LZ0PrKJJnnqmv4z3nEWkQssa3TAduEVh0uNqxtOLFrWLC2VQUtZp0WuhXFXCrQKsHARgJPc6IGaMZgrOBbwRjDY3E8qIHmCLWAqxDWUK+KIDJQkyDSSMbxIWeMtWwl0tnGUBX1HdH22FKRcmFrL8SQSbYy5wbekCStKdOaCUOHd475WMjIioa2yilOZGcxRvHuc18VC91AKhESvPZbppjpMSxVOVvHWSx7dezsFqTRmZlo7WoQ1YI1a93R6NoYKOWBzm9wrRGzYfFHzv5ClZmkjTlHDAltEd8gVBg3HY9ReWodM2CNrNMbC9FUoq7Bxp3t6EzDqkWaww4j0TqmmtFssKlRhsZ2I2zPE66OVGew1pBrYa6OwTV2ZeaFWEQjWOX9dCFVT+vCCtmTRm8qVjOTrtOAYCK9mXlwPQ8SOPNZLih/YWuhzSvZ6OXW8g9/tWV6fuLn55mPeUNSQ82wxBX6MS+Z07wmRCcC6gZKtRAFlUZpjSqWqnB8irS2oYaeqsJSGvHjEZNnfndVaXkmXgxL9ew2L9HqiXlkzlui3WMYaGYglQEazKXn54/wfOmwdkvQG0q8Zc4e0Zmrwx2//6tnDtu/wcszhkwpUNVgVNaxnApGLaWCM6CurZUh8kozrILRBqbQd22VhcyVUT2KILXiwx6nhpc7z9ejwduFskyc48J9KlyCZ8oG1bDuyvFr3dFYNAuuWgbjeLndkZ7MGtLylhQz2/7CNtxx6J4Y7T1GZxRBpUd8RzAVkyOiHUYLv315S3SPeP+OwoljiTxkg6jHMYJVVGZyv3CaFJFA1zJFZ7p2wklC2gvunztCqGuVUU/0wbPfRF5vF3KOxOSp8xfIecCViZeHhRe/Fe6Wa/7dHz0/Xzo+frAEeeSvvthgVChJ2NiBzg2cVTmnyt3R8Dh7inhwG57MnnexcEUgWCDP3J96jmxR2zN0Dmcjp8vEkhaOc0QrbK3n643nixdfsnk4Yy9rqr0Yx2OeudSMqR6lsKjFG8H3FrKhxIqzhlgNzho6hIUNz/FL/u0f3vDdr36L4UfIYO0tydxwrDc8pxs+Xjquj0JvAtgrYt3ysGz4Zen51Dp+fxso0zOhRqwoxlUCnvdH4cN0IEjPtBROUThly9vZ8EW348VO8fVHsDOt8Dmh39bqmX0N9oyamdoSMZ/Y7DNfX/XYB8uxNAqJThKBwkYq1Vt+0j3/8a1hnl8z9srNpufqRcewnHicnj/34i0HD93QuO6fGG2BcgCN4E5Yk0nFc2pbuPQkeh6XHT/cveE//vmWn4/XTGFLCR3ihdA5NqJs4oIxjmgdnvWmeTBKT0HivIbFBmHcBJZUMKVH44TvG11VFIfxO+bqOOqGqW2JBQZ7YRj32LJAM2Dt2gaJG57LnsUeaGaA6hARLnPk5cs3XB6fqCnSxHE+b3n7ofLyH0FrFwZfVrKcf8uNfYGf8/qdcQvVwFK3PN5d8csvt5zKgYdzZKk7rNuiaaYzgeYNrU340ghacS1hRLhUx3GwLGIQZ3E2wH+lJxhPqw0tCsVSU4MBMJkcF5a4UI2h63tWx3f5XAVtiCqOgneGrlutgmIM3gpDrVgax+J5FkPnBqpkVHQ9VHjHMs+Idxhd5TmVxtE6RDNfb0dCtnxrLM5ZqI2E4HrLZbmnVfMZAWypbb0ECZC1cKmNsevhyrFcCk0szTmMWLS2NbMiBrGOucyogFjHconst3t2usHWxJNYkuvIVcE4jFZGU1FRdpueVNcXe4wNMwbQjpwStc4U/4Q9ZM5T4SSZWGea5PWWDtS60DnYmECn66j+mDsumDVL5sF0ntrqam00jk6UsXMMQDvBYmGxjiyebCyZxigDyVwj9sy1M+xT4X08Id2wTkM6x81e2OoFiRMjhRvXOA6Zf/PukZ+Wa9Sul1bNid4FbCp0tdLnFXzVTFunDuLprKGl/Jc9DGQMsTlO54Kzz1y5M7+6ecXD2zNT7fFktrax62DXd7QyoGw5F89zLhQcznY8PnxiaY1sO85JydHQjKXJyhIvYnnSkf/8mAghosVybgeO04b53nJeLDF2lDzg6bnyC399M/Ll1Qv2mwvknpgO3B97FjOAvGCaRy5nRZoyfDCIv/BPf1tQdyYYAXVr3kBldae3QC0DtfZUY7AsqM5UzRjrsLp2qpsRrOtIOYDxmFTZVSHNZ0LwvNltCKPQ2QdaeaL0mSX33LSR94zM1WBlxH+eDuSma4uiBqiW0Apf9hNffjmwZKXajNGJF9tnev8R6gMwQVNKtNQ6rrIQDKY6WvTMOXBZDN3tntA/IfXjSvMznsvyTKnHNbDpPLevD8z5jtPcU8uGhKy3obbawUxnmYoj6xWWBuKwOTOfEzEqZclIvKBjwxhPmx9oWrn75YF3f7bkYYP1N5TlLYMI2hrOOEQMXhthNOjO8Om84am85GHZ82G2PNSOY7b8ur9wmTOaoTc9zQldB9ZHxBu6MDI0h99ckavnfEy8SxF7Kfyz1wODRD6WxBMjPy2FpJkvDoEurhUs8MSYMM2DGdBqMMaR4gkvq61s4Yq/ffuS58sbdrsHSht5Pr7h/ccXvLvfU/xLinS8ikKqM23OXM6Z+zTybhn4m8eZzWHHUB8xWukkY7pVvPP0EPjbtyNbtQwpIRLZOMslOn65y7jdjtfBoaYi3tPaQizPWBnI/Z7N4QqboOgzU668fXxkyg905orrYU/ViCfTm0ZvFbMfePzF8P5xixn32E5w9shfvbiw48jd5chkVjBX7wv7ccHaO14PJywe07aYmklaWfTA+8df8ctPr/hw2fLj84F3p2tmuaVaQW3FBkeigPGrqlk7OuM/B+9mgiQ8iraEWlZRTTE8q+fubNiMB3ZXG8z8C0Nd2AC3dmSmx/KCmEaqcRzTRy4xse33lJQQddjm0OporSKUNctQK6IBbz0pxZVxQaGKJZsrfvro+O/+iQE34WyjUGntglmDUzRZbwupej48HLj/+IL5uFtfvs3S288vN2PBNIqbiXnBlkwvjS5HRDIqgWNTarcGC41ZJ6i1FjSe6Kj0TaAorTZiDZS2cJ4utJKJIpTeYweHyY3WEr11a2LKguqCsR3BeKoaRrXsrK6fTQmoH8if9921FVKpeGl43zHHmdCtXgERQyEzS6WZLUMYOAxmxc/PhRQMZyksza0mTevwRljS+l0z1mCNIeeFRcp6IPOGmiui0Nu1kFpLYugtpSYalqxKb92K4W3QiWeLrIRW3dBoiFRmMVS/umXqnKkixCScSsdyUs5+wK1ibGK58NQXsoFUoDUoziIGatG1UaFxzWmEnosWku2IKdENA+IMhYYagzWBVBJOLJbGGGAT1oPQsTQWUaIKYgOLDZxloYhw2Da+K4nQAtk1ohpsq5h2YRjP+AHGlNmmIzchIV8f+B/vHY9l5FMV5CicU+G667jVIyGf8boQdMaqYtUQROnlL4wjNmRK9ZTWkewLjqVifeLL6w1DVMpScK2wnGa2m8Q4OGw78WYzUDEMPtPSO3Q/U8Ww6IaPj5UpGZqrHM+Gcwl8iJ4HNrwvG/7TIzjnOC49xzRw0oG5dcCAFM/Ob3ieTyw/zvyDL2+xLpNKRyw9fnOg67a4tmNbA08uM2XLMt/yhz++5/e/folxM6VAaj0xCSUDdqTUgZItbhjouwsbfsAyo6ZRS0atoZmRJb3h6fwtT6cNh75jl2f8pRHaFm0wPT9xmRL7/YTWJ1QtSwF1AVPXz8uox7sdQiG3hVQS1nd4f420gc4kRD+y9SOzCpgznTwgXKhWQdzaYdcrHj9co21LHyzB7liml/z9jzvuLlvEn/nn//JXfPVVoi1/YLRHNrsNtYL5jBl2zvCbL3f8/LNjlg3PdcbYgSFCNY5h2xHx3J2Vm7GhTrBFeDx3LBSMgy4c+KXtsDkxGIseF374tOU5d3S7gLWG0VtMmWktrtxyMdS2KoT7zrP1gY32xHBgqJbb8Ir4eOF5euKSG6UWgl44LYmlWOgC00VYMiypropgKrcby9dXlquw8td/f2v58Bj5H//2z9xzQ5aR83TE9iM6bFAsWEuMDcjE6UyrymHsCAp96Ik5cJq/5vtfvuef/uMX5OI5zVuS3hJlYGoGbywxWz7cKTl5nGloFWqGS93wxw+F33dK3zJWK2oMTgaydhwXi+qCL2dMSni/oTfCYxLePylvvu5pbkZlQVqCdmaJMyZkmn5LJ54mex5PW75/CtwvC4NEbJ0YPHR2XYeJVWpeuHvrqO3A7d7wuy8Cvxsaf3W4h/aJzXBP8xdsFxm6itFHSj3h6KBuIGVcp4jp+HB6w9/+6Xf87Q/XfFy2nHSL9tdgBWsjyoK1lgZsnfJKDS8spBR5QhGzsHGReMksy0jvbpm98ImO7+eRx3yNvwT26czr3Stu2wcOJjGIo+BRRqpsqMZSzAt+eLjn1Xe31OUdWguSLbYVRrOyE2PNGHUUTZgWmKeJ/RAoyVOrYWkT57JHKrhwAcCKrNW6Nq3NEGlAxZo956cDJV4zuoFlqYTiGYyQrOCCoxFp0nBeCTXTUaEmTLf2w3sxPDTFGg/WkGujlYmezI0IGyKK8Ny2PM4biglUDfR+wVA5lsoihbkWtILxqxZX7Sr/iVopNIJxdE1wteIUNlbQamhmzbTQKlYFJ5bWoHMDpTZc8DSx0CpXPvBN8GxUUCM0NXRdYCOssi9nKQuE2ghG6MPAKQlzXPDBrTjgWinSVs+Ms0iNuN7QRkt1I4NpxKnwPFWy7xBY2wfKirPG4tuA0Q4vULjwQ8yU0nGJjTlVjBOMHTgtiWaUt7HhbcA3g9ZKLud1vWJWzoQRWaul0rDGrdC3GLmokgeHF/BL/OzRYSXhVqUPhpe7zSoHMkqfJzaaGQh0wZC00bLDM5Br5dJYzV55YTd0vPAdpyIEcZRLZNuth29pijPr4YA6c7CeW/+Sd9PAY+qwMiAYwvJEHwzSlFYrrjW2WPrW6Pkccv3LHgYKrSSWOvDz2dF0wy5ErsNCbwLZOmgHxBww7v5zeKahLnLoHWN9pJV7mi5g4dx62mDRocd1hXa9pxTLh2PgPzzDD2nLT5ee0a0j82BGvPYk7YEB5zyxgLDlLr/mP/ySya7j53vLnz9uyeEGeWp8GRq3JbNnYcmgNvB8HpimLS4cOMdb/u57z/1xwyIHqhvZb3e83hW8nXnVv2fPO+znHbZoQV3HvPyK//yH3/HTx5dEet5cF/7loSFzhhpXyIUKR5OxO09wW6J2fMqWD+eFp/zIzXDB1hPIgHXgq+CykEqkmYmd363cbe+AQDyfiRyZXMY0h6rF2IzvN2h9xePT1zyce85RiWlgmjZI9zVZRtpl5t/8T3/P/+Vwy2B6apsQnSlFEaDvOsRmuu2Jb78euJyeiEtkRkjOcrQF1UJdei5p5IvXSseJh6eFR/OS9zHRdQP6rIQKu2HDV97RXTLnSZjcCn3auYq0xvHUg3/NUgaenxoqV4j7jnd3X/HHnzbM7pqT9kza8XjMnJLBFEtWYdzAm6CYKXLKlZ+flMcyUO2O1DbU6jiYE//wxYWw3K28cA+jM3x9teH/+k+/4tvjlj8+LByXC+flzLFs8WFHi4aSC1e7wOvNnuN5bSOIEahKTpD9G/6nP7ziy28WrD8ThoGt8Xx7OPAwNZ7uTwR/xbEa1BqEM5SM0y2lBD4+ZX71ZWMYZlxdD4ctRdCRBc9glSllXG1YWfA+0GuhJKgsqD2DmbAyY5cTxSpRC4+LwdjfMV12/Hi84j7tOSH0LjHWiikFkwunWdh3hftTY148o4fff9Xxm6uFN35iDI04Q6eRlJ6oJDQEQuixpUFplHnCtgRSsGHH5fyCH+9uuJv3OLPn2jtiq8TmqKqEbSCWhT4Yvhs9vy2WqzSTzMQnP3G0T4wh8GnZ8stpQ9KRi3ju24Yne8MiPa4FLufAcV54+etXeHm7yltM4Ngsj6ZQm0Flw+M08zRZenpabCvwTMAN0OYToiuWuBoPXoi1UM2wQnYwZB+QcIC26n7FgtaCMQVaXaczxJX8qIqWQNcNK/2vKjEbZi80D7JYtA40UUQyViOQVlOi6Uiho4ogWmlqaVppupr0Ao1r3xPyiSkrRjecc89kA70fsHpEZCF55T5dWFJa4ULOc4mrWCdIx6KOalauv7LmEIJp3HYOKaASUAVpic5YRCsihs5+DlKvPVQ6Y/jtoeeVV1IsTNWRnGPjlYMzuNIRlw5fEoMzdJ0hmYZXpUigaxCscq6R2TjENEZJvNwOXPvGjOWnpeAp7DNc9xvugXOFUyt8TJGNU5z3SBGqNkpJRDLHHLnEC3NJJIWyZLwBT0NqplrhXBrBGmwrqwNALaMfICeKOrCy8jKMYI2QreVcG+8vM19uR377xSseL4n9eKArZdUi+55NK9zYji9KpI8nxhrZ+h0nf+GYC6l0GBlXP4NZa7C1NKzyWbTXqGnGU9kGZRNA58LAgtHLyqpJjflkWPIOGPHO49SiTbjoiRKeMLFBXRHLLlREGspfeDJQU8ZohzWWp+p4ngJXfeKwSzDNxGXdnUzzhhfXh/UBKJXeG0abkLjgXUdOK3mspMSUDzwvW5aHgHWWYBr7seO3csXzY+BDWW+Bh+ARcatXu1qWuIpUrHFIsyS54u8/PZHqyMe446wHlsXhAFciA5mSGrVZkvWUeuDu6ZrUnnl63iLyW1zYsEyOhyUQNfPF9RPGPFPrI7llCgljBWdXffK7ty/5+YevOHONhsD94yOyU4xAUUfRTFJldj2uNN7sL6CfOEjGhiPfDjOv9pUhWHTZoNrWQ4Tq50qkW9cSrgd1GLsqQk+lItphVbCmB5lIeSTOB2LbUGXHc+z58NRRZbuayUwj2C3z6QU//XLht1/2aAsrnbFZUlaW6gg+sDt0uO6Obf7IEATdDSwXg98o9tnjWiD4Ixt9oMvHlXudHpFyxf28xajnyyHz1XDhy3Ykl8LANY8ifKzKcMn8/LTjXfvfUTgx68gl7piWDVPcM6UbCq9obkcp6+ooamDGsUegzIx+xg0D2+2G+6g8p8JsOp6T0ggY6XhxrWy6zHIRpqjY4On9QN88vkR+LfDixhKr44LjfRJO5UICwqajlg84jXyaLHN2n//ehFQypVr+fPkt//5nzxfX78kzTNOZZBT6LU4KfW/JXeM0zRxzWjkaANUQzci7ZGm2ce2UWpVj9DzlHb/8pLw4vObl+JrezsxP91zfOLZ9T7YNtYq6iDUJrQknlWYLahrnPPDTuwOfnl/wVK95psOESi8XbtoaIIwl87ffR+I3W/58Z0it48or+nxHuF4T+8U1xHts9WiVdWXiesQajBo0n6FlWl676Ov5fouaDcZs6f0AzOvkS3ZM2VFjZdw7tj7ya1v5MhXG+kjmkaFrPI6W96eev/mT5dQOVOmQrsOPVww6YnStFaa2Y4lbztPCeNPhcsJiucxKqYqzFmxg4YrvHx554W7wDUAo3ZaFDklKMIIFkiqxNExTzst644+tITYxBEDb51tyQlh5H80oivm8VnDUbAjdhiVBCFAnKNbTCMytkWxDPgttqJFaC0kN0Q5E2bGopRq4Mpm97VCpnEmMNnBrLTfWgvVcSEwIs3FUJ2S9sB0ac7nwmCrPy4IYi9HGKc4rsz/BvhvX1QPryzMZofkebxO3GHwVLmoIxuPEY6VhxZFrQ9Ww2e5IajmnjAO+3B3ociJ3lWyEaitXY8cr77g/Qnh2jN4z9oJIJteGM5Zd6OmbYmrGqMVZi+sMoUz4fGaUBZt7QuxopuKbMvq11VANnHVdpUZdjZG9aWzSevHaFKXNE30P1TSmqaxZA4FBlI31zLWyiCHVihOwxuDEcLCVg++5iwq+R52llIQYD9ZSvOFDEuIp8pvbA2+uD+zNyKva6HLjNgS6anHzET/dY3Rh64RrB9FWHmkYq5SSCGrQopQiODWQC+IKXjpysSuh0Ta8LrgAoU5IEJa044e7K96fNhi/WWFUztIK0CyPactLf82uS5TFIip4acgK/f3LHgY0rnjbvmtEVzChY6kzSMG0SGvCUnve3Xl+++0XvLj6RNc1OruQjkdwDqkO360jFqkD0X/HH9/veYgB5+DN3vFV3wNKy5kxHBi0YVjVlNfece0sx+cLJTqW0pG1A+k45QNTgUZH055W3LrfavAhZSwbZmtJptCK5/3pNZ+WRtHXJF6R/YDdeJgcjrf85uYO0/6I4w6jT8BKnDOmEcwbpueR1tZkbmqQcWhVcr6QGZjV8Vzg2YNrFw7bXzDt73i9P0KYocu0Jkj6mlh+RUqCskHsiDGBrNd8unc8mJH9OLAZB/zYM9ZI5pHaLrRaWJLhdO7R2uM2htsRnma4RIe4HqeOapWZCmx5+7Hnt1+MWOlp+rk1YWCpnilb1Bau3UKTZ5r5DP3YGa62Z74c3vFqSfz+m8LW/ghtZvty5Ku25TS/5I8/fE+0N3yz3fJ16fHzBSeG651irfAglk9pS0m/5v/3/Q3NFIoKxm+hDYiMGL/HmB2ow7qy+uhNIEjDL0owa7L6MkU+2oFHOmT/ivykZFGq9NAsd6fC/zpnim5YmmIYcdnwVbdjfz6j+Yj1Ht8qG7V83e04mkY2DdcZDts95/NCvcB5WRh72I4jpVqOp4aj5/FRuB2EXBvYhpGFFBVpQi4nWrsQ7EipAfUGh0Jc+JgO3D3/Cvvc8d2V8nK74/v6ip/rLU/s+btPStDIy9Ghl2de5A/8q26gHk/AKudBwDSLw6OtrHvSsuVpvuVTfMlj3XAWCHbG9sqQPuIm4VwCP/3S8b/cNVq5Re2WLYbl45k/tzPDP/BsdAGZKBTUeqIWfFtvmYYKTTFisX4EFSSNbObGVxvL/Z0BY2gIplsDoqfnDcs5sCkLX71pXOdn/OWIbWequeC6He8+7fh//KcdT/EL9vtbglHEC5cacdIT3ECrCaxSzI7SLggeJ5nWQGsliEcaWB+YbeAu7VCBwfZMtbEs9bOsuCIBSODZgoUhZF70De/aitaVmX/85T07f4IaMS4htqIoWsqatLcWaSNl7tcaX+cQAWxHKxCkYr0lm0SQzJXPDCbhciOXwMeZlX5HYx8yu23mRV/wKXA8NUZRDoDOJ1JL9L1lmoSJQocBEik+kJh4XgrNeIJZ+QRFM9aFtT5YGi+2G1LRNVCnSus22Go5NMO1tZyy0oxHJFBboYh+znI5UpPPkjrHwcHr2xe48xnRTCGCKey6TC8rkvuSLBo8hUTTTCTjOkctiVrBmobvLS70qAG0IrWgrayVcOnA2DUcp4lRKoeuY6mZKoazyFqFNBnfWRyCtIw3KySr63qWZSLVSi/CQeBV54l1R4kzs6uIKZgm9CLsjNI5IbfAWdeq7BgGMAa7OXB3nnmqC885Mn/4yO9fv+FAxqlho8JYF2xZZVetzgiJzgZGH7nptvxwznjtoSZsE7zxPD0904uSjCOlTFQDaUXKS60YOaIlg7Mc6w1/fH/NH+5f0ewtIw6n0LcCsvojprblXq7WZ6rfYirYVvHBYsxfeDKg0mi1UOMF4xxqhZwMqS4MAWwxaAs8X3b8+C5yuzf03UxaTiRVlA5jLDYIWoXGS356u+WXxwNpHPEkrlSJxXNKYMVjo5K0cZKKWEWXxPV44l9/seF6fM2fHjv+9JB4PF4ojBxjokpD05FRE6/8Ey/lka6Dh7ilsqFKpN9kPpxHbP9rLrNngf/NX1418NsN3Lg/o+lvaZporAlU51Z/wrxsqXmPtQFRRTCIQmuZVCqzhdgPTKmRzMTt1ZHRvUXLO6qcVyRsARGLuoXcziy1Q/yOXA2xbrh7uOH92w2zBl68HHhzI3zzRWNvK1OqlOa5LImUlaI7GhvG3rB1M9++PPCnt1siPZjP3WsjiHF8evKkJTAEizNCw2KboVWLCSMxJSqX9SZjIGqmMKP2A69fCr21bMYF0tOaaJWn9XTd/z3//PdbTHhNvn9FfPcPaeeBjsqvujP/rFn+bXa8O3lwIyIvsF7QWlcHupH/P2t/smtJlqVpYt/anTSnuY32pmbuZu4eTUZGoRJMAslMEiwggZzxLfhcHPMNOCAIFFAECLBYlRnZRESG925matrd7nQisttVA1HWuAY+0JkCeqHnXNlb1vr/70Nao7ZE0xO7YWRgxVLHulBLYmMTuwA33Ya59dw9CH+cIyedmEpgzp5UMrZm7k0lbnbgK9ka6gJe6zouNz1LyxQ1qNuspDA7UHPG9oZuWHeULU/8/KsNfb/FyUyOR06zUshoS1zve673goqSUuN4zsyHRs0dp6Ic1XI3e46ph1i5vtqyCcJ/eaj86dMVllteL5Hvnm/58LHjkRcUtyNWyyVVHp8WBhkwZkFrx/lx4XQceX7d09p5vVQr9H6t8kndUesVKjvEjYAyNTimSLCNEBdKmzg/7riTl6ThmtEUnHlEu8CHe4d7H9n+MuH0I+rOqC5oE6waWj5jNGFspuX8hffuIfbsJsMvu8j73Zk7V1AxnOaOu0PgPF9ju47z0z1lm3FhRuPj2sEfhN9+7vl//N2GR37JdnuFt8ogM1lYL4OSCQaKrsBjtT1PJ8/TfkeH41gHSjM4Y2klkfI9wUFv1glAKfNqd7SZvBzoyIQ+YDe3nM6Jaiw/ex75Vz+vjP6AMuPshb+4fY9r7yhmtUNqXS8DooIYUBOweosprxjpcXbBBIvKKpQZ+y1qK9lUXt8Ir90ZHw+IOs6lsc0eNcq1L+z6iSb3dD5zO1wh0fP08cL52EgFYiqY4Bi2nqktGGNpaSFrptmGs4Hc1oHwlxkEaF0lRMZgja6X0dpQw3rQq6evmV85R+/3ZBVU2zrhFV1XDDYQ6KlqyVrpXaH/4r+wTfFO8UGQGsEKS/I022HsmkspGkm6hpDbvOBchzMWdR1Tg9waewlUVyhGqW3VcmcqViOdGGrNLA28WeVGd6znwhON5B2pVTKRrYNeCzGu2RCvX1or0rjCUDWwVMejRGaJWC/4slIDA8rOC4s4Zs30wOj8urKJGbRxUeFDXhie7nn13OGaI2CpzaIkINOkIYCKIjIT6olRrnnMlWAMozQ2JlHmxMVcKFtLkJ5lEqp25Gz4dBTGrqfmgfNl5N37DZ8e9zT/gsE4QquIKLYUnGkUaWQMn/U5x9pjRBj8KlizogT7Z64WGgrSoKUFxpFcFcVykT1NZ6amTOqoesM//OnCzfWWm5q4ux84zhvm2CPVsPUb3r4ceP808v7zFsKe0A3Qjry+7nkpjXqZ+DpM+CvLkuAUDY+5MpXCNkfe9MLIEz/JltYyzhZiVg6Pie1g+NkV/He/nPil/x1yek8uno9Lx2MxbPcjkix3J+GYHPtxsyp0cXzKiQnDLlRcvZDLQmuZSsE4C1Sabvhw95o/Hq54dDuW1tFMwOiFVECK5dSE67eer82BX26e+Pr2HcKFZgONQMnrL6Rxbr2hmidy2lA0IPKGeXrG43vP4dSx2A1PH4THqdAPmdfXDZscLQ90bcdEYeg7WhZMfaLrMl+9GHmxueGHEyymUoFYEl0Q1F7zeN/RvQhriAyLqqELHVktxvQIyxfBUV7HipqocsFQKMYiWJpYai4UBWMqLTfEL1hOhBcjTx8qrV0T/Jmvt4l/4xYOqiznhVxXc1cRMCbTtWW91Xcjl3gklcTGCs8HjwkbOGb8tudbM+PSBU1PXMnAL7oNHZ7JdpwbfD6cyK6x9YVgM9H1fDwmslaMEcbOM50PzIsS7RWzDmTxtGJIOYI1uE4ZtRFiXsVDptKWiVpOOG0sDU4iGKkcFqVJz1IWPh23/P6nwKeTYI3F3r7mp6eZH+53nO0WJbFPCzc3noWE6W5I9PzERD5blBHrR7ztEW2k4Lk0oeZIGJWtjdSovP+98ou3V2A+QVlwZh0FOruhpR0l9eBHjPRQICLcpYXqOryfCXOmzBmCw5XVQthaxrkCOB4OjUtr9PaC1AmsICJ4k0BXvLcyYXtZKYtxA6c9PsEre+BvXxg+AIfm+Y/HLe8PHd14syqUFe6ejixXQnAVE5Rlc8t//LsrTvod3e4FwTeCjXitFAziBG+Vmi5gLVYs1sIhd/xuuUIzxOJIy4KVI7ejBRZEGtc9fP3SY9sZ7+54eXUhyBHTErFueVoiWd5ipOPN9om/vvoNkn9PkUhxsAmGVOwXHseaMSitrZkBgaqenPakaUsvyrV7RAI8Ds+4OygtNtRUjM3cDJYrXfA5IjXR2YqIx9jKXk6Y9ojbzWw2Fc8DRSz6zJPZcshbDq7jEYuEQFdW90CSShJHpay686a0BlUrXsAZQzWeSzZ8Ps30tkdV8CIMQfC5sZfKz5mQtuPAhlk8sXU0rXR9xyiCKzPYDu17Qoqk+IhXJYSeXgLOGKwUllKZo+Ak4IOn7wyUTLKNGgtS2/rZGE8zwuQcWRVrGqOxXPCUklFJmJJQLWAcg1svFZemJBEetDLX1bFBTevEVhQpq9dFc2XEkl1ldJmuNrxCjmdM6xh9j3EebROmRpyzBDH0Rsm5cmmWgpDbin3ufEcRR9VVrtckY4hYMWiDGAvGJrIumGDWamqtVCK9numXnsqAOsFrpmsHaj5hOkMzyukyU6sno4jt+TQ54mT4dCfcTSM1b9gOGwYVfJtYjJDNQGkVbQnfdxwnJef9einNZ7xzdNS1Sur+7JeBdc+zxBkrCecMqRmeZssyGVKGiCOqI057/v5PB57PW+7PV1zajlMNNHWY08Jvj9DmHdFucH4gNQPVUE8n+pD5xig3PSR5YAobfoyGp9KwKK/2A68Hy/z4ifHyyHfeocYStOH8FUUcUj+yb9+zPf0dcplpfmBzrXRXlR5PPnzF5/g1n81ALBe20kAnnuzApzqwa0daVVpzKG2VPtmKDR2pfMNvvv85fzi+5OT3FDOSY+NZNRwXpcuGGXg7nnlz+wHvP2P5RC0K5mqFI1lDrmn9YrXTCiRi4OOdcroI54fCdChUzeANsWz5kC2fbzO3/UKLjrb0YPZYm9lvjgxyYJBE8Eo+P/Dds2fY7WYVoZQLc17Nibe953S4QfREt7PYUCklUdJ66XHBYIYtrm4prWFMxlRLMx51sjIVmiCuQ0pGG2SFpGCKUqzizYXu9UrrEQIsDzwX4V/4C9/+5Z56EWjK4XJHzQd2vsP1QrWfCDvl6triZEFEUUbOKVC1wzzdo1VZiIirvBqVnetINWJGQb5a8OGJ3TYSc+b3P01MVx273YacC+dzZDAjMe35Lx8jx/yWRQQRR9RCKsoQOtLhka+2N8Qpcv/wgdobjF+DnOf5jN/22Oo56cB//uman+4yP95tuas35G5lnH/+vme69zj/DPqBWBO5nDBF2A4bSnEcc15rZ9LwfUFKRVujWYPre2ZtpGXixRDZ2YWZwtMnS5oc233B2Iq0AsYi7Kl1T6xuXZc0oTVHLJVj68ndS1wo6DSx+Gsu1q/41ybshi3eTPhSqZPh+FTwe4sTB2px1uCkYUqHqqWgzFWxcotd9rjynFI8tpx4a2C/2fOH1HE5NiQ8w4URZyylVT4fB36aesz+BY9F+KcfB/54foUfX+F9wJkJY+taXTQecLSS6FyjlIbBgVkoOfLh4cJ1mNjZC3/zjefVduZ6TIQg2G7A2YnRv8ebj4zhA858QswMxrG0rzhHy2numKYTz4czgzxROK9gMRtoYmh4nJf1u9xWD0OTSlXHFG+5PN6Spo7Rz9j8ESHybHB43XNMI8VaavJ8/Djz8k1PqyfMCv8jqCWViHEOZzqsXghkWqmglm7sGUpgmQc+XTY8GENUSxcMJS6gUGxHAmoVNn7l4Hc0brSwEeEhKZ+jZfY7ql2R5qErjEOlfckV7IfAm65jNBs+nyreVK7E4S283o3c+pGyLJw0c2mGppFSHsmz4oPHab8yCC4NEU9vDYIidSE4h23r1KKrFtsq1mYGazBzQyTQgpKD4z4VmjTO9khoMDQDbV2JaIXXHt7HGbUW49agr//C3hdjaFUxWDp1jA2yc9hm8WLozJ5LLWQfSHhygaI93meSLXhTsUBLhqrCuVQmiYxOsS6gOeGNrhME2+haJKW1cp9aBFdRD9oNxKhceotzbk3055kmW6KLIAmbG2J7jqbjiRvu81q7d84SdFjPgAkuZSTZAC6Qq7ChMXzJS0RTwHYsKaO5rhX3pa2gPbtF6+qbqLSVA/HnvAyot2i2VFWMZozYlaPdALPiWVEF64h15Dj37IqgbkDaymNfb/oDp1bADNB1WO9AVtWnVSFIxsoa6pliJmKQtmP0q5J0bxY2OjHYzD978ZqXNRCXRMuFY3LcqSFp4E8fK7+8GfFiSQLoiRLPVLOj5YLx4BVa7vByxpYzz7igusWlGRGL8X4dtZm2Cl245ocfXvHTp9dE94LsOprpVzFQ6TDdDpMqvthVVeEPtPy0GsZKQCVjwjq+DaZHyplSE+ITzlcyA5/PgbIKqehywmok5YElC9+/y7zartrnMl3xdBkwV42vnp1x7XFVcdaOYBpf3SrPQmYYD4zhyMMx8eG+ci0bZB44Ls+pj5XhtuG7iLaZZhxpSrza9Di7x4rD2Zli1/FjaQvOrw8DpYEXnBhybqizxKosS0PPT1Desx+u0ckifsDZEXtY62k7TbzgyE3/SO6UqX/J1BmG24n97QkX1gbC/dEwlUDOFuev2F33lM3A5ydhPjmeToGlBIJt7Erm+Tjz/PojzvyJYp64+XZGiVhpgMfYWyReoekrvhp/xf/n05Gj2VG0kree5ge22yuGOdIfP7PjxH7fka63HFr6ot72zOmAcZ5j7fm73wifzx3iNrhhR6oZbxIJQ3gWiNPM0A2knFbsclJMKXSm0NmCMQ3vBG2FJh2tWpr2XJYLkYx3lVs/o9NnDDvSxTOfDFdXBecyNKh6xXR5xTl1mK7RSSMtCyyZQRq2XajVYqwn+x1P4ZqT3GLU0SjcDgbnKrJMUD33d1uebzqknRFt6yqJASHTWiJFxw/nr/ntT6+w7YafbfYMdsK1GavnVfaSn+PtgPiB2ipBKkM3cL5s+Nhe4XB8PAq/m25I25d0xmNFUfNlRGwA46FYnFVurzxba5FlwlJw1tO1yr/61Yavxwd25k+IzkxqiWZPNUd+/vKekb+ntk+kdiClea2OmYAPe+wckcvC3llGmUAq4nuMzVjnKKkCjpIirQGmp1RDMSNzvuLh8SVcXrAJDuMmSklQZ/pwYjNWHs6OuXrUOH66bzwfM8/E4+qFJo5qoJlAosP3A8acgXk17YnHWsv1bYc+NPKpW8ftomg80Utk14GannP19AK3XeC5hTE9cmszrst8apbfn684xw2pWSqeyV3Qm4DakWYyWMfNs56N3xC8Qg30DUouvNht2ZSIHS2fzssa2lYhTwckT5hkoO2Yil9ZLqYR6h2tGpZcOFbHIoLJhU5GjBiSJlQNoTbUNqjKlBu1KOI8x7w6b7yDrsmaw6DgbGN71bG0ntI6DI0lTXgbSEAUzxwHUlpDh6oCeWE7gN08Iy4QW2PCUEwjUwjB8KHOdPmEpTEYS2yFai1VhFOccGldy3jT2FTlBY1RItnA5DqqdSTneWqeu6eFEK4YO3jOgs0T+9DocmJxE0kyKmsG46n1/OHsqG5LEIttjZ2zXCaLxhuccdwOgcdpphi3rlBJOOMJ3lKwtBrIZVlXL8auU85iQASPIViQJf15LwM5KCU3ilFKnikSsOIRMasVyTpa1ZXwpp6qGyruS5hGaFWhFUTbOg5xBlVD0YrKanfKpVGKwVjouz0fL5YfH4RFPOoHXFGGesbXhIpSykRMlXkuaK2sMNFC6IX71FPtczouUCOikTmtHG4T9jxdAo/qVtSsrB+CrQXTJqRGVAVBcVZoJlAZODy95qdPL8BuGOxqoUoUVCqOwtivlEWnq4Qjt4FSekpWag3kEjG9sNsLVjMiSm0FlYQ3M91WuLxbH0C+A+8U1YaWhda2/Pih4+uXz+nriXgPjzEg08I3bx3WtBUMogHFcftsQ1VLMAeCe4DRcToUbBWCOky5oVSzBu/GBT9ESjXU7Ln/3LHv3jCnmdk80d2MeHOP0Qt9WFY5Sk0Yo1ijmL6ntZVchuuY00Iu75hnw1vzLZ32KML2urDxCy/MhRdmwZ0Sp2nk/aEyvzBsfWRZDpQClwmmtuXYdnwsN9wdArvq+T0Wn3vGMGK9YVkKx8fK8qGx++z5l//NmW9f/h7vHgh6IaUJRDC2p0WhnBZGa/jrFy+R8S/4h8MVPz7NnBYhxUZvFsxygvlMiRfOTXnIFXu1ok5FFOOUWC5EGTjGLfRbQDBG2ThDsMp2VKorTDkzq0JQSp3IRjDuS03NCdZB0YzYAehpGtDq10mMAWMW9r5RzzOaOpSBzw83PHv9AqyjtJGnw7e8++lrzjEQ+oXn9oGf7wecAZFCPD5xPV2Q+siwveEQFx6wdDIgemC/ifxsWHXGP508Hz6NfPPmGRu5w6pCNeSc8VJoBKK+4d//9pf8zz99RXUjv7hVfrWbeTOcCfEzy7lyeYqMPUisBD+wHSwpL6gdeVg2bIeZR/Wc5YrQB1pbBUNJBSdC7xpiMuRKFzY8nSKxXvirZ8reHbjaCJ0qX3c/8ZX7CVmOqOkojFxkh7NHNuYT2o6kelmFRrLa5sQ1nBnJy5aiPWoWjB5IOmO7wDrParSc1lGs9VR1pLrn8XLDh6cdD+eeq+1rvumFbv6MsFCtBxvINeI3hjLBXCo0oegV//DDwt9+fcV1VzGpEPKEEEjiWE7wfLzF+og1FRgorEz5n90q1iQeSlwPSLPQW8VaR66G94+Zy2HmmSSeh8hoZnw+oAKvrrZcqFjbc5hXONk5XXi8yWxHTymCeoe4RmtnRqkMwSFUls5ipBEQpCk742lY0pKxpkN0xqVK0jPJ7plbQKUwdMJUHOezIboBo4mdsdixEW1lpqBj4PFyJpsOC/isOC14VZo0moETSt/3a7ZJV3Io/S2X08hDdBQaSIdtK7dBEB5ny5wtxne00riSgX1fuerhBxYuapl1PXuMWJw0ihguzVNzZNN17M2w+hlMxVizwtFYfRq7WHlmWH0SY+CT9txNnuOlckwVY3fsxTI6MHHB5MSoln2YWILlTOQUlBYM80lxXrEmY6lAo0rlIo7eOoIqWha2vpG1YIKgqQAVsQak4ZqsF+e0MNodWRtNHGrWGoFTpf/fNhj4334Z8MOBvbni/hI5VbiYwo3bkvPam4yqVGdJVSmq5GxIl0wqCzVbQlmrIE3hsjSQDG5HyT2LeqzqWg8atjhpzL7nXXJ8ygHMQFXBKUhp1CIY71jSRCwLsRpKtag58nIQ1F4QKtGODHY9DHKpqBuIOEqrLGaVJVWTeLV3DE2QbkO8gNh7SlKq6VDZkeuWw2nP/d1bcr3lpl8YueOoA59aR/SWviS6EtEYkWx4/5PwpNcMQ8DJTK2Qs5DTI8b9xLabVlQ3GVpD7MK+S4x5nax4DxtbydlxKAnNiVMe+Pf/tfCrVxtM+fLW/BT44x+f+O5nW5xOqCqqkdDNuHaF0x2lPK5jyX6LzQNuuuCXNQ2b0sDhDPXFDhMqWguHz8qpjBynHYclcPXNxM9+FQjuJ4JNgEXU0jSv6k5jGToPulAE3HbBX31GbhzxoxAurxl85Nv+PV0fCWVAnpR6vhD6LbebK8qrJ3726oGU7vHbK6QFDk8HPj5l3n8sfLr8jA9ux3PreR0M226kq0fEZpYwkk3P4TLwh3ef+cXXG7w6ajRYAtrWWFWm59Keo21DlMb7S+IP98p92nFMFuMrWyuY+YIgyLCj2pHHaWY3Vl5djcxxIjRLM0otlpLXcb1hFZpITUiLzJdCilBUWMqZZoU+GOaacG59I1JdhSQSAnlZaEzriE8qpVpSyww24lrCF/B5xnSv+I+/+wWnIDy7nTg8jdzdP+cw3TCLwQ+RrzaVb4cDGxqlwaNcCOcHNn1m0yX+Jll+d7cwOc/LvvLPXmf+Qu6JaeI4b/g0b/jH9y/55eszHXdotdS4oQu3HJ48//SnN/yXj79k3vwc4yxHc+If7n7P8nLkimtKKphu5HobGecTYnpyLTQiaoVD7FnMLYuuVVAxBmMg6bL+nqhye7uj0HFKJ5ayUBFUD7y98bzqMn0nLEtib8HPHa285Ewj7J/RlR6f3mNkIUvA2C2tFtSs31e8o7WOnAPqBXERkTOpXv7XuqAVh1ZDajvOZcfTNPD+k+fT8ZqHuCMB/2J/jecDohFFaTagVqh2QzSQLNSkOIXaDA/lGX//3vIXr4SX/Uw4RUIqJBHuzoZsO57f3qLlHrENmjI0y6/Gidd64ewNx9TI6ihqEdPw3Yin5y5mrobK6AQfC+5LALA1QyzC6TEBAZpl1oEfjidutxFjV5KqxIXLfKIrPWKE5labYW4rIl1KYlCl1Mr5GPHGsLcBlgmtC3noOdoN0+LJOTBns34fqQwt02ui2EihsjjPpWTOgGjDmxXreyUFVwvBemrNax3PrX6HRQPNjvzpzvNh6knWYTtls9vSBWhpYhdPvNEZdjv+8HjhVCq3o+cvesvV+Ym3eO7xFOtw3rMjc52PFK0cxSJdj1DWTFQIXErEOoNTZQRurGUrys5YWlY+Jcu7vOf9/YLvO4IzhLbQ5Zmdc/iakLoyPl4MI+/bhSdpVIXnXYeWyoYL2IKIZ8krmvlSI8b2+Oow1bNxlqeSSAL90JASV2Syc9QGThWPUtNC57dcsmHRRv8lk2Iof97LwLdjpbRP2J3nP1dHrZ45z3QCsSm5CEs1zBhEz/zqeeR//1VcfxBxhL4nlURMjrvP6xf0fnriToXiR4bQ8ZXL7JjQEvmwdDzejwReYG0m14wjsTGFIDPBCF/bwlAjs+tg07HfKsE84sKFc1mQwUIRvAk42aF9phBYkuVYlUmVTaf0rqxK0G1gZxKaEtPRMrlnFH/F4XLFYbml5pcMy8Re7xD/QO2u+eTf8rtLxC2RlhJ2yRAbj2XHu2mP3XVstz2bsaxTEBrjckHEoGqR9TFHFcPolJt+5Jg6Qot0tjK4xq1RKo2nFvj+vKd1C1vvWHQkLp7Hv3/J46z88usnNmFC04zW9/gQMKVh8ViB0AeC3+CePuPSjMFTCvTdyHRQNs8FT8YsQp0NeQ6cjy+4O0/4MfDN10+U/AjN0KrHhBWQQgW0EByIrnWiblTsJnA4d6R5xDXLoA9cnk5cDm/Y1q+oF8PxNFOv7vjuq4Gum+k3FfRAWZSX28rL3ZZvXr7k//3bwj9+OvD1eMveG84p0YzDD3sCFm0WYUOaBkrxmNrQ3CHV4LqCklmqMMUrYnnO+8sN/6//WvgxW+g8wStvrzzP5cJUTxQMUxHOKaENBmfpjK4Ut6HnlM6UohjnyK2y3cBXV1Dned3bDh2POjMvE5WA8Z6G0kJjqZEssspqrCM1RVyPSEU0r3/E0qSuClMHUhIuV5prfHz6mh//vudnP7MsKXD/UHDbESOFnWncdoldfsKnRqkBOTTsHGg50dojb8M1P+8tf5gzgYXr/kSfjnSucDVuEd3y68+Osy4MHlR60uEZl6Ph6XHgmH7GMbylyh7qQpNA7F/y+/nC7bgj5SPIiA2W7SYz1fPK99BGE8u5BA7zgsaJFhMT10S7anOtONR0TLlStOC7Hm2GmBVjO7b+zEaPtAWqRELz6KljybpOGuKAv9rSd6tERpuiCt73VLEgGet3XI4dpXpMUFZB8MopyDnRd4Gmntzg8+k1v/9ww8dDx6QbCDcUZ2mtUItAiFSdESzqepoZWJYNc2vgMl2z+AKlGWYZ+OlgOC2Ff/2rLV8NR2xNTKr4Frj/ZJm/69mOA7VBzQpPE8Nc8RREDclvicvIkoTQWyQuuAbPbwesLCtdUgXjHEUCd2fP9+8gRqHv18urho6fLpHHKogUfNfRpUJnhY62/tvNkDXR3KrLdSi2XqBmPqtn8RsGnwntgmmFahwfLh2neo1KR9BEsDMiBbzSKpAvBCdYbTjjkeIRY7BiGLynW54YvrgU3OBIrdCqp8yBQ6ukZFlSR28dm96CLytlsYBYS6eVn1NWv8S2491y4c1QuSoXzOXCM7/j594RTU+pytdOeOsMc7E8dQLB8jBPTG1BnIB4Si0MNHYiXLXMTe8IWih14C5u+bQYcFu8Nrpa2BnPnpmuRoypaIBeYcgTnevIxjFL5bHM9J0n5IyQaNIRvEdRkjFrtTMm+hqwNGznmFvGM2PDjFrIWFIrBNOx8x1lzsSSyBqowZOMWXEA1v95LwN+Ed48F970PR9/XREyGx+59ZFrn+mkIMZyWhLbK8P/8dvIZv7MlAQ3bEjniaKVJQ1sNzeM1XK9sfzCdRTbaOnEVmfeXoPWxvxg+OX1QMwLN4NDVXE+8ddvDK5eiNNEbyxv9hbjBqR3IBesOTKEBx5yw8iOrJmsFxj8qgHVLafYcc7KpJVglKcmGGns4hkjmeaUc/Rkrkg6gHuJ8TcEs2EwJ7Z2wYfIRR+5LNC3G9TARSumzlSz6kuN6cjZcsng5AIur5z/bseUIrlmWsl0ncf7kVwDvfcsVDyrmSp0yi437ouSTMdZtvz6cKIvSs1QGmh9xW8/Wf7q255/8ZdnHA2jjt6feLVTvDi0VIZB4VjonMfKutaJZSE3iPT4TeOqE6RU6mWBJFh25GnD7/6p8PJ6xAwCzVGKpcaGC5a+s1DPoEBb2wlpboSwQFeZ2sxGHa5F7CJ8uBceVBiTJebC3Xxm+fvKt391zdUNSJ0odWE6LdgWseU3/Ovn8C9vfkY9fuQ+Wf6f9wbCC67d+uBxEjFaiEsjnZW+W+Um4g3Ighqhky3OvyIMf01JL+l3V9y0G3JrdPbCq7CwD4U6WqzbsG2ezVm5pJl0PnNsQtJCLoXzeWZbYexG5loYQ+N6HDhVJcbGdLkwi4FhJVda19BWWUpGTCO2L3Y6aVQKYldBi5YMmgmuw6mn2oHFehZrMJsNGE9m5P54zdUJ+nEPXeXSCmOf8PbA1mRkWcjVMpdGmzMlN0QN8wK5Jq5NY9ca9XTEmYRSsKFxe2u5qcpl7jksL5nomeh4dwoc5x2tbhjDi3XMHgtVF+Zc18pt6jmWgmk7UlKKSWy6B4iPnC+NWnu8NAYmuvyZN1eR173wD5+f+FN6zdnu6dTROcthyiSpKD1xqevqxG348Hhms4e+M/QIIYE5pzWr48MaZLbCdujX5L9RjDUYAlXNuvcvt6T4nNp6NBtCb1ARMIHULogmot/y8LThp48veDrd0OxAsAEJG6R55nlhuhxo20qzGSxUDKUGlqXDqnK1qcyPEVMsRjxLE4oNfLxs+c2nmTfP7RrAKwUvBs2OaVKudgYfOmLM1HmBrIgsOBG0WaZyw3kKdDUxyIzGjNiOauFcBNoGzSOnFPjTvGeKA50NuGpozVFbY2qBmDt2mw1z8bDAaAvd3lBLpWqhaCaKkjC4HAllQeqCsOHjJZC6K25dph8yh1Pg/rClhQ17v75Ba5tZguHRWFzt2JgecmTvGkYU1yxFBe8dgwhbG+hU6Jyj7xzvph0fDjtK6ym2IUbog8PUAjVREZo64lTwtnFtB/bLkSqZbTAMpnFlDeS8hqF9RmvEth1o46srx35SnFEahblGNrZQUCrCYBxOLEPLbChsLQwtE3xHZkeLW6gDwVoChb5lein0tkKbqF6Y7Y7LvE5grbF0ziMmIVox5BU61yrFG87TgvMjS2kEyXSdkPIFay3VOoo1XAqMwVCINLXQKXlpGK3s/cBTyVR1FNbArgGW9meeDJyLZ986Ri/87c87vj9U9qPy86vGi+5I3+7QlpmKod963m42fHzc88Oho563HJrhofScdIcxAzYvjLbgJWPthRt75llXuH9Upuj5fHGUNtDZyo1f6GohdBNeCg+Tcph7znFN+xrpydXQiuHtq4Hvvt4QTKTIQukyzawY2yY9MW25RE80dU1Gi+Uh9ISWuKmJrp0QVXLbkLXgggdjGa1jSHB7fYW9KM0cSDkSLz0ZobbCIw7Xr13WuhWKr2jnUenYDhnTHhFmtJ0o5LUSFvy633EbYtrgTM8mWHpf8d5gbKPfetKsLMGwOMvW9Oyc5ykXltpwdsNc3vLr9z10J7ZXwmUKtOr4+rrxN29HfPvEruvIfqH5iOkEWSLGNRYbuFu2fPop8s0rx1WfSb4SU6Sakea2HJ+2nB87Bm8R12F9IGbLZTZMFZwaLAlvzaqBbRmK4IySfGFpC0Eu2H7L0XSktqFrjmnJfEwdT/9zx29+sPyrf73lq5cJ1w9QO5aT56d3jsP8jq9fCW93gdf2mvfmhv/xyRG18TIYRBsqnhw7lry+XWltlNRQ95qUr3l39yv+8beveVwGjmZkKR6rlVIaQ9/TeXi4PHIXDXMVFgoMymYEGzyZwiUJUxJM8wQMz52SXOD89MCjSZxi5pLXymY2aybE14KZZrxRjFaytnU9IHalGrYCFjKFLjQsKzrYq1Dp+cAz+mtLmpSn5MjO0rLHlQV7fGRoOxKOnDJjb/D1BFpIWmmlUc4ZWSriHUlhiRnaxMAtaYKHc+FVB6ZzbFi4LUfu8oZitgzDhrBkXPWktsWEEdHGi30lLY37hwf61uhkoacRSKvOetzS0szV8Jn9aMl0lHCF1jNv9vc8C99zPU545/nT3Q3/w2+E/999z8KGQSxjXRjLhBIxdWBqgVRH/vHpOeec+OdvhLe+YZYzUiqUStNI9YaHh0LoAq9f+jXsbAVpHrSHFlBeA88QE/BGCFSMVpoIDg8tc8qN75+u+TztWWSD2IB3q49AtVFxHE8L80tBgqU1JVW4FMdjbFSZ6eSI69fpmxFDr0o1hkkD95eZ+gKaaV9gPIVSG8u0IDqjVekIZNMopqzUQ1swVKaSOVVHOUVsyJiaqDlzkS33TyNz9OQkLFk4d1t81+N1/b00mPWwaIY+eIIVDJ4sgRgTGcfV4Eg5scQItlCaYrTDqcNQsC1zqiOX6DjHmVvb+HgaqHVVQQ8mM5hI7YRz3/PusdHqFbduw7Wf2ZqFjYGsA5MdMTJAzfjwkprO1Kqk2XJ3vuJz3ME4gk3YdqK3C4MDNZbY7Jpd8wOkxNbBhsJlngl4vFpyFp6GEe09i+k4LZ5qe6QteJMobaaVisiCFV0NpTawNEHVUKuiIoQuQMlUVYoGmuuIEWzzWG301hOaotnALjAzcZGeh1NPpuOg63crYKFmnF3f2kULwduVF6GNWlZaY9aGNMX3ioisQDsUjKXUSnKNpSassZjOQTzTi2FjHU46zlXwzmFQtNU/72Vg89yyH6/wdubn+8anKRJr4RwXNkzQwGhHCJZnV4ar6577x5HT5x7trvjpAveyZ5GO3hi8GM55YsfCi50ybLYcc+V49hxT4GSecdGRjZ1xdkFq5HAqPMZMUk9tgiJ0bk9ZeuZFyLOg4cyvvu3Z64SxM37Xo0lIF2VpHadkeZgzF3NGnaEZIepMGC1drJg4k7SgfsDIF2xobRgtbFrEy4wbMpkZXwuDNdisJHH8FIWxC1Q8JQiTVVqrbBBaWj3frZ1o5URtM0ZW+yHGgdmSyxUubAk+Yy34YFGnqBr6XcBEwVAZVNm01co1G8E4i9h1v/m7e087zjQXsLbnwyXShcLXu0jv7hHXU4ZGrg7vCk4A3fDT/Y5puuLDUnixy2xMplqIDEzNI2bP8bhnGJ6tD/t+Q1aIrZCWC7SAUXBW2XQdgwtINgRTyeFMihl1GdMJXPdMd5VjhLSs6tlzvuHz9yPVvOff/dvMYI6IeqajMux2tE3PT09HzGaH58AbLP/mqz0/HU+cl8pP2UJzjPQ8xTeEduHpCB/ut/z00HN3vuZhecu5fMNSr+h2zwiDYCqkVjhOkfszfJwtc90jtuMSz1SzkIISL5HLdGFpSlPlu3Fg0zyPp47qr0ErG9vYbhxPp0KzAbPZYFpk2y7sTcW3Bes8n6fCp0vicppw3pG851QXsnoMM6EJnQidepJYfpf2fMpgaJSN4yKNKiCt0UtmSlBbz6lmyhjAJLJOYFZNdWuGrELOmWh7Sgu4VgkaSVn46VH4i58LgxT65cDGWpzZckodfhnIxyNXdd2pnpvldmxs/SOts+w64W238Mad2YjjPCeeaiVZ5WdvGq/LByQnzqZj//Ut+/4Ol/+A5juEGafKd7sX5J+/5FINf7os7IPhZzYS0t3qGeg2/HFqTGr5mG+4/3jHt88y12FCl0jVipoVtNNKY2Lkv34/8vL5Dc/2n0ETuQo5dxg2oHuc7Ak+YIgEZkytIB7DSmyL0YFuMB7GYFdbnslrAwjPWRzzXFjYMoyZUhqXxXBqHQdZQ3W2NIwbsN0VNWW8tThJWFVKheJg1plsLMkql9K4ZJhzJaWZNIMNhiSVJplFhKqBy6VxjI1kAj50uHQhZvjw4FmWHUsEwdGawaonWMW0tW6qdl2BLPPEm+cDVy6uo2XZUOxALgvjVUefAlpnBqe4pBjj6IYejZ4uLqAjj60ntz339xfOdVjzQzUirWFNRiVzfx54ys8YfM9BZipPBPeEt4Vq9iunpnoWTZyjYeMCW2sQVQiBjQrFVhYKzRuqJjrW/9fc1kphbat6vsWFgDIpiDrEeO67kaVmcikcLpVz3VJdx9AKyxIRmUmSELMemE6EwSgLgVgMxY/MCk+tsrGCpML5vNDvCoOHaUqMxtG3TCczqomnJig9x2nD1PaoWz/r3ru1om89oxFIZyyCtEIferysUwpPhVYxJqAI1VRcUcIX6uKsmaSN4r6sX4yj6w0uHQgIpvU0W4kZNh7snxs69HLY4pvDRMc+ZP7Pv7KMrrJpCyUtfD4VLtqhJuJT5u5ceErKWT2PhyMPFyhu5X337cyNn3i+yeycsiTlD58DjzqQ2BFLT26BJsrWVzadYcorujFTGXrDVW8IbjUMXjrH3Aql9KQ20LtHOjeDnKi2kbxB9lt2bqCayIt2YmdWfOyzbWHfF/ZtoEwBKYIbOuZgSEY4JWVuwis/82L4jHcHRBeW+UgpNwx1xKmgOO5rYOpGnHF4F8BUVDP9oJR4QbqEs1BKBtMQ56gVtHZc5gD9niIGnIUvzG/jwFnDm+uOc3Q8LpbtVNm0DJ0nDT3dOHA+3dPyzGA7Jqk0cWQ1iIw8LsqbIfB4/ADVI+wQ77mqnl4yfXbUXFnKQNJrnmrialNxTIjtiHQ4WzgvL6jfT0TtsNueq5cN4x8IAbQJLc/EuiCpIk0x0jDlTB8C51QpYhGr3N4YOCg/PkZyFcQnOpOJ7Zo//Wnm3fvM6+szLAmxHa0antLAx8PAP73rqLJniYm/+KtP/Lu/2fL+fuLvP3l+fRTOOvKPn7/h10/C5RR4fLrmaG84tBENO/zwDJdAEGrMFBrOOYwNzDlSlkqhsr8aGbdbkhl4TBdijITNFVYEUwp/+dUrzj+cmKrjpAbRLTId+XqEve04hRsOpSfGiEO52Sv+C7K6B37+/BpyXSctXc8Ph8LHGFDbsw1thVEtnk+lcTEG3W2ZU2SmEUUJ1nI2W5wrnCMcSscxOn7z8YlfXm2wckIpzN4yP7/itFRq9DQ2RP8MSQ4rSlLPhwfIP89INzOIwyTDQzJ8uHimsmWYobYvxrcc2e17RDOHmFApvLk1/E1oLMcJrYFTDeTcuLYL23kiZbBhR40zt9dlJcrpKvRyJJz0pPSAb5GbsKUPDmLBpIgJPb0Tur4REWaEYDcE/cSYH5CaKOj6YDSZmjNLaxzLc/746RvG/gHhREqV2gKUDeUkTFro9wYtM5QjFoPmDm92aCvo1LFVR+wUTGXXDN38iGkLlwymdsytcs4jEjJLrZwTXJaVxREcGI0kMhhwYcAWR8DhW4PaOJVMH5RzgrM4LiKc0sAPP8ApWpbqsU0IrqPrO+al493dDXeHK5Yw8kjhw/HCq3BNSZmH5jDVEkxAVVarKZmxC8S5ElsgqiU1T0sOTYZn155sGlOyHJeey2nhaCOjs18mAYnOjWgs5Drjm2OnCyxnTv2WSwpM0aHVs3EVNcJZC75WWoFz2VLNHrBUtUTgsRRGkxi9Y5aVhGrV08yeS5tXLkBLDLow+sRjUk7GfxE9WbwNOAINwVmHaQ5jDDmC+IDWjuQ9WTwfa4+jI+ZEdQ66DtTRFc9DhNtOiFSKAQlhbcRVMEUIZsPZdDymBhGureVVm+m0kM9ndv2OlCc8js5knBaSd7xvhuPFQ93ThRFjVnukkcYgFqmVUBYcQm26Ml7aheuhQ0skKKgxq3FShKINYyu9ZFKbKaYwxwk6g7Kuip0VNuMamu7qFtqCuA4VaPXPjCO2eFL1mM4R/BN/cXvPlXmPTPckU3lzu2MxFRcS2y7hygHz1YTdHLmopWlkExauxondeMcYzlAWDg8D7x5+wVS+4TE9J3PFokqOJzZm4uWzCiUTq5J19U27zZbcBabScTgWLlPkXB2xLbz0YMqClQutnmmt4IOjGw1CJvgDz24fwV7wZlmTwC3Q5lfk4S1deA7VQIlkMSytojJxtTnQtU80s1BKwTrBdD1UQ6hAqcxl3W2NvvGi77kaBXEL3eaIzDNZIw5Q7FpZ0baa8toVKW/IohAS3dDwkrA+YlwHyWH1zDc9PO+35LnQS2Wz2+A7S60nNuOE04XNAMcS6QbLrBe8gxdj4jYop0OiyJqqHn3PmBRB0NhxGxPSEtEEmhuYU2I/OkwVQhUCDs0jZX5BiRsuj4ZWIy/fRkaXKKWRTcXJegkAJSaPxCuqXuE2gdwKuQq9bWx9JrhAtx/pR0+5OB7OjmW+4ve/e+Dqu0ZY5tXWhfKffrL8518PXJYd2Bu8dPz643vOy3v+zT/veRE89XcLP54PbHQknXZY2WE3N1xiYDI7nBmQmqFlck00DFGFqSlWK65Oq2dePKenR2rXmLQwqzJnpfpAkUYPnO4OnO8K3m7wLlDzChOpT/doveJpadyLxcqOmBOHyxFjFG2grcNFQyiFjV7YbRXbttjaYyRg6wGxim2RrltQW8h6wXSC5koSy+yEP+XApwTnXDmbwMUa/uvjia+ON3gx3N8fucww9DtOwRNaT++3VL8Ke8rcSLowa4+1CjIhu2d8/L3jjx/3PJgbdBgZAnSwmgzdQDCGajY8HSJapjUB3iopN5ZSKWFANDNqwukadFqyIZ8L5kXDNEsxnoqlSSC353y865iWleTmnCXVgPrnWD+QvaXoGrLFCbVYWq5Yv6BtoXjLYiwLA0X25NRRjWFZNgTdsd64M6nJqqC+G3mYQd5Ent9aWgZpPUbX75oi1NjR4kLOlsUYXo23uEtG84JJFZFKNTDVLenseXxYmJOwOmHKWhlNiS/MulXaFq7IBYxYUh743cNCVy+0Yrm0gepWz8k87VhmyzEHouk5tcKswmXZc/+0oZgN2ixZLZTKaJRda2ycZxpXFwEIxSjeNAITxm84pcaCJVWD0R2fPx/5q5cGJGPqQikN11mE1fBjJFBKRZ1Qy8wcLzRT6FyiYyBmxyUFarQE5xlDT6ZxTg0zF173wnehkqYzLgR6JjZc2LaC5pm9ddzHC2XYrhMZNbhmGavyFZmNPTI45dDv+P+ehQ+px7kRJw1Vi9ZG7z0lw6AWr0pWoPdk56mtQ1gxysY7pP7/lcuJS5l5V5RHHKkGEsIyKUsTWlmDrG70pGppzWKtITiD0US0galWFhOpews5IS0jBKYE0QToITSLzQ3XelooNJlxtdDHBdMi2gWqtbTayDHSD45GWzNf1tOcY24Lagw4SPOEdQ00Y42SSsK5QKGhAr2r0J0Z4wlTtyQrXFoh/LkvA4+x0mqhULiyia/jAdVH2jSBqfRmZnctBH/CtCewldD19D3EFhlCZLtJGJ4QTtAK6mH/1RV9eGLTwfZT4+P0jnBV2D+/5y/fNt5szpwfEz8+DtylHZM857ef9jzVDtsKpqxMeyeRl9fK17vE6ME3RzYjLVjUwHyaCBuQ8BmrH8jpiUzBm47W1n7mAozWErTHU3m1tzzvBy7Lka17otVHEgrWUKwndh11Ngw+YOMFIw4nmett5pvnhau+4TrPnCdizTjTGAao1dEkglS8sWjrMEHp3Ef2z08Mk2fnFVXLx4NnyZZqBPLMbrTsXvWEZtFu4JXLtPke0z0h3rNYYU/jf/dLoTORJZ/Zmju6+o48TNQQmJcjJvZ0tqdVi6+ZrbUU4yhtQZOlZIHBs/UNKxlTFlzxdHmDRse09Hz+FLi5uTDcTqgTEolUZpp1ZJ7x8eE5H99fkeWa0Sq3Q0bCws4afHfCbBzL9TV/uE98eDRcdAcy8vFpS102tFyRYeAUB378aeScnmH8BkMHEni8bPkf/m5COPLfvI38d1/fcHne+O76SJkrJUfeTU+ET4F/zI3We0qe2BHXUZwIJ9kymxuyEeacMRrpXeUSz1zqakszoacLgaVBqguOxqksND+CGIxURNZdKi2hkiiiGC+Eluk9VB1Ixq9d9yIYEaxtOE2cl8piHJ3rERpOoNWElZnOpbWBLIlaEtZabIMijY81snGO7IUklUkbmj2/niwPxxOnZaDh8K3Sb7fs/BUu70imo4wj9y1TTGSSgDhHxPLuPPAP72448RXF9DweD7y9VroKUg2nuXLYd0gfSCr05sLGzZh4QXVDaY7TUrBOaWrItqd8CXq1VJC8or+r23MpA0UHfnj4jk+XF+uDTQoxReZkcWZHp4aYl3WVpivhLrbM54vhuHnDzJlj8xxzx1PtWLYvueiAyoK3Xy7qdXU4qOu4u9/y+fiMp3zD9M5SGfm639GIUBKiFrFfQmrtjKihqEd8Xv0ozaKuonnGWodaz/uPcLir+KEjYihmQJcVEaxjx4Sn5BHVnkvxRPGk2vEPnzrGfst+w0qRlIaTE6qKTQtjszRXyDgelpEPFyF6j3MdYoRCw9YOTSe2JFpuaGeIo5BVKLVAXPC1sh89xzhw+RI6NvQ8HByneRUiOSlY09gEyxg6IOOCR4BLvhB4xCpUbRhX2HeW+RiZGREcrio32z0mJy4x4Zzw3YueXwKvJCJ9YzQzXTkx1jOuRT4apWfDj0VQtfTAzi982xe+cY+E5Y6tLSTzyOPyisPUIcaSm4CuwjJTEjsMmzYzyjolmklkV2kl4qyCRoSCqeBNptRCDguf2gzNkLUjKyxaUdsYfce+OqwEbPUEA10AWzPGWhbteCyFQytEzgQHgyqDJhKQzMoP2ZsFmTyx7GjZYkLGERmxaPXkaqkmQD0TUEyp1Gyo0jAexKzwrayFportHbXMOCf4KqQKIgasW8VldWbvTzzrRn6cr1najqU0YvkzXwamZe1Fm34NtS3zhZmIbboqQ5NipNL0AhwRW1B9YNhUNr6sLGcymmdoSgGKgtiJ7fjE3+7+wC/fvKPomc7d48MFHxQ5zWQT+Par52h3w5ze8MPnt5zyNWo8Rme2PjKEirQdJo1Y2dFkJmrP4ykxpYVWC1eS2YyZVlalrGLAjXhzTWvXZHPNJe/IKZDVUB/PXL812K7R+bxKi7yhVUMMnqe4UM3Mtmt83R14c+N5fV25Gk7cbDKYjNgd9nLCdwesOWJMxnuhoYgBz4wx95T6jlovq5lwfI6rO8ryilr3TNYQjaFIIKfI22uPT4amBSdnsAeaFI7Ncx8T0jVejPfchDuWywNLPSFuxrQZfMaUDGoxNqwCF5MIvaEV0MIqUtKO+4fE9rllS0Y1UYvgm133WVmZzo4YHTWvaVanmUZjaTt++/0Lfv0fXnHJz5DgueoSr/uZ3a6xf5Ox/QObF7f8hw8Hfv3DntT2mN7jbGYqG875hpmR45Ph7/9geHwIdKp0slDLwtIcTYSH08AfPwc2euDtszt+cdPYuQs0A/GJl6r84vaa/9R6/uHhPdc3yv/p657hcMd5mvk7ecn//XPm1G6xbuA6GPajsBwTh5jJGJwKpSVqa0hJiIWLQAvKrIVExZm2EtAYydJTAMtah/VGKNWhK6R1JVpawSC0FphbT7YeUHKJYKDqZQ3PdRABISNmfbB5oyTJRFZYFwhLOZHEU8h8PF04zzN0A0UsU01Ee+L5i1vqvTIXy0mFqRtIDDzUjgPXHKvwn76/4rE8x4aAN45SKs4WQsvkOjKz4/vTyHLKZHVc+Y6NvUdiRFtj8FuGHBn7gVMZeVgaSSuzMeydQ2hU7fnhw47vDztOeeBxecVJbim2oVzQuhBbIuqCiuCsYd8MphpOuRLskU/nhf++jnx/3LC4LWIGjAjObKjOsneJTWhfDvW2Yp+1J6VrnuI1T3VkSkKsC8//+hlBfsK0SknTl+xLoPdCp5WpJYxbyO5CqYnqK94qG1uZjxOHzxbanpaFQzJ8WgwzgWKgKji3oeYNRkZ6N+CMUEultIFp6Ul9ZhMSz7qZTQeSRpyHVi1BK6MKgxqk9hi/oaqhD4FUEjWDpzG2mSIdl2rRDtQ7NDZMW3fS3pzYDQOPpWDsgKhlSRt+9+MH/tk3grGVLhhuegd5ommm2RUAdMgRJ4paR2cstWWq8XRNSQhZQZ1l4y0bMTgn9CaQZ2Fg4et+wIcLez3iyoQ1mS4vvNGFNxvHD7KhmolBEjDRmZldV6AKJi+rdK69YlcNSRWpAWMNwSh9VYKJbMoZRyQ7x9RWn4q1Dm0ZbZFgld4KxDNCodnGpAuxRIx1YC1OLLlmakuMzhDKhdhGrOlxbQ3iFVklT0dRJqBJoRqlSCWiuCDUErGsn/3Qj6SYUBnBWJZY6d3aLqlioBb2NlBqQdWztJVMmYus3hHTcNK+vBC0dR3SFBXw3q8OGTEQesiFjoln9si1O3Myz8nqV4fDn/My4CTjuw3brfBmV3m+HRnaLWW+IF1PsQouY9wTWlYgTWkLqUaCdesOvDY0gfWW2mCudd3BevDDAz7fQXuCdgEE6paahdoM6APSntiFC3/zuoI+keITwgFrCsYMPF5+zuP0HU/lGSKF4/SZScD2Hrf0aG0Y41DbUXNab1/Wge6Jy5bLk+JrRoolt5HT3HHmws9/1q+0QKcUXWgO1Cc24TNvrw/8ooPQTfTesQnjaujSylIq5AnfHnDc0XSmtlW+0XUjwfcQI7YdyPV+DX2JJ/MZ63/OfLriUhRvHU0MKj1NKzVNbOyXfm2JZBzV98xNKT3sRsswJDieKadHJjsTxoAEQ2kH8BbTb6hppqmh28KYKzYqVSyLCsVYGiM/HU5srwKOkWk5cFMUlwu+NKRYrLXMMeM0YhpYueHj+1f857+7JV9e0XUjVgs2TzhXCKWyXA6EfeVwX/n1b4S53RDCiLSBSuEUO94drklx5sMH4d3DniBbOmm0eERcB2pY1JKz4/v7Sr/b83B/Jk2Zn19b9qHDaoX0hI+Fn5nAzfMXbDdH3sgHjLlwRcaI499T+R/nxDEMtFo55ZmcMiUHqlZynLC2IK3hXSAvhbNVWp6pcqGqgHPcZSXbPal4siyIFEpLTKwAGMXQgKqZOa2I0KkIMU/UWNeLQksgGdEJbyvNGg7Nk43gWHf0RQuqhVgNisOxJp+bOpQVPBWCX+tyoqhXclsoHNg7paSe5h3FWCoD76aO//DxCpc83z/eIm4kaCO1RmuCsYqzjkU8KayhqMdWSOkDb1nw5UJaIi1NbF3hzdjR7UY+PXT88SfHbBXbZf7ltUNnSE+ew+e3PLVXHNUyS49xjW2IDEEJZub1PvOqf6B3M2KFRZVjVA6XM703XLstd8fnPNxvuZ93bMY93q0P1usNvHWZ67FS2sqgyKLEqlhfMb1nORkWDRwP8F/fFb57NtJZsC5BWaitUr1dpxtT5ZhmxBeqFZbqKA28DZwOkc51aNb1zbMYkDX4WcSsQ191GOdJJTH2FhMdQQIxNZp2XI4Fx8Juf8L6tMrDnKMsFXJdvSt1wmFpknCdQ2Ql1qUUCV1D0ozHIBKYa2LRgveBqkKOGWFhN0xs40CLgaVCk54PD41vvxKsywR1dAh1iTQy1UHywpID0p4zqRJT47E4inQMg9Bq5uIUFcOmg2fuQn8jhPAcSORciArTrIy3N/huRJYZ2ywh3nOjRzQ49v2FgQM1GO6bYvuOKS+05ilpy+PJ47Lgtj0pWTrjsC1jS6H3hc5mxMBiLGcRYiuorITURibnjLVKsA1qprRKNRnrDKqK0TVoJ03oDPQOXDwzesvUKiKCsY2ohdYKzZm1pmjXxVBqlSJKEEVMJZaMpyK2QL/CpzSNuK7jYjKpRLQleoQrFbw4JgyPLdC0p7OClAgaGZxig+FSK6muv+tq10qi6kp+LUZp0q1/n8xeFqyJJAZS/TO7CawkrFSWeeISGvX6NandMMuZ+7szWRu3YujeTBjztKKHJdBqo1SDFhjCgHSCGrBasax75t43VBakXFDNqFiMOihrbdA6S2mZPGesy9i2mrPi8ongFtRbbL/j84PjNz/MhMeFZ88zz54NbNxEzneUNhG1klqm5rp2u1FqNZw/euLjQD0JgyyYvGCLw8i4wlJyBZdW3rVrNJ2w5sKuK+ykghSMMxi5xfmvICuXs/I0CcYWrt2M1YhQaSmv/m4/YjE4A7VMOAqpgUqmhTOpHViqEpznxQYyyqclk9I6fj1z5OZm7bAX0zG1SLGFq33guhM8EaYZ9wVNGZlJshBlQZ3BdMcVZxs3pGQIg+Bao3dXZBm4lEJTw7EaLqWxM0LLlhIbplSkVmxTuqDM84nBwtWwI8WRpx93kH6OpcdrZRgcNQunIrSz4DcbVBz/0z92nPJrjNthZKV60XtSG/jwIVLuzqvtUi3W1JUKScclGRKGU4lYSdw9XLi6qtid4VMJzHPm2d7yxt3S2cZDnPljO6PW8dYbzGWlP9oAb+2F/+uLDf/t8SN/f6n8yT5D3UBoPftlnUS8GAdKPlK859AM2pS/3GRe3xT2XeEU4Q+T8uHciARCu/CLQXC6IF7J1hDVkRtoXejtwi5YWu3IVdnu4Hq3o7ON0VdCW6c3c4P/8jDzP10cVYSNVXpbKG2F8aDgzPr5nmvFSeYkmVLMKhkyFkWpTii1EssB3xJbPAXPYnqSsxxjzz98DLzsDW3T43PDLHUNoVVHSo1qPNCBeBSLiqESEK2UFplaoXaWwkIInrR4fv+x48dLx8ka6vnAt89m8hBY7iq9eAazXoZMy7zcNoYwM3SCc4lX24/87fU7bHpkKYloEnXTyLeN1gJOb3k+/HPen/ZMHzM1JcQkhu7CX79c+Lr9kavuSDI9pS2UsqAkum6hHxLmOFAaRLb89kMlzQObMWD7kVQSOfec84a7KbBkx+cS8LL6N6ZimbFclkRAuDaVvV/HyNuhp6seLQ5jwoqLbRkkfVkDZAY7ILnRKpRmiU+FuV4Y3kbO8YmShTlVctNVk15Ac8PUBLJQqiLN0spKo+v6NVdV60R1PZe6VmNdy3S4FfDTKk5PXA8DORsyqzU0asfT+czza48tGakDJYGa9YBPpdFqt4KfyNxZOJiAtZ5+12FbIDRPvGR6l+nLmdQaqEecpaVGrIaLGcjnwtf7DVscNV9WYVPJeJ8IrKG6jOL8lrkaTrrlNN/w+eGKj/kVvemwWTDa6JtZg4O10LRhOkNRQ7aW2ISmbX07zxGkAZk5LqgVsJBKImmj0bBq0FaxEnAKW+PpRBDRtRnhDNLWUF9Tg6bC1jrEQG6F9mVlU1sj1ULnV27NrI1aF4wT1BdSXqhi11WszSuxsDq6BENbG2GrC2RAsxCcJYjDlrL6GdTgjINmsJIoFDCC1MZSvlgcpUPU4QBRRUVp9c88GUj1zDJ5YotMZaalgK0TnI/YaSIX5dN7z/kvN/zs6w3eR1Qa1lo0F2pLLNIYuwBUTGv0xtKHsI5ec4MmKJZm3Sr0aSC9RcqGpqAECl/xeH/Nx49Hbt+85GawWOeZzyPH954Oz/V2x26AobvDu8ylJWbOUFeylpgO57eIhZSuScdX1MOOXi2hFXzJlAixwal2fPjc8fr1iMgJjICuWIraLlgL1lmwZq34GUtrnvNTpoWO1qZ13wNIs1hGWraQBtQEqDMWg5EBjyXrgrKg/owbFrYp8jwYxI1MP1ZOpVFroxmhdoVuk0l6Yu8T266y2MZAgGmGFHFjw3Am1TNqEkbSasnbCzZ2pLLjcfacVLEbh6s9tg1Ys6zhMDkydIVrXzFTxRSLxWJao3w5MDb7K0zOLDP03S1v337FP/6QV068CsclUsuGQ96ypzFNym/+7o+8++kFndmjrawGRKs0SeATk2sYO6z7XhrONnLwHHKmGtB25JtnmZ+9KHyzO/Gv/nZguvzEzm3Rk+fwOVL7jjFX7hbLQ+tJwG/ihV9d99z6AJc7iB943Q78X4Lyf9jv+b/9uPC79B02j7wO8NcvBva68PQE0RtOIty8KPzbb8881wdcVs6z4/vLyIfbgW7r2LgHXl61tTpnDGCJKKmcqeWRPjS6NnI6Dkxzx9Bt8CZha4Gy7q7D4BHn+GY3ED5seEJ4Pva0GHk8JLbDDZ03OAtRhZ9OF3qpuLa61NUoIoZaF2qpqChJ4+riMBcEj0hBRSji+HS27EOmmQnXT/Qy8HSaoWWWeWIOGxTFGfMl/2KxaqlieGTLrJEoA2d1PDw4KiPHek0WT2yOSzX89KRM+0AtlY7CaCIXr4St8N++dfTxgSkXjthVeFQiukQsCSMLTSPeCemLDyXHC72eedZtOS4Ja078bA9fhz+wm97RcmY2Craj1ghasSbiTMTb1T1QxbIUoZwtnx4TnzZbHvEUNhwmQ2wBVDBLI5ieUlfjXDJCE+itcsgTz5h4ZRpjJ2xiwzZHsgY0Ejy0GnEGXC2MTUgx46qhtI7aPMenyjJnQhdp1TBPkamC+IGcB+ZqEQdoYskJZztQwUvFOktVYSmZSy1Masisj6rcllUi1wSTJ8byiRtbSXTMzmC7gSU+IrUQ8OS27t0FYYmZ0sDi1oCsFJq1IOtlsNDANqwRem9xZcbkI46eqpVSKlbbuvKRQFkyH+sToS/Y9ISUtIbzWsZLQoioCFE6fvdwxcNp5PEwssQBMSNOC+S10udaxovQOYtRv9bNXUeiEs068tfWKDVTNeMCqFNO6YI4SzXrQVlYGxCuGShKbzxbsfRYEkr6EioVtxoUay10wF4L19bwORZO2DVkyRpQPKdE56CWgteK5Ijay/r20QxiZfW7KAS3w+sqFXIZOlWqVEQHJCneKsZbCqvNVFllVcYIpgnGCNkUDrUxq3BtwBr7BcgkqFbcn3tNELNhImJc5LubiZfhnmE64vSeIJWntuf9suWf/r4x9M949iyRk2GJCecdfSfQTuSmtNYwKogN5NkQ7ICaAZwnpSdqA2SLNZ45bXj36QUPk6GKQ+WGeNwwLzvefb+w3Srf3lyR360bhtE+cfzNkXgL7rvCm288aXGYL3ZBmsWYEWc94iDP10i7xsmICxWHIDXRBI6xcFg2HP7R4fwV19eHFfmpidYK1vXrW75aULe+SYnjfOyIaSTZdaxbO4epYWUKlECJe44PG8Km59mNJxAR4QvQw6xyGBI2XFa1LJBioSXD0HrSdAemMS2N7asLeztRzRru7Bz02VDPglQBk+lDoUlFSBgKrTZqm2iDoMVz0p6jDkTraSxIO/BqvPCsu+ebqwf+4nlkbCN/enQsc2A0a9glJeEff20wPwaut7e41tGZgaG75tVzyx/eHZlyoOOalIXoKqVlzvnCw+kKcRs4P+AEjBOaF3RfIBTuS6Xrt6RqiNGi7ci2S7x8BTs/8YtXE//sV42b3YWNX4FG+twh7RkyOsrWUY1gw8jlB4VDpCXLyXT84eMZf2PZuWvmp8q5XrjeNl5vLP/2zZ4/fYbZeFxQ3PKIXh7xl8isEbft+earwo35iD8/UmNHrh2318LL0WO4Q8o7xpYwkikiZOOgLgzhiA0HnFUMb4inr0nakaLlEjPOWRobEhtcFUYbmKph45+RpONwmbksjmieEZdAnxNeFjKFqoIVz8ZuyBRSq2jONFVEItYaFuM5GsEOjaVNZIFqLMUI95Phu1uPkUbLDwxD4JnfkU9rKDR2HjdWpK6rFa+GWCpFLT+Y59yNz/jxqfB5Uubk2diRV+EGvKEkmGviWGdis1jjMbkiNlMQtqHxUs64+hGbhNw9ZxxvCC5TJGKqrvkJvz5Iz/WKHx6e8f7uirs8QnD4ImhK6JwYzILXQquVWgu1Rqzo2uYgYyy4AWxLUBRTImKUHJWDCXxoHVVGVHrUKLVVQGl4alOsHWlqKBpZyDQbeKQSKDiz8iaG3CiaaR4uqjhbMTHT8oIsBjN7DKtULDeodeDdhwc2bxstRwwGK5ZcGikJtSpoplVBnAcp1FrxNGIDJ5bFNC4lgrcMBvqckVaRJogZIC905ZG9NiZ/y9QsTQDCul/HMmtmMo2aKylX1Nj1IlYTl1LQcVzXb6IrH9/WFT/uOlpdMKIYlArElBgaiFaUTCuNy2XiaThg8hlSRk3ASqKVC80sNLfl4+OW//j9huafY2u/9vN1PQyTgqXgzSrpkVIxFiodSWBuiSKCKBhdsx0iMC8XfCeY3hNLoggrh6NBbYqh0mHZV7gSg//Sgmmtkr9o26uUFc1swGllk+f13VACixpiy4gCCHPJhP8Vh50oZcaFDjBIc6yoJ0PThDMG30ByoWuBZAKlrQbg5hvqKtU4si3rzyRKbgWsXc8cVS7OcM4C5UjAIG0hy0TGkYv+eS8DufU047gJB9707+jKO0KrDLbgLLQgHNLI+2XLh4/PGIeIMtKcUHik6AwqtKLU5lE8WvYsy5ZUtkjtSPFIzBXTwbiMWL3h7qcX/PT0itR5kIwaTwmO4gTsllQiL+8N7lTpbWVqEcme+4+rpWuztfRhw20XKGXBNYuIUPOJYALBDvjOop0HXUlfUiwhW3rTc6iW82HDP/2Xib/65894dpMwOuOloWIQa1mljQOxBqZ5x2EaqC6iHHFuwYWATVeQI6Z6OkbifMuHz55xHPDuAu0z3jqQAWPW9YWjsaXhlkioym4RpnkmzRfKIFxxYZ5/xHaRvusYfSBppKXIUbcMVx51FWMT5AXrFKhY7zDNEGtChwTjAnViDDOv+zMvtg/cbu8Z/APOznRGqZdXOPsXTDLSmUStZ0ru+PHxK87TDe09GBkZveft9cDLr7eEzYl3H85c7iYwGygJqyfYRHQ0cBPh6USf3Sp490J4bhmHjmkRjppJLhDIfLf7zL/7N2e+/RacnTD2gHCmLIk4ew6z0m2e4yr0zSFTAqmkaeL5zS0v/vI1c77Q1Qf6qWAfP1GmRipwmjOtKi9vE3/1Gp4tym8nZaqKuvWtx7r/hbX/+LUs2/YzsW9Mt8w2x4RLczPvfe/xOZEsiqIoCSioUWoJ6qmnP1YNlRolFEoFsuie43WZN01EHLfNMtMONdZ5VJcF3OwEkECciBNn773mHOP3+z6lt47kG7+8dRx7T1lHJj1w0kA0ii9XtF3YeccYHEEuGJS1FVKasHZCZSWL5ednx99+/5Ykb9mPHinrlni3gbkq1nX4qeOnh4lfL8LqDcIerEE0gxSaRlybsSZBVooOaK10ttHVlUEyX9469sOeS0784dx4RjGyMNWVJZ/IZkTtyCV7pgpB4rbnlDPWN/b3FkmBHArIim+K5EwphU6EmpTPMfD3p5lT7lnFkX0AHblKoBlHMUouAApWULcydAOjs/jq2PWJA5GyRkLZbQ/w0SC+Q7TgrTIYTzJKczt++vwlf/OHb/jh9J7V3IHptrBe3fHr7878777YDqxIBsr2q1TEChKEVYSrSdTditdGyJbV3PAiA2ftKHSvaOJCa5X3+z22WUoLZJu4rguqDqMVtQlcpXnDIp5zmvBG6JplYU8zHbFtltVQIz6txFhptUfcHgWys2TZ8fnc81dfdyjxFb/eITWQV0e4Vo62gstMgLVQSyWI4WXNFDVI6Pji7o697fAlYi6XbaztHaskQDdNfJ3o3QHneq515ZIN1Xck51ir5SEWUskgZst6aaF6Q2QLqzojWG/JdXuQYv8xwwDFjlSguvwa3FT2fguGGlXyHDnlQtdDC4GkjuwzeDB+x8/nA//wQ4/yhiADvbdssXmlIFQjeDEEMRui2TSqXcFBEVDvkZIxLWN0xUp9bfY41rwSum2K21pDEKxYVIRaC6JwL427GkEtK5YgQtRKtYZUC6VmgvPkknGq3LRNynWOwqQeYyrGGJrCUgtVC14UpBHLpkJ2ojjpMBhyyYgELBlvHUaFhqdZSwuGiiWVK/QWNZ5cZ3CBmgwmKSINK45rXDmVSqsrfl0ZyogxcTPL/tcBCP9XtAkyYD29RDpzpWolIVhjcR30rjC2hq6GmI6QrogKjUx/9GDatsMUD35HaW+Ypg8s80CLGWcyswqTWMiFxwredDC8YVwOvCUz25XZV9ZciG6g9ZaBkd3zC2NVutZwrZKtITvHZTlwflz58Fbwa8XVQCvQgsGMHdaMHPotzbtMC1YiyIIMFhvttlmVgrUd0/Udf/+3kX/5L2d6f9naAK1DzGbAavSsa0+OK+oKnbvQ+wvOJWgZUYtpO7RuOtDOGNI68vRguP1wS2szphW887TXvIWgdBZ8arQ8cbcq80ujBE/r4OZNwPiyVVC0ImWiN5WVhdgp6jtMA62CE7NJQaivhxf3eja98u7dz7y1F47jlbvhZ8bugVLP4DqsvyW4I1d2fIwN0484/xoQFU9xO/JqMBwp1TMXy/clc5lW3h4Cf/1P3vGDnXh6fuauNcLlmUvO3N0fmS5KWjJGG+IHbLLs44X3/hPNz7ibgWEc4Hri//TXiV99+58xfqJMGXUWzEgrPcu8aa53h7e49Q59sqRl4fG68DgXLjbRrx/5638WGPNley3OnrYabLmyl8o0K+ui2EPl2AdqDlzXeQObyNZcKLWRl8bL+crUz8xaOMvCLIWJRokL3kSmJhyGe4LN9GZlsJl02CYItSYWHJUjk95Q3Q2HYHC2UKwwiTDXwm1TbszC4W3lg7M8pszT5YTzkV23InXCm8xNXzj2jTUFXq57ujByDCs7F/FScG0hx8RD8dzakUdpdL2QbWLxjcccOdG4LPApeu52jkkb57WAj3TeMQnM60zoDvgcGWk0KpiEU6ETwxvfM18rzTgaPU07kmw1unMrJOO3oJNz+P1A1p5qO0y02LbiZEYAL45gOualsBwqrjPUWjfkqt3xcP7A3/3uA5/mL6n+A8b4bfoxeHK+5zEuvCwLhx2oRmqdwWeqZFR6Vu35eC2c14iayOgT/TgyrcKVe1QsQSrWC0E8MTWCzvzyeMfltHCpVw6+AoE11/8CFvMIS204lKFlBu3JRSligQHxkWwKVw+7/Qb8qVmJqVDtQMHRuiM+LHhfKMaxcsP1U4/EIzetskcZbcdqHRnlVFe0CY+rcKn3iEKfPL94t6dfQJJS1ZBcoKiSfELKVlcbbaX3hrPd8ZBnVHsG7SjNMgTDLuzJLXHKkVUOrEVxoYeaoU4b6lYM3gSaVvZeCJ2hFke1GddlOlnxaeWNc9zbRmLmuwCfxzue7B3rVCk48qz4OIAWXl46VrPnViqWGWMsq1VWDWTrMdbQdKIUJViH8wqSSG1FxENRgvF0BtYUKXXDf0uwCJ51jWA3jYq2tl3mGljrN5ofBScOMQKVDRhUG+I3CB7OMtXEwTlSzhzttjLxzWG1Aeb16woqQmoNazfS7La2q4gX5JVn0YzfJuEUnBSCFVpJFOlYRMjS6KzF1QLV0Gphlhd8mnFpYRw6LrUSmyG1Rs2R7joztjuMnSE4WvtjrwmiYH2jk4jYivUDLSrNJfAF2yqaZRvRtIDGfht3t4y9Nxj3TGuRyo7L+p4ffr5jOo3cDMKbIdG7BKay6MCSGsYUkqwM45k3l57hWujbjAmFtjeIwuI8I4bRVJw0nLUYsVTjmYxsIZnFs54yvu9RHFoE14/0vcHZRhcc93fK45qI84VSXpDW4ew7hErVRA0bHOVpesvvfiwcD5VUoOkdtXqGsXF7r2i70tkzdBll+3cKTvHaWPOVrh0YuEN0YCkdtRhydNvqQkdaK2iK2GGP0T3aHM6uWBkQWRhcwWnAyEiTiPMLroMujHRYVBOxbR9+1+WMr8Jh2OFtv4lQdKZqpBmwXpAmHPeJm/0jQ/dAS58o6fNmIrQOxFCrZU5wOkeeUqU2y7OC6To0bKnXkQ5Jm8q6pELOietqKRFuDyu/+voth/qCfX6kI2NK4alVnj4mnHYYccRswDS+Pk78P/4vZ4L7GSWgNnC5vLAfrpibiq6wxkBdtj8P+m1PKIUpXznKDc2OPE6Jc7nlUi2frx2Xa+V4u/Cn44y9OOS8YtbCYRAOfuDzqvzt50IYLe8Pe96UDle2wE/1PTpuUiVBaXmmP/5MdzjTp4k3rm0WMRUSgWvdM6+Zt8YgNWGJGEnbeLU2xqbc2Z5QDRw6Om83uJNZGbvEsGv8gsouzqiNpOGFFApL+R6xnxE7o8YgOOq8UmuB4Y6vdt9w071H1s+gC9qE0gw5ePpu4H53wyV1XNdMDDO2rzTn+Kl6/vtL4Mf1jhdRUonkWrgLje718G1oHDsYfeXFNUyGbJVSHJILd+uZYEe+m6643lKt55ozmZ7iHE33fLp6fnu5IS33/P408Kl5/GAYhki5WfHe4cXSquUSO/7m08TgvkDkBlF4WXb88OkNE1+i/oAxDtjCkcVue2NTdjzPgX3f0fQ1O6bguj3z9cB//mHgt4tHrePOGAbqJjjbH/ASONqRUVawWzbBWUPfLrz1C1+/KcS4gBNMf0DtjpfZEIsh5ok1ps06R+E2XHlrKqteXkVTjc+L8qkcETr2uwMmWeg9TrrNKtoN2C4gKthqWJ8t09QTaoeYijiD9z1RKoWEOE+tMDXPUxG0OPg4IzLz53sD2RCbJZWAHAbMuECbcVXYNRhMYbKWWAfiNDPahksdR28JUlBnsKbn93MFsyOXbTrW247ONMIwkit4DH1d8QawgncW7xM2fkLrila2CYEzqN/z9+stP58Cze1obquu+jZBntkNgb1ZOeSKrwJiMa7jUiFJhwW87bbJmGkkCliD9dsKx+kmYmp1W51YsRgMsbyCfCjEZcUNARS0NMQ6vBikZq61MoeNLVCbvh56t8yH2tc2kDgmreyc58ZszoG965hSwtiBopms5TVLFsiaUN2Ecd4HmgrV2I2oa0ZazQggdWE0joPpWIrQFYdIJtaVNTXwSmkTEEETwfnNSisbjVHVYJpilomhXTj0maurqPR/3MNASQ1hZnARMQXnHWIV0YVqZoo3zJ2COHLseFne8PLywiU6dvOFX/6ZxZgzpe34zW8Nj4+Gr9443uwbO1txpiAuECfBthU1EYJD0iO9H1ANdBb2ZqG5Cq7S6Y6bWBCzJaZzGJiqZVZLQUm6gPQ8PRyZ1WB2txjX0S6Z/tR4/z7y7m6hGzPjjaMURYsQc6RStmSuWCa3vZhbGfn7Hz4w3txyjY35ZUdJPcNQ+Ce/PPPXf6qY9gjaUBq2NrxteAHb95RF0TpS5j26enw19K8qT81HlrlnzgUzjOz2Bzp2uJagVagLfc10OXCVgbu7HmtnjC0YqVgbtgCmOFqb6P2K1ci6vmD9Ld70m3NdKlorGbfxsA14u2JlxphI8wPNWFKuWHPD+VOj1gviRm7uCx/PgakYOpR+CDjbMCVCXkG3k7AXSzU9Fc+UBLWVr2/f0g0d+eGKzSv/5uXKlQN731EKm9inLtzfZQ6HH6F+T1VDpXG4VfpxBOlppXAtPbkq1gqNBXrLcewZOqFOVy7V8Nup5/HaSE1ItWNKyvd/OPPll5ndBBRgOoFtoAHxd/wh3ZAuhhgu3Bhw9spSL3xqK7FWmg/0TjF+xfUvmPY7fL9Q6plUCt7aDTzVvcXKt4i+g+S3lLMWKtv7AxtwPsDQcW0FiYWORmgv3JhPfHmsvL98gS6GZKGMytC/4NbfgTxQTKKaDSkrWIK1VGayJkpz5HQgqUedI+L4OHU8rXty67adc4U+jIR2om8Xvug9N85yndkooyKIXbCy0DtH8QWbFFNXKAu1KKlVqoOFSqrg68SRlW87Rzk4zuz4+CjEV20xbuBTfcf//NmQYs/ndM+5CHZ64e2XJ6p5xIdlgxVNhaWMPMwfmPVIs1tgME4GrTesOlDINBOJaZM5LSmjKJ1NpKZ8PhWyVTTsyHLD6aHj4+OBH883LL7nTXDsMQwSSLWwyhVrOygZK41gR4wYYmtYlEEioa14u255jFqorAzWozHRmcjd/YZLjnki1MixcwTrWWvlpEceznuWGEjiWGtm58AJBGewAsnAT0uAZHh4iKzTjJPwagMFrdDaTIlCVaGshSoO63ZoFqobKa3ndz9/4ptfegYn5DjwKR14ODt8c7xxllu5EPLEQIe3A60INa7gIibd4JJB2xWRxI3zvC3Ci4k0hI5MpxFXK/sKg7WYumD1So1QRch4punC284gLW4jeyrGj9TseDofWey7DQMeMsqFXGeMM5uYiYyRmSEZTBW823GlkaQnqcWpbLhna5hawtA2nLQtKA0DlFSoWhHrN88BUFvEmo7RgK6V0PdUlFYM1igWQzSWx9q4MYoYwRsDpdL5gGmN1sAA1TjmZlBvsFno1DK47ZIGEZW6ycgs1JpRMWitkJX2uj5LGB5r5SsVjLzq7NuENx3VGSwOayCKEGWrPhapWN9hcWgpNGOppqOsDUxFqmBqxbaE1bZJjv4rH/P/1YeBVguQGELDSENb2n6YrmI7YRW4PCeWbPjNg+fvnjpS/gUa7pCnEz/XK/fHF3AXFnPhV982fnlckGwxTXFSUQkECSCQidtYxkbqfgbTsUgEO7GTzTtuxGHNO1rnkH2mdo1cE7pODFT2RinlC57Pf8JTFF4eGs33dMYTaLx7eeb/8M8qwazYsWBKoc4GYeAcO65uYDUDCUMAxFnmJrDutuAGB070nK6N8nvhT74804e20eW0w0mj5QuZFQrEMvL5521EXSJQLIP3GO1ZpsKnxyO//tRzzYF/8qcDf3FfkZKxHZANkio77/lkMsbNlPZCrRPGAkXoXEfnGsErOSVymmgVliSYfkdpN8TakcnQHBXLQESDMPqA0mOcxzmPpbDOSsqFrne4MPHn/2Ti4d+eWOc7Qn+HI9ObippKbhFRT2c8tkUWbSwlkEtj1QWZJ7wmDv7INResh9orF40IBucMnTfcvOlAKlXNZh1shSwCsWItZHFEZXsDmEiVFWssFn3V9L5w9l/wffsFl+uArZGqiSX0PJ4M9a3S1hesdDA4NC0IiWyEGXiJZ7z9Pf/8YPjF24W/3M/0JbKuSm2OiifrZ7RVGpWcr2Qt5LbVjMyY2XUP+HqA9Y5aDsxaWYmoZMQr4m7ou5sN8GIyO3fmID9xtN9z7D/SiSHngapfcm5beLSzG9lRzY7WHK1C0w61nqqNpHs+nY9cJiXVA83dUYwh6sjT0rPoQOg6rGS6sGxwMFnpPZSS+PoovKxKjYK1G6nx9pA42ELy4MRR20xqgSUrEwWxht7Buq7sakKb0vkByEzTTC63GAqdFRrCJff8NN9y9HuMGQDLZYl8uji0sxhTsbmhL5GuXVFrWLLnZd0TcwBjcdYSfGPQxMVO+MMO5mfedzNvhgu/2D3wof+I8InJVJ70Db/72PHzY88p7dHuFmcU1woHU5FUUGvIJdKZC1EqqRnyKsTm0dYIQ08fCvVyRlOlaSW2lZVCFk9wHskJ2xrSW5wz2HWGXDAYBuuY4pH1ebutOm8xLZPTwn4/MvpGTo3H2Ph/fSe0dEMrPUfX84bInd9ES9EFrqWwSkCkp2sJjGOqsgGnjKfZgSkuPM9XbvtCKoXaLOe62w7caun9iqlPGN1h1COiaIq0lpBkyRmEK0YaIoZbo9R2IZtAkPIaQk5oOdH1FUnPoCt0B5L0PCx7npeBxxD40itHPXP0yjV7zkuHmI7eGzrrEM00KfhgAAEqSQurK9waS5gbrsINjlNdSbZDHTSvZCrNO7LN2yFAQFyh5YyahJWtqmesxRvZLhfVcbfrOCjUeYUQiNYQpZGoFGtZtNE1xRthZ7cKbsNuEiC7TRusCClHMmnDB7eGlYHSBGMywShrvqC5IghNC1oS3jh8CES2sHlDSN6S6kgtUCXTQqVqJCaFqhQLRQ0lJ4IzpJyw3iGiNLtZclXCVl1vYWv61LpJkIyj/dcBCP9XBAizYRw91lRqTdSUUNnh94EmCfGBZHs+xx4b3pPaBzKClo3O9sNDZhxGbvuP3H75wK1puHRLikeaJsQXWlNsdfTGEqyS84SKYENG9MTARDWFi3pqc+SSOJUL9SD0wyM3Q6bzC78MZQtbnN9w+U5gudlQpjXSYiV7IUpPewo8nARrZ3JNZAdtt2fVnk9iePKWXBxUQVSoLVJM5WAdQy08mcQTI4mR5+uBuDqcTYRuQNRTayatFWs6Gu/49e+PPD6+x5k9tq0Y16FOiHrDo77hbx57fvNp4FJ6Vp/5cgft/EhVh6SRCUHF41SgbLso1Y1aJ6Yg9UJnLdTpdTyXtmCeNErzXOeBrI5q0+thphFCYFoWrBmw4hC3hZ3WJFxiQTtImpF25ub2xN0x83QKJECKQi4c9h1SK5SBJTk0CFUiiNCbrQtsNDN/+gHiGZHGrTvw4cbx8aHQhVtubE+WRBeeKalukKW2naJVLeepUsyZzo9IkO3783EDOalH82b7OuwcQzVgDMHukVrR5sjF8TJ7nuZA8IU8KVn2lDYyfbryB1t491cj/+zPAn03oemZsTzwvi7YnMFsmuFYOiYxlGwI/YCRPS1vOObWKuu6bETHFqH2tNgT00wZwisZbkcutwSFf/rmI9lf2HdPjDxi2wNZn6jsMaPj8TxwXTaDWRPojh3GVQw71usV6XrwHcqe0+krfni6I7u32K6jCxbrO9KsiLfcjzu8M3hWTG0Qr+y8pbOBp1NGXKPznlYOaFFsWfgQPH95GznPE2sR8tKY23l7+IS3xBYQEeY10dVGK4lmBx6XystlBRaCMVAz9rXCuETYpws7L0TxrBjO14Y4wUrDSuN2bBz8TJTKLo+Ui+GUhN4UjqFhRscpKuQZNyZ+8eHC1/1v6fkJaxKWCcwVkR2n+cDHp47n2BEl4DRzCI5di/RtodZIEbOpYJsheEtchFKUnAzzWrjkRnm/BYxrrojtAYNaR8VQyQRTMa2+htfK1vOuZWtIZcfHR88SA64zdNqA1xS+jfQ2k9JMrYnLstJcj6fjoAZTN3iOjDf8uPZctMN3PVYbQxgYjKNWx9Ua1A7E4kj2LT9eG+/sE7ZvW3ODgBrPtSZOufLWNg6tMtdEqwuyTmQ5Ia5CtTQarRlAWUvCDZ6iadux1wZVsGWlrmcMK806przj08vI5+XAEnbQCp9l5Bd2x5dtQiYll57BJ5ybMLoipqIWtm6CgkBxhktV7kzZKJJxpSfg2ZFdQKzSUFQKKptYrFCxulFwlYQzjWAdxnqWmFAJeGtRqdASlsboDaoGbx2qBbUgRraVjamb6wBHsDtmVYI1jK7bQp/ayFVJpTDUlSErUpUSNr1x1+DG9Cz1TDUFMf+4LmyoGmiyXXRUyDKwNkciUG0BEcQ0ikRG63Ba6a1hrkrKinhPUNj7ntaEo9vT5i2YqO6O5k8YPGrs1sAzf+TJgIjdoC91C5lWqawtE5NlDG/AfMHH5z3XNWxErK7Hh8z9/sRffnXmzfgHbg/fYeyvwV8gDZR6JKWAdXVzLpsRWwMqihfF4lhKQ0bHMELMhWQUQ8AkB3WimoQ5OiR+QjkjZkJCxocDXdnxWGbEr3x1hG87T8qel6XxXDz90BHEvD4sJggG11mWOZHCTEB4p5FWFRobYa2Hf3o4so+RJ4GnxfP71WyVGjvS9Xt8dyBFYZ4vTPMNTd7x6x9u+fvf3JD9G479Ad9OdLZiH2H3+CU//ij8/kfLLDvOSXmaM+dLYf5DoyzgOo8tt2hcca1weejIy1u6kBG5oFypGlnzitUECMGNxDRjjVCyJy47/EFxYUFtxmjhPF2hQamOw3hE64nKQmkd1ijFFowRRCC1lf1tw3yEuUFuFrtW3CAcb3ps7VifM1k3+lZXEoNRLIl6OhFenqnxhRp6bNcR8oA/jORWGMcd4nuQz8TiUGsJYUBzxVqHVegHu/kcukqZKyWfaFo4jPd0psMbg4iQmgPtqBW64qhqKa0xpZHffr5B9j3Xp8YaNz94izseW+Bd77k7XDEslDhjcgYNNIHCjJNEY6K0gRIroW/Y0GFr22yNtm0jXytglVbS1pzR/cZUl4m1Cqnc8qYmvuo+spSfIJ0xdUJ13uqtIhRX+cPzhVQdxjcGv+PbN7c4v9WV7C6QWiWr5Rpv+cPPN8zlhioeU4SpCsslMaeNfz4vCnXG6EIvK3tXGTDUYlkkbCN/VebWMDXgUsfTx5/54s8mfjX8RMsr1ni8u+HHlx3//Xd7fjNZbA/FejIWWmNNhWsrFK14l7GSUDWUZimyVfRuh44ohlNV1ARSC9As0iq6TIyi5JowVHoS9+OB0XYbG75t1EyjHXWFXJ+5f/PCnfwGLSfUeKpkcEJMA48nT9QR8QZaZVkT9+OOu1EI5ZVHL4kmnqojS6yUZkg50nTAcssPDwv/U7vw1++PBHd5vRUbjBOMNYgYrDUYjZjWqGnd2kbGkDXwu58P/MOjgf1AEGU0FXRDsi/XF7pdh5QrXrdQ39wynbEMNTPqFg7OYjg3x9kfMM5hW+HWbhhhsmH0By5VcM7T1HAh8YfrifubRrYbDwRjyRx5yntu/RlXJg7SWGVmNqCqNK6UYUdOG0TJiSEZz0sbWX3AYvC2x+rM0Z9YmMEETuXAT8uRSxzx4x3Welad+ZgtuQbQRpihVo93kSYr3nU0o1QqlbYFSK0lamPynnNJvB8sPlf6lvAkvAyoVKxTvG6rRWPtdvHSRGfAq+DYcmWmwU3XMxWlqd1qmc2x1EI3ggdKa3jrAYMFjBTUFKyx5CjEuj3gm1VS2WBGqTVqEa5F6dPKrjq+soFHNZxkoCLgKsYpKUe02S3jYtnYKVnpvcc1SzaQuw3KdpaO8zajwHqls419XanAj7UyWwMqDN5irYOk2FK5zZ6DgaqGWaEGT3GyNSzsHzlAuPOCrVDaQLMH7G7P9TTy+NGCPfJce3543mGCoXMv3Lorf/Grhb/8xSd24dfgPpH1M80tYAwmHDkvwpQHrAwE2SPqMcajOeIVTOlZViWqYbhJdE6gQV9O3HcVe2jsQrf18OsTSSNZIjXPGKMMrnD/1vNGI02euSTHsu5Y8oHUKt/eGG5t3NjRWFy/p4jj/gj3N4nOWbxJpHViuV43yp4d2Z9WuiTc7OBfv7nj0/lCbjMtW04vkIoSS4+wI+WRp0fP9z/ueWx3tEW5GRvOOk6Xhf9pDhDukGUgB7jkxqxKqo10vlCnRp5WYjMIgUZFsLx8HPi7f3/kl39dePdeoaXN5NZvwpNKfK1leVLZk+o9l7nj/pjp7ESTbTwv5pbWLHF2iN7RjTfYsMD1I14yoe+gFpraTSvNmePNmZfLa9VHHS/zyod74W0X+XLneXoQytIoyaKuUViplzNjXjGtAR1l3ZHjHumPpFw5t0pvG6epsq491rPlG6RDiqUPPVJXpvlpu7mpwZiGN3DTQ+8E1Z4ljvy7v4efHiph2chszRoMI7kFvvvpzKU9sTMjWj21zkhzqB85PdYNpS0BcXcQ9mhaMHLGjYWaDSu67e9qROqKIVFFsM6BlP//m8ploj2zmErmFq09KRumBtPquacwthOmpo2FKQYMGGsR23PNWxixiN0Oq/OCsz22RlordK5jkB3UA2F6z9dyy603FFtpUsBYrjWydIHiC0P/iUM44WQhyIV9OOFlZp7h2N/RT/f8w3PPufUbDtXf89Pjz1xOLxwOV0RXvDgckaN9gy537LtbvAFxA0UOiGGzvMmACSOxxE2o43tKapR8Ydg1tBSmvJDMnhVDbJ5pMuxFcDXQNbZKmGZaW7BmATypNvR1HXPKhiQDvnjIigyVJnbLZ4gjlp7f/rjnx88dE4YWLOIClcz5WnG7gURk6ixn8TykG357CjxFBw4OoccyktixmBv+41PP5/WJv7zLvN+tIG0z1UmF4FFNmBbxuq0UndlupQ/nnr9/3nPxA6HLaN2kM6K6Bc3iSggVaxKuJHxz7J1nrOlVftMwdWMyeCOIkU0WZAJZBRcSnWs0DJottVaqNsR4rnJEqMROaSmS1aLG8Gj2HMcDRs8s5YXi4EdvqLEnVU9pHaP2vDWNvVSq67nYG87isdbhqiGYLTzYW6XmzPMyssiRcOjppaBloWpmVvgcG7dDx9udRYpiWsWIUs0C0sgtbTd2rUhVVIRsLQuerJUhKD41PBFjE0YMqWaMFKxsro9gXkVYtRIUvBqSOhqOgMF7oTahWEcxjmqFK4WuZdRtoUdVsMaBKtc6EasSCcQEpXQYK9T2SNBEVZAmm5HTe7ITqDNfWMfbbDk3x6UUjB/o1FHVQ17oek+raRMXqBJwrOvKfLCcTeSFkUkrRSNHt114u3TCGji6HS94VAJZLFNOaBGm65VDrFhNSJ0pohSBarbJyWY2+CMeBnobkepY5sCsX/DTg+HHj3dcL45hf8t5SlRX+Zd/6viz9yd+8fbEYfwelR/I7YHcLiiZWhrWb5rM69p4mJTpEazfEQ6eb44dY1X65ijaYaOnuBVzl1FJ9CFjykTlCmGh7/aYFkjmmWwazbUtb9CUzr4GHldotmeODm8OHG7ecbO759v7B7ickDYx9Dfk1THFiev1RLezGLU4V7By5f7O0NkOux4Q/x66PYjjF+8t/+fjl3w+DfS2UueeUAd6847PD5XnqbG7/cBXx46TVt7cWf7y/ZmBK9fJ8v/5/cxz3bFLM50dqbJnqpXSFjSecabA7si57jhVQ6lXbFnIpefxxyPIwtf3B3J+oVaHC9vOqrZMbCvOveXj47f88N0R5wfefLHg9RM1r7QKe/+BlIRpUZoOXKvHeEswZ0S2amiTuhVh7JVfvP+BtzcX4jrw08fAOX1BcPBuKNzWC6V05FnQWYgtkTrHKStj53EB8B1nO5LNAdE9fvGIgRde+OogfPPVwEhA6kiqhSaGYI/UCap9QesGylEKvjPs+0BvFUvmMo/8h3848Hf/+T2JO6xZwfdbqMsrUhqntEPqHdlvaJScoeXNNtafIecBVUhJsHXkTd8jkhG7UDpLiwutLWSnTEVYlwuIMIQOIRAcWBqlveD6F6bzzDm9x7meZCxJLKmNWL9gixKqQbxj0S2lLNoT13t+Pg3kblP3qmwPJm1n1Gy5hXUdieVAXHbMj5mhLAyu4YdblikiGnnfMisdcpj5xVe/4cP+bzHtAdWZRtzUuqWDeqTqL/ndp57/59/s+HWzqPNUHfj42Pj22AFKzQCep5OhhhF9RRMvpZLMAefYtNBRyEZopiHlmeB6+lKx5plf9p5xmsjuyIte6UXwLkPpqFkwVdg5z8rKYgLzesNP5yMXvee0box4EWUyjsUph2L5+PiJN186xAqpKnN0PF56Xl6OqO4Izm1MDVtJolyWTC0Dq97x3Qp/OMNTPHJmt5klWyWvkfceXM04G8julp+qYTk5/kW/cJTnjQuRC6XBMfRoXP6L+c86z6fPmd/85Ch+wPtCJTO7Qm4FayHWRNGMzhNf+40foQIrBhsc67zQBLwqPk7cyJGFytws1gdKXqgpspOGKZ/xZkdy2/6bBlFu0Hyh+g2uYzSiJrO6zG/XzGHfqLISc+YsPatAKo3gHKgHGmusdIeeoB5h2IBTPbSSWeyRqyqpRUowqC4YW6DC4D2xVBZRioeHtOCtIHuLLpGmgdoi3inBFlJLQKVVtte4KLM0JiAExWjDtEKVQtLGwXliLQRvqG3lulxx1iFasC2iGJIdOLd/9A0onTRKTGA6jHqaCqlVTFBqLpu4rTRSzsxxy0VYkziIsEzbKsu5E61esVgKHZdWaGRWaTRR7iTzvhupF8NiBpoxOOeQYjHSUdaFO99hAOyOisGWyHk+YYcDaS0oA4ZGXp7pnBDqgpZK7wydNcylUNZKM42cEq0J3dDh5rzZgMVSlf9imRTzR4YODT20HJimnt/8fuH3P3as9Q3VWK6XSs6N4Gb+4oPw7c1HOvMRw0RqFvEDzoKow+SF1hpVlc4LhzFx2HliPjMl+HS654OO2FapDVSHrccZXhDbNhKZW2iSMcEismJcQ8a0jVDN5nrGGDhXclzZXp+FvttxG3ak3Ugk4WqkZQ+6Q9MA9FiB/WHAhkbXGWyd0TxTi+GaE7YYqmkERk5T4OQyo5v54nagRMfBvcUaT4uFFA0yWUy58m3niX7mzeGR/+abZw6hEtc9h77j+VK5CT0tZf7HP1z5TTsQMbRiWYqjNMPSAi/FYNsOvy5UWejejJTTnnqd8a7HEOhNxUgk+I7oYS03/O53Ry6nL7m59eT1J3xncWowxhC0YNtAKQGNA2kZYVTeHHtKXrHOIaHR0oSW7wjD97DrMdwS9Bt++LTjuBu5NQkXF9boqGmgWye6mnm8GuJ+z/5uwOQeKVvwrajFtMKoM2/vM3/2zcBf/pnjX/xqI4vltae2QnEePx6oJW0AFdvTCliTN6rcP8J3/I7zfMdvfviKl/Utve3JJjLXQOwGrlVxdqT2lqKeqySaUbQesemEMwvvQyZcFEwmyC2Xl8jTrnKzd9uhyET8rjJowg8KWK7Xtk3DjMGJxxtHjTNODP1eiJ9WPhawF0eJAylYWjNUyytLIqDdLS/lHQ9L4DK94fP5lpd0A/3WADGidC3wdDlQ+YLz1XKa91zzjiY7hqwMy4wnksQQsxLSgi2VvDqKndn/8om9/w0lPlDZXOgYCzJu2maFD3LPL8N7Lt2Bs2563pdXArcYpcnI0ynwcNpj3Q6hJ1alMNCKElrFqTDUSsmP7DrDrrtwCEK3hw995v/2JuAfX5j6xq8LfIw9wRZuAc4F61eCYQuG9W/5/uMdv365Zwp3zNIhWjEktEZUZ0QrP5wG9v09u50jJVgX5Zp6TL/b6l31NfQshgLUtpEA0R3Tx0ZOgXEYaQLRQRS7HXJqpLcga0OtxXQ90b7h4/zAYWc2up8qNWXc0NPEkuMVYyyldPz80PFSRugywUE1lVkTq23YpqhTKg1bEorhZvA4tUTrOIuljXv6VHhPwmul0xWjPWoDiSvWFmpd6WyGVGkaiWKhc9s4GWGwQpZIricwBSsFZOFaYT2v7EbHWlZMXRHpcOJwGEITnHbEtNDiyvFwx1wsDgdpoZBQK6wVUq1YD5BZ6wWtm5wI21NLppjKlcJjLQzBUUsj14x1Qs4rIg1eH6rWBcQYFJha4irK6CpgkZRR3W69U14YPay1oK3Qdz2ijb2x9EVZU0LVIXaPNkVSppPG3kHOhVTPaFvQmDa/QDCEocfXlTid2TkLJDofEH3kpg+sqVJqIgRo4knWsEyJ0qD5QKUx1IyaE4fxhofSmJtuIiTZYE0iBq2ZG+tBhEkMiw2sLWPqNold84SzQufAacbbjtwaDaWzhmo9wRqGILTU6INBSsTQEBVwA00DznQYGzDZ/HEPA6iltEq3u9lCG2UB/ZlfvD8QXOLl9MhxrNz7ho0dpY00b1E/kNpCa2daq4gNaEvknOnCytvjCW8LjZU537As31Ivdyx1pdWJXB20jDMJZaG2CFoQZ9CqYCDpBC5R6ys2VDOWQFwqWhrGGkxQnBc0Ci8Pz1QPFwpe7qjLTE09efDQBVzncAGG3lDObasb1kwzFe8LeOEpOX5+aVxlIvpNeP35xWAOA31LuDXSp4JtlroapmS4NY5fvffcdj9t0JjQ8d/+8y/R9iVyjXAtvLn5wL/7HzPOdVzygWuGJjtmE1DdgoGhN/hlwSwT/e7Icpk5HG5wLJBnjKt453DOsZwOLNcbkEBOSkqO9voCBUWsp7k91e24XhxLdngZkWOPdx3VKNISooXWItIy1q7UBt58Q98cexRJEykpj5Mh10AXZ1x6Qd0t5+ToXODNzYEQQWPhfhfx48I3dx1//t4x9pn9XeJ2B8QdkjM5RlZ65qUj9HtECrF8IsjC8TBsJk2nWK20qPz4G7N9r3ag0qG7N1yuM6sE1ibUWAm+UqrH2+31PHSZ/+M/Hfirr1duuyc6nSlLZo6P1PN7Pj8K+ieB/hAQP1FZ8b4hslLzjLOGKWa8W7kbj2iGtCoLkeifWH1Ew7YycnYkqaAtM8cTxRuSP/Ldes+/+dTz4/Kec70j6QhWCLIS7MTBRrp54eGTJ5t3LIxcSiDbHdZanImMudJiY86V6APS3IZQbYH5mhEs0hSRgVqU2izOCTSHCwMyO8rS8KzszUysliyByxoQt8M6w+W648fTLWfe441nqIkiBVMnXDszuInRrvzqoLw5VO7DzLtD5O5gqc3h4z1fPlxx5oXGhVtRPu5uaW/vcZI36VKZaaFwyQf+5rPl46nH9T07s3LjrngiWi4MoXCdzkTbM+me//DTB1wXcMHgnCWp4Ye58iTbAakoxFaITbnveqwars8XxnrgTivrciGEjrNTslWmOtE089VgCGsGOrzr6LtbhIjxJ2jTJoTSzU5pXEOrcL1aXq6W5zoSg9taJK2AAWFDnRejGAQxGev95qM3gjGBtXqeq+XFeBZvMe3EexsJ+YLFYXS3jbzJFBNZy0zTtjFd3IDKVl/cWviJNTeMKwgrJc0oBawSS0aKktPKrjtulVLTGGxhVxstC1V75qVSj3Ub5WcYrONaZ6prBN9RpZLbsoUPNYLbyHuxXSnOUMWwYDi1wrW8oMOBWlZq2yp9zgkBiGulaXkNZRqa67jqyp6CHbf/iwHjHZtprEITrOtAoDOwZ6F/naaUPCNtv9WTJTOuC768YN22yj3rlhnx3qNWSemBzkUON9vPxbuKqy/4FkENt+Oe1DY3QywrSQ1r23Iv1noaFW1KjVd2uxFThKY9YiyieaMPdntaXvFWsF64FOVsBtQGpKxbmF62z2d5zUA4p8TWc61CUkPwW4aqpE2UHozZEPF1ATUU05HajqYDjh4tf+Q1QS2G3e0N4i8cjwP/6jbQ8cTo/gZrJ/yfW8Z+ZJ8OXH7ecVkTyTxx8+0Nhzf31PxEbY1YM6IVGjhTaPWE0QUbErvBQAlM60hUByawNIdOlVa3ypkRQTSQ0oIJG9lJK9TcwBporw9/6chVsDsBo9jeMM2Fhx+eWepI8Y6fk+Dub3DVEFfDPFvkKNwfO1zfsKq0UjFqaE1Qb4minNfC80l5SR3ROCLKSuYhjZxy5ZvhzGG94kvFtUa1lSkpaynM10R9mzEq26goR0yaIXbQHG+Hhf/tW899b1nkyM/lCWt/4ssvDvzV0RFi4/r5E48PhofF8fEkhB/u+ZNfWA7+xHr5A7fvPOISxgZs2zMOBrorb29HtBbUjMw6YeUNP//4hp9+2rHmHbbbU2WPphd24z1D1xDJdF3CWEepDpWGMZDywOUUMKXh8/azuDTheQb3CtbJrbGK4Rode+O4/ZOO/bWwN8LzsFCM2dzeFS6TEm3m6y8Eb/YIjRgT03Lg+cmTRdi9e8vtu8j9XWbnMlY8SkXVcnruOb0cscGCzUxrIWrF9hnbJboc2TnlV+9vefl0wXYDU8r8b/7U8q/+/JFj+IykBNkjCL40rqfC83XPDw+Rb/+bN7z/MuN8BJvQuvEtdruBvGwhqNIyne1xfiCWSMmRzEQICz07rBU0N+Y88JwCzy6QimyrC3PLzeHNtjaQFcuFfXjg3WFmzBEvwlB7rmujmRkJnlZX7rAM8YKNiaVst9kXKiGYTekalWuqrOuAjjuolfDqoEC2w6CnA/ZcLx3WCLuuEnPjHAOP11u+PwmjO/PD54FP5QMXs8NootdHgl348jjxV+8W7sMJkSv9kBntFVMnglP64UDkAynukRY2zHB8IsRKTVdSH1nfJPyuR0vjIo5/+Nzz+er59l1mt/sdb8eZg3+mtmfEFUZvKGvg33x/4N8+fcFTuUNsj7UFrOOUGx9rZqJRTdskWFbItWC6wKk0SovkAMFv/IpGwsqAcYJ1SumVp+kjh7DjKELvwrbqjco0Fcad0NRxvTZKaniEOXc8XTuWPDDbQLNb9RbbaIDDomoBxbZKsB07GkEKuTSK92gRjHUkMVy853M23ATLIJn7OkOF4nu0JrJEFlO26R1KrJHEnooF17i0jFpLbhfue0cpE0jhSmNukVYsIpbmOvrQo8vKvi3c9A6pbeMGmI5TWsluSy05CjvXuGoG2URtqoEpvuBdIWrcOC/SqK+3/NxgkdcDkCyotyB+C/DKNv0aRr9BshSadSQK52bYBU8wDX+0aCwIAecEayNG7UZ6bIWdDdh5xZQztkZcU1waMHT0stK3lYAFk9HBIqlHRakeUosYHM6lDfftLFG3w41rF6wJIBYxA3UtoB2tJKwdUOtRta84+YbB0rVCz0YMbChaK6HryVXIsn1vOi/MdExiqa1uyJN6BWvwYjc+jSaMCNkMnLOl4tFUsVYwaqE0vBp2Uuhto4iS3J5rPtLaDqLZchB/zMNAc+mVhqQcw8CxO/HhkNH5kaIX1BdqG5hiYLq+5/FpYNa3/PCy8hf//Jb9zQFrG7Y0iiriOspasXWh5gXRQnVX1njlMX/LZTXklFAtdOOO/ae3vLmf8e4JZMYbQyozrTWc2aQfog0xBqyQEzQsdJsyOGWYnjK69nhjULU8nxUxhvdeoBrS2pONMuYI+QkxDdVl281IIBfPkjqefhbKGsCNVFXm2pgMrNLz+dlwLT1/6TzH+oRrGezlFQrUsa7bDcGIQzQgNoDrwXRUU+hvnvm//iuDKQVdX/jiqwtvdws78zOyGlgs9rbn8YuRf/dzx08fF+ofRrT7in14x/t3PUfb0DRtyVVj+eWf9gwHT02f6YiUpSfO7zk9H/nuN3um8hX4nvw007mMMZ7fmvd88eEdWj/z9Yd5ayia7UOqs4HTxyPzT27D7Y4bLjlpTxaP2x24nAKdG7gWx7ValrVx+AC7MXOolny+cDUjhPfMOvBiEjo/MS2FnRFEbokvhdMPnssjLASePhre/3f33Oxm0AlqoxbDvL7ht7/+ksfzPVMthHzhF18GvnoPhy7Te2FNW9XSlIXlaFipqL/hz79YOO4CxJ5WHC0njHE4t6PkkaUdmMot//7fPvAXtfDNNxFvryiOohbxwk2/x7hGy5sJzYcOp7Am8P7C2+NHbs1IqI6np5XPMZP8DXN3j5rM/S6ws0LVH0GeQR8I9pndeMKw8vnHwBK/oeR3eOnx4uk6z40k7uIDNm+66yZHVB1JBDd2mJqxxVFT4PNnzzdv77DBkNNKCIFqCkYdvd2zthvmZUdtyugyqgW6gTl94H/4m8Q/+1PHagreZ4K+8Cd3PW8PL7wdn7j1n9m1T4TX/r36lSAdLQNl+9CydMSUiAjOVEQrjkLF8eufZ56mwK6/x7bAOXd8P3fcv4V//Rcv3IWfkPwDS34h1kiuEdsqw/6Gf/r+L/mPP90xyy3WH1FWLtPKKSWSMyj5tRJnKGScUYxrfDSe52BYiZSy4HuPMRZ1cYP8WEumMoWEK5lf9g5ZK5OpTMB/fhk41p7ztVBlh7lWjAZiW8F53NBBKqhpJCs0Y1AEZ8OGJMfgRBlL4su+0LVCNtsOewg9a654ZxBrONnAH2qjs4p1ytjgmpbtgOMs56LsBo9/pYGuRTHe0Iyl0TGXrV1Ba/z5bs/TtPI365kmkI3DuoE1J970e3b7W/q04PSMHweiPdLMzYZVNgnlCdw2jnaq5Kqb06VkOutoedq0uXar7c3LhTF0eDaaINaSW94Qva1u60oRgrVYbXhtYN2GHhaPimc2mdQqSGM0hVwjQwE7P1PSC7iBkhNRE4YrziaisvFUuhWa4GUkq6PYytwWZmu4BkhksMJUF/ZdQZgQXan59ffXmSAVpxnVhSyWax0gGnrTqHbziTSVrXJpLGI9miJBBS+BuTaoBVMLUjw1KdEqRldcB6O3RASq0NSzpIlu2LNXy46GlAIuoKVD/ECwMPhAva7sXaArgk8njEboArEMXLUjI+woDPyR2wRGoQue1CqX2ZCT581esD6gxaMtI84Sm3JdEzVbjL8hrm/47e+Ur799yzhWxIB9rf3ki8WXQnAWWkGqMs8Tn9fMw0lotdsIfs0x//YL3rxUfvFFx93xZ7ATzow0zVQS4nqsNISKwVPiW2o6bCd525GmQi0GE/zremGDHL3MO/rOYkqiaUdcGtdTwnYrfecw6lAamB1T8lzWQKo7MIGmSpXt1LtmWEzFieenS+Hd3YHDUPFtpkneGAc4am2b4EQ3Dam2AppofkSdwfsr3x5OUE+k5YEWLwwGWIQ431OXNxhvOLy75c927xluhLwk1u8SZxt4/Bz48MUdlp/I2XCeM4VHapmw7orTyumz5Tp9w8OPHbYcue161AWyEagRXRo//6dGfLrhcCd8ef97QtcwXsAIzoxM53tqO+L6iiPS6kSJmTXeck2FYI6YWHmoHacqfGOU/U7omuP6sLA+RS6hww4jV0YeRQjVkeaGs0AcqA+Z7mHhvgkvMvB8GdD0GrQqbWMllB2//+49v/7hS851D83wJ7/I/O//aWXoLlgpTOcrs2bm5Ci6Y+hHjseOZAZsKIg50KRQdEJbI655q9D6Hp0SVi1rvuO7P2S++nDCmBdUhVKF2hqEbedpMVSnhDAyhpH1uTAMZ267z9ykN/Dk4TRRY8fLcMese2y/YN3Mwb7g7Uc68yOm/QhyotlI8Lc8mG9JuscJ9H5l8I7Myhc7wby8oPGCykaUtCWDGqzdYYlYXWmqXC+enAJ9F7C1UO0/vv3vCO5LUtnjrCdUR0iZWE5Y3aN2x88vR/5FfuFffWWwSXmJlXdfPRPkH3D1AVcjvVZMs0BjzQlsR/AHSoVcuk3yEw3VvEqpaCQHJw38+Hng774faF4JbkAJhNs99+Ezx/6Rrj2T80JdN2ucUWhpJZXA21H5wgu/XwVqj1RLLULvPVqvGGtALbklnBU8Fefhc818IlOcYoPBSeQoSqhnjtITE9S+J4XGuiYwZ7zxSFMmFdZ65OEl44Mj1YgzAckC1nEY2MbtwVDy1kWvbjs8rkXxIvTW4xp0teBiIoyBa1GupVHbysEfWGqitsxq4MX0OE00pzTfcPueaZopLeFF2FPxrW1CJHHMubGqYLtNMFcR5uXKbZ+5sYbvUuIlCFNa8Rg0F7RbkW6PiiXVlaqZcHuDlo5cBKMbeK4yb6Y/42gx4a3HmUrMC1UrKg01INYQnMO0RrB+E8QhpJJpuTLSYfOC9Q4/9nB9wQLZrUQXaDZQQselQMBhy4Wb9UpuDkkrQiEZWJp5XX1WvLPbgaspsQjqLDUqpm3OjNwuXGQjy0ZpW4hWG1HB10rs9ljtmPPMpVQcA8nB4LbDyaX0rC1wi+CqUrSxOou+rosaBsI2UfYYvFQ67zFF6XXi/W7EEtkLiPfYzrHUxFq3RoWo4v2m+R6CxaeCAClXjARidtQcCTXRO09dG9IUW7Zf1VhK6FmkY82bhVfrH3lNYEwGW/l0jrysW+Xlw91myluzQWRAy448B86LbuNoGviOZbrj6dM76gcYxq1zKtHjqt/qPnbFWkG0p+bAvCRyNYgaamM7SV2PnKJjbiN/bleGccYExco/2qostihWOyjvKNPXiL5FjZDqJpOoRimqxGZYEUopDNYQY4/DU61FWWhVCXhaM6j2rx/8Bme3kZOxjmaEVvW1x2mwKpDWTVVqVh7XmQ93jr45NldRo7VtL2qq3VK3kihFOT1krtOMOwbCodCZK1bOiIuIGqbYcKZjSjecXr4gEzGT4I+B0RaQSBXdbuHngctLYH/ryGq5LInSLtTywuEoDGbP9GnP5XxPK46xGyitsK6V0hRnNxrkobuhnRxLaxjdEbyjlYxIt+Ur5Ba/vwF7pRKxkjHOsurESylYe4/Kjp+L8FIS+4PB24biqa0iZkWq43lNXF3PpRl2amnF0IqQLhW7Nm5co5RKaoVzs0BPM1tYpySl5IHrPHBKPSlm3r478IuvhLH/x3Syoq1nvgixDJjxBu93rFGZWHk7KlktrQlrSti6JXGXXCia6bTRciWGA6Xu0BpAC6hg6WkIrWaaKGKVKUay2bDDxhWCTVBemB5/Qn4akUkQNWRzonMzffdM5xbEzqAfkfYjpT1STUTRbS0jjne3nr2cuBuV71bDx5hZr0KXBJqlamNJkdlUkipTKoy1UlulSqUfB0z12OpRc8taYS4eqbfc8g4XI3ej8FnBVEOTV15VtpTWs/eNe/uIDwN7OxDaBeMia2oM/hYXE6ozxireOMRa8ipUdizXkSWB63aUUsmymUlnu+fn3wbm0wD+HrLldE6ggssL50OliDBhMeEttuwp6XmD5IhjnQPLEvnTd4F/85vIwi0Bz2Ab2USsGbnkE32weCxFC5bGmldyYWNydB1NK6Y1XF22vI9TFjtybpWEoXnH1STubxykynxWsusJw8hKJPSBdZ3p+z2igh8MOV6wUjl4RWi8xAhhRzMGKw1K3DDHVqHolmkQy9wKzipWV/ZuZE6V2grVecQ4EnDNM9iKHR01Wbra6EzF1Yh42HvH5Zrp3IARg5oNcINYTF4YaubPx56neaJ0hmaFWApTuuDJqA0bPyGveGaaqVvOJE8cxs1cmOpMbpYqSs4LwVmc2VYdCmhtmKbsxNABnQDiuJaKMZsuWSQgTcltxsTGUDazYmzCGiPFGLJ19GLpnCHME8fcONqAE+UKLO4GEwZKLmhRtJVtV288yXhSW/BeQQ2SK1od6g3GgReltm21UgrMtXAOhqrCUh3nDNZsEjRbHbUaVAw7I6DbutOI5al6ziIggallJhcwxE3/DBgq3iihLQx14TBY2to2J03eDiNGLFkrysZhNLDVsGumFFBXt0qiBhpKBlou7FyHaY3UDFU9WSyxdJTSobpVM/ljcwZ6Z2ltS0fGpNTW8/e//czdjcHYD6yLZ40jZRkZXEfvEzlGVK4YzcRzz+R3uEEw5YLEjLVxU5raSimWkg+sU6Askbbq5gPwlrU2ammI60Bu+frDG6z/jGVFfNlw732PTBVb77lO35DT21c//AJGaLXg91DXlbUZFtm40zsr9M5xLYZkPdLvMWZk7BacLqgJtNbhBRTLcdjTQuDykmhiGYZAH7YPmj4YvJ3Q+oldP5F8o/9wj7kcsI+Kq1ttpyw7mqxIN3K57Pn+N0JSoc/wxcGhNUI+bYIM69jdH9HF8unnju+fC9XuaEvApsROla6BaMY3CCYwnXtu3x3BNXah8PbGYCRz2AcsPX1/RKvFdlD9dXMapEaZEj2BwUT2JIL1OONZL0d2Nx+o9QUreywfsN2Rfh/wyWM7wXaGznVIF7g8NYQeP95xjjPJvvDuzYjGM1oNOUPvPT4qdT1R9rrlOlqlFSWVTM6KdQYjjVYiHRHTAg+flNsve2pK1HWzjfUBpE7su54bD6LKT0/K+dQw3tKSIaaOgieeIrv9DtXANWXibeHSMiVFUjZo3YA2UzLkWqjxihQhvLYYlI1AJqIE12+eh1pxLmDaBndpCDUnRBut5q3ONGXqZYHmyNphgnIYT/T+EVhppmKkUNaIGIOI3VCiOA67K06/Zy/LxoJPUOc7Uj/g3EC1jShwaSMnRhYz8pQc1R6YTEHdxO1BCGaA2Cil4zwVFnpsd0t1O6Qoo6/sdo0H7UirUs1GuTRWGTuLd/L6IQWn04o/BKx/s/HfeQGWjXFhbnk8H1jTDVkPJNlRbI8xhTf9Vp+TZrg2z3X2tOQR0zgMHayVJVbiOXG+JpZmcGbAGMG6jYsvTkltx6cfbnl69OgxcTfuWWy3CWFMRbVRNeF9INUFzOYIQRqX9fpqimPziLQKJdGbRm8Mhrb9lGulaqV4y6OB3BZs32OzeX0NeIptzDXR7UZKKZjWseYFK7opfLVxdEJvA0+pUpzDYtkFz6ANpwvGdCxVmZpuyGm7aXVbSwxdYF1A2yYqqigxVzonlFapLdO18rrC81S3AbZKqzi3NYbWuvFG0EiuylArb63wbdfxnSozUERZDKT5QmhwN3hGlBoX1vLCdY5QZ7i88Odf70hNWbSCMZuTg4x1yoIhYXHGYlA8ll6gEyEbgwo02aqR4gbC2DFfYNLNaRJQnMimFw4e43tULFmE3CneVbqc2HtDrhaD2wR0bMIhqkHrdvBoxjG3jJfGYDKHccdy2ZgyrVbUgGqiaSG3RKkzTEKuiWobGSWXsrVF7IBxI6K6+WKkMlqhk0JVYdWOaCvJKBcTcA6yGKw6DJXOKKFt2OfqBOcDtRparJsmWhpJ2YiIRig5os7SxJC10VrBSMJJ2jJNqli31Te1Gpy9YRVFSybVrU4pKuB7LPaPfBgYO4zAbhhwRpjPke8/d/x4uiMVz3QNxGjovOVdX/lyMDRVsl4xrbBEJT/tqFZ4+8Zh5Bm1aRMtMDJddry83PB0blBm/CsHOlKJxpBUQR3l2vPwsuPm0NOFvMkY2opvlqrv+OnjBx4/vyfLji++sHR+xbYruWW6Q+BdL9xkg2Pk6EZCqlyK4VOtTIvinOGNvaX3kdE28gKnqUOwjJ0hmA76nnA/4jq7PQh7xbrIu+PE54//nqzPdENiP3iMVpp+IEyBozZ8iaSLJ/KWJ33LH37jiI9CGJQ5Jvwu8e4rBWvQmvB9j/hKixC6nqmNZNNhqiLzjN1Z1DZsUXK54vrA+tLz9PnA8Z3lw7sb4rLj9Bw5PWZubgO+E27fXlizMtw4VCIxnSlT5DYf2K0FN0cCitYDD//Q83P6E+bzzPuD4esPH7i/uSNNYNWippJM41SEp9kxZ4MzylwzOuywdaFzSlkm9FpJpXKqjYfYuOpE30EfAkOp1NhYciPGiPQD/X63vcFTxRT47teOu6+O9GGl9x5rA+/eHvizX33JNA2ktPLTTzMzlUveMc0r477D6Ib1zQnS4xU0MA6emxvhcl6o10a5bIjh0I00RqLrWXRBm5KuK7kZfv448M2vbkCetpE1/YaHrg0X7IYOLpVazKZErQ1Xt0aKlNcPQ2sRB62ciZywzpKb4dAf0LKjtQLoa7o+M7ortQpD2GPx5OJIMnCmQztLTj1TE14YWfRAkY6n3FiaYvtAb1acz6TUcMUgGtirp5ZGJNF6i5gjo7G8GzLfzSc6N2wPpFC33b8WijZaiszFcWLH6aUxBuUvbg2dP0HZDhj/8bs3/C9/e8PUbrh5M3L33jEcldFuLaTUV7qO7VJhhf1+j3rPtF5AzOsWwXK6ZC5zoe8zHsvgNoKisT3Pn9/xdz98zbzeYmPhy9ueU61ENYgJtDJvICdrqWm7bW1j6i1kVWpGXKCV/Prw2WA3238Z1Yg0j9FGtY5Ty2SEW7ny1nc0gbllatuwxKqNYXDk1bDGTO9fdYlasRhGLRi/56X514d9wxmL+O6/hOsmjWQAFXKrGGcwmvEUdtJRi9LE0BTmGOlDR9NGyiu660kinKvyOK0kDHE+47uert9znRdajURvWUSgKW9DwNzc8uvTidkqxQVSUiQ4ZuM2IiBK85FFL1sQsjau14V3uyOShOt6xZqNvNk0YRSM7yiVrT1hlGst24O1VubSKEYYvYGWyE6IxrA0pYhwaIp1DpsapSnVKLEWXoWfSOigFTpRxBpEhVy3DFpTRW33KimSDcfbtgftpa64nLChp5REAxCoLZHqgrhGbYnSFBP8xqXQgreCMVA10fCbL4UOOocxDVdXRon07Em1UqSxNofHkbHk2ig547uG0+3eX6g0EgC1WiBQNZNbwTv7ukoWrstK9Zt0L7YMJuN9RddMbxxaINeOiGM2lk49Io5GwLRNLS5ikfJHPgyoLZS0kGJBg9B5xyluK4B1Ktjqsc4xe8v3Rfl0VYIdEY0ca8abDp0EpoUvTie+fRsYXOL51Pj9xzs+Pt0x5x2GgPVbJ1itJUujOs+St9Glmj3Pl4He9Tg9YbHIdoXmdPrA3/6nL3l83qPW8fCY+au/2HPcXzFmpasru3EldAuBO456YH12LOcNfRxz5Xxa2fWFDzthf+eIRfh4geYsvTdoFWoRatONdmUcRiNpeSL733N3/JlmViRkjBrW6mku0B9u+WJ03PeCKz2pBC4fPeXnSu8a3ipid8RpY4KLEVRleyOIoPlAKjv05o4pG5yFLSRaOewswWxs80EWZFJ++k3PNQ9ou+X6PPL0Y93qRl3jL/6k5+Zuom8L4WDBLJj6jJ0N8e8K42wYW2RAWV8W/vDs+B/+5i1P3PJ//+86vPF0reJbZc2Ns1rWpeNvPwY+LwPF9nTdnrjqJhUSQ06ZVRYkKomOh3LgD/mWS+3xnwq/+kXgPkyUpVKiJTZBrGB3nnltzEvEoFwe9/z4feWbX644r6xp4Mcny8dTY8qZFgxrC3yeCy8RahvZLx37fc9lWjidlBQdQz8wLJnS99R4h+PAvh/JM1x/ztQEnTHs3T0HLCUW0hr4d/9LQvxXvP8A2laM2Vju4tp2ApdCo6BtO6CJ9pupLM0E25NSQnRmtw/EcsXqhNUAdocxAWMDYixOLS1njICScWFDXqMG6w6o7HjSRrQjqYPLmoid0JcLjjPezOyCY28dvVtI9Uo0t1jJSMn0UqheSKVQSmV0e3om3shnvrKenX9D7D2f00SRghphiYnWAlfNXMUS3ZFUTsx5YfBC8/f89vc9//P/d8dz/Arj96CFblz58KXS64qphUUTp1R5PlfCzS3l6llLo7aAvJrsBIglsBSQmogFinT04UvOl4H/9J/f8t31G7r+QF1X8vnML+57Hi4RLY3RZK5lpbx6SFQFK3W7TcVtpSUozjto7b9w+413lKrUGjE1MVgPtW43U+loNXGj29c/iyVXT/Q7VAyplNfPyo5YI8bKxm5vFVsS+76ntIGpWbIWom2M7vVQJo0ijaIF1UzwgbmuOIQPRvmVUUo1/DYKi+nIbkuS91g6gBKp1jEngWa4CR3ShOC2Sqlryt4IxQemVqgIzlgCsDMBsdD7HXLo6KzgjaUPHV4scs00TRRTiR5e1pX7IfBOd9hiWOoV1wtZCnXNONMhriOnheYMKo0lR/a7HS44ppjQ5YJ4R66JWRdi3TBYWRs3puGlEgvU4ol5kw1ZKyStRKtMKL0TTFGoG3a32sCSV4xxVOPI9TXIWIUsbK6LwbCzHdLKlmcsjdPLAiaTSNSSsWJRuwF7DErnt6+15hkjHc06cJYmFqds0xObiCIYFeZaGbCsuW4NEoFeQFrbfp+VDcBcI9YGAoVAoVDx4kAc1gqXlIjeEXOhyObjbhoJXjn0nrbAaY1U51glMHe3hLb5a2rePj+XdaFrf+w2AYXcMq02xFTGDmbXMb04hvOJg4lU6/lce37qbphj2Ox3JtPnSKszIYyIWsbzxDePT/zpl1uN8P/9Nwuxe4/rAoPL3Hmlt43RdPR2u0XsdGCu26Mf6RlcT6kV48A5Tyo93/828PmnHpUdxQY+fiqktPDf/us9nf1H0EhBcyGJQzqh1kLNM51tvL9rHI+Fd3eWu53Fj4Z4SRg6LpNlNgfOlx0/fl+YlwUjltBF3n3p6HrPO9ewJtJqouWytRrMSusybqfsY2SsBbsO1KdA+Fm4XSB5xdx0hNFiW0VKj9qeXA1Ve67pDZ9/PXJ+GDip4SIey8YLMGvhkcp9F7g9VPb+TJAzlY5lukHL11x/rth12/vNs2f+emTcB5bF8unzA8NRt/HcVAg5MOiOvnpMuzKGwvuxY5yOnHzl3Z3D6jOmGEZ34BqVdRr4+Rr4cbqldPcYc0OuUKmk0iitsayVqRVssqTiSXWgmIFq9yylcY6VP7k3kADjsMZjjKOI4aKR6jrGfmBulsenwvtv31JL4nR5w3/49Y7nZSCFQKFxuWay7JnUY9yRh5h5iIW1bCN6HzoQTyXz608vhN0RJ2BWj6jHB49YoRhewUhQzZVUFI1f8/f/UOldRPRCGBNOFbGNLBCCw2jDtgopYhBsTai7It0NZEeLmTzPyBcDpRrQRs15q8K6AVd6EFCBWpU+dBRjEfGgDhM6dq2xtBcWk/Bj5ION3OwN1s2IXoGVYC1702MkcFHPvBqO1mEQkhS60GGLo+WCyU+MdWW6Zm7LLbDnaiK7rrLWhPGCeihmG1M62dLtqo1VCxMdL6eOX/97T3u6536wSGqkVNH7xs4nTDwhJqHaeJgP/O2Pe+boCMFSqkfGkVIrRZdXBXlgjdDYiIF/uL4j5o7Lqee76z0vbU+nHtcFkq4c7cw3Q8PMF2y88pwyfX/kwXVk5zAm05xnqYBuSNztA1AoYplcR20FdIMyWWMRMRjfsSyR1QiIJ1AZc0IwLGbP55LAeVpreDG4ajD0pDzjjEG0Yg2YOuMb7M0BNVDLyqKFWyf42jBto31WUUqrNCPQhB2V23TGVKG4I4U9F+mxFnxV3gwju1BJxuKLo3eCE+iCMPhN/9z5jmAazRlWNSRk4y+slaMGeuMxeXsALjnRjMGocjscSHFC6wwCyWQmaVzjEzs70/tGaZm6RsQL0pTT6YlqHGK2UJ8zYEphihdE2HIM1aC5B+toIqg0kghLCIRW2PcdrsKSMktpGNn+Pk228fsqipiKGwz/P9b+q9myLEuzw8ZcaosjrnT3CM/ISFWVVdXNZgMNECBBGv41H2HGB5rRjNZt6GYXukSqishQLq4652yx5OTDvo1nPKSF+WNEmF9x1lpzft8Yy3mhSmBxhjNKFwKxKlhPVdkmQFYoJF7yzJQqLynS2hYgLEYpLW1hYCvUWrHG0kRIqREwOBdYW6HlrQaJMdi+o0xbm0ClZ5NoK1ITzTTaq3o6iGLyirZKMZ6KobSNWKhha1R0VVhKQdprykCUBc+LONzOImaHl5FERy4Z2Q3sQqDbOZ5WTynXFAfTvFBapdSCuoLvhE7/wpmBJRpSllc+v2B0QU8X/Dpy0xpX5UwLIx9mz7MGsAcERyuROSfGqzvOdQtDPVXPh58MfzwbWgt8sldY6RitR+zAZ63sbGQwZ94PlaGvtM7w0q45L4XgDEYsPgSMJMRYYgxcTgOmOnzXiAYWDTyfB/IaGb3F5IGqjWpGpjaw6x3VrKj9xC9+rhyvJkKYwTZCH7Yxr13o7h3n7498//uR59NITFuuwLSOFoWnuPL1L3fgw+tozEMzIHXjbrsJY4dXN/YMxeKiYcwR4zyZwJwjxzuP7xRRR2t7Yqwsec/nU8/5fGRaLIvNpNcdkFJxUggBVl/w15a7cY+sjxi3pWmX757ontjUoxpYmyXNhXO5MF0y7N5Q2gVrIp0ox+MRf3GYDJocLRacRt67iD96rvcKuuLV0rsezRVTHbGMnGrH5BrBb6OuitK5QE6Oy7oyvQYtFwyVTX6kGIo6piUzrZXQdWAdFMFlTym6garcRiwjwekSuMxXpMvMDz+NXOIN+D25bqPZvd/T2MaKS1w2vaitDA0GCWiL2NdXXDWBNRmcE0pzWAnc+B3OGiwVL4oTpfMd5MTSIsvcE186dl2l1IzxntDtWOdtT+2MgmYkbyPi1izFOF5KpuiBWDs+/ksij3vu377D2BXNyuUcGfoOZw602rZtqCg2BKofkXVke3M88fX9d1j/Dc4/susmOh9xVFpcUa00o2yt9gOxHAntLTYe6cWAxg2yArjWY2sFnTGSwXZUPEUrYiauZSCa1x48jdZtjQHbZgwJcYL6Sk2G6Z8zfNhxNLutM68Tosr587ZDD0PGSqFpx8vnO76bjmCP9Mhmaxs8tfZE/3rRt2cuiyNw4PE5MNc3TNHTzAjXV+RsqMFhO8V6y5ITXVkZ6guhzCR1XFrjRjqKguSFKUW613mitIwrBbznokoxgakJjoaE7SUnVmiquBBoBZaaUdn2/ramLV9jB6ZaMCIYbYgqgmHwO0re7J2YRtUGbcXajgavbP2KqsEWpddt+pmA1jZ9TycdnQgmK5RM7zKdVU5qyFVfTX0NUbaAYd6MrcqWMyhpppmtrmeM47ysrLGRzMbOv5SFohlvDHEtZF+xXmhqKetErI2VzG6A5gxFNgDbqa5ISDzVxKrQhw5xBqFhqlJR6quYKrVKCI2U0wbeEbvVu1tCS918HALqICsU5xmC46gOV+A0n8EamjbUOlw/UsqCesHbwvWxZ6pme7JqhdChayPnirielATVgtFC04kL+TXku00tvMBBLOJGQuexwaHiiaWQXcai7PsDO1PoQ8d917M3SusCyms4PE0YMWTJTFbJquz6njpNeLYWXbWG1Tps58jVbMF0FZacwDt6M2zK635AcBxKZegVHxxRRjrt0bqt65s3GG/wtbDvPXGx6IulikVdwfpKF3aEIOzW/2MO4//Dl4E1eVKCw9CQ8wX8ShDIouxFGXNCnaUP19vrW9NmfW1Kro4SoRrIJVKLYs2R0ylBtbixx2plF2AURaRxZTNfDQs3cqYlZUYxfkC0YTHEqNgOMII0T5p7yCN98IgmMpZqO6ztCEUIudG0BxM4PQY+rCPLcuH9dWTvz9zdLDj7jHGNrIacBuJ0YckfCcPC+rzj8bt3xOLoeqGzCgpL3RSlP/75if/TbwZccKAJZwPGbIS3WqbNyBd3FBzqLKat2NZQcWTjWFKh272wO85YB40dpTXS2jPUHd1hYG88g3Ykt6eJ4F2jsxEfJlKbyGRk5zHeY5whzRn/cOKwGozfxmsi26EW5wvLRdHasz8O7MwzY7cpF0yulFJoNpBSRFLl3k74wx1GgGrxspnTeteorGQbiNWRFVKALGx/dzyxWT4unh/HK0Q9qznwYgOLWKZUiSipDfzw0nMOytUxsOshLZWaQYwl9FvKH9NYsieVG3JdEY7c3Fyz1h5fN5Vst4Kjce+EJypONhd5iWWzmzWzMf+dows71vJCrQnve7IFq8IoitWFThpoJQye/XHHEhWWjhpH2lCw9oZUPZ8/B+qo7NoT3kUCjs4Vsu55jF/x48Mt55cdZbH4lok68uPfC9fvha+/eubYR3IUzifl+mjou0YrDQO0pBjXgb8mrw5vI1f7P7E//BfEnLdXANvXyVRB6/a6yDVRdKHaQOPT5vXIO0qeKV1gzm2r+jWztQ56oO/QNhBj4mAKt5pZtdKyw9zeoJyxZcXViTFsobbeOcp3C/4x8uW455KF1TY8Dd9G1qUQ55nhmOn7QGm35HJN7a83voamLcwaVvLiKDaQCxQMj+sNbTXEOhLrjuR6oKP1ATk6qu2obruASSsEYzCh2wJXLiDeQw2YqPTBcZCMKRHJCdeEUivNJXZ+mx5UO1Kk4qWx2EoRA1i8NxgDpcFUKwfjyS1s8iTNr/mE7cBqrdJaxlZBnH19L24J+ySZ4CK1bZnwRCNJjzOGg24hzXMtFONRo4QKiGwf8i2hNWFdI+YMbNS9pJulMiJEFWIqWKsUthZONokkZtshx41lnykUpxi3YW/3VvF4sjrWUihGiao8n5+3AJsqWhq4RtaFua4MZcVYQ9WO2fmthUOjim4KZMsm43EGi1IbmLA1d4pCcNtOu9W6rUVRqhOiFagFr0ofet5dHbikQqzb12vOmwmw1katEfVHYmzkVjk4wzzNLKtS1NGaBVU6q3xxvcNi+e70xGLAdg7nHR64kYHTB8HVgc7umS8LxkRQy7wW1rXQGcfR7hhKQEzhOSklFrrhhjfWczCBxQ081YVqhOgdJVR677i7uSavM9WwrYNEMd7TdT2aGy03gvF0YcA7xyCe3XmipxLXyOfTA48E3O4GS+Zhysy5UuOFIgP9WAjmipYay9SItaOogdSo9S/sJihiKabiUsPOQKdYaxAPIRh8glwTnSt4X1haQZujacYQKNVQNeKtQxBqNVR2YBSRgrMwdvA3R+HWJq7MTJdn4pQozVFoZJewYUMxpgheoNDAdqxxU+D6fU9JkdKU4gxj3217tSSUJtQwMj1ZTiehyMrNmyd21xPOnDCu4fyO3ljW6QQ6Y31mZy682Wd+NB6JgksXvL2AP7LqAeTAy8eeH773/O3f7Kn6hDEWa7bXhB0qhsZyVlQHdm4gpUeaQhZhbobUGvsQ2fcLamGe8jayXiHMkSBPuKuBGUPub1m1EtwZx0fmfOacGoiQbMSNCXXCkDydKtfO0owQncOo0JohJ4viiYsjtMLOJsZOCfuVloXzubBMAS0jvs0Ys9DywnoKhHEgDEL3rBw7JYeEOa2EakhFOMnK3ALeHagFihv4/mXB9x01B6rpIHjG0CNlC895Gp+nnufVctbEr955rF23iptziCglZsRamlqWc6Dmhs3CzWDIxoP0yDLDWiBHAnBwHfO8Ipppy9bDljDS3EDrHL5zWP81qZ0wnaXf79mLZZczpnbUso0TpW/0vWBSA/WcHnYsKP0h8Pl5zw+flXdfZQ7XjtqeqDJT3J7nyx3/6Q9v+PDhFtyB3jWO6QFtoKnj4UOHdZnwzlJjI54Gnj6eePeVY7/b05IwX66YP11BDOxujvjxRNfNiJ/ZljEb6yJWJfgOXomZTRxNtnBS5xXbBBULFrIm7BCw0jA2k2VBXQOJfP3La65OlePLM/u18Dnt+NM/eRaTub+74DnhhrxR74zD14HlsTL0gTfvhX9+slx0JKgg2ZMFLrNwHC9gj6S0ox8Hdt2Kd4l9eOHgFlppJHGU7orzYljLhaeLQ8yeanpWAioOKZVaJ4x71UuXFTRhpJBMo/qepg0RvwWodI+RK7RZ7m8tkkBXIYjg1VCXwroozWWaE6IqufPE8cBLTKRacdYQxDJQGXvPkjyT6Zha3Tr3Tigt09iENjEW8rrivSW4bRSNg5oSoS24DOdp3j4nh8x1d03fLGItU1Xk9fNb2dYG0WzrIbUO0YY3gjVbS8E4y2Ibp6LQD3SvQCObLgzDgLOey+VC9Y7u5hXOJhmriWCUnDKXNOMVumqpa+GxFYr3dENPydssUl49IGIy4jNqG2ItqTrWpjQMJvTYElFtiAhLLpvh0eh2ASp1+2x0lqmkbaWqirUe99qEeEmbMdaVwr1kfKr4mFDTyCXymDJDsKTS6E2H6MS+d7ycztjjNb0P9C6zb0pwFRscd4Plxi70RFJSPmS22q7bOC3zeaG8KKclkZdnhl0gVrisQmodxnuaWMR1vHvzDkX55sOPrKnSycLoevbNsOBI4nlqlbQWfAuEfmC37yjdQMzLhlJum254nbfGxvn5CWzAD4n9uGN6euBynjiTWVvi2Qfm4GjTTK3zRq1cM6ZBtj02Rn6Z9vQvmTwfueRbXpa0Ka7LX3hNIG3zpNMSpoIurwEvu+1kxHc4Kt0rYOJMIFaLwWJsh22GWhtqtx6qtw7KVnlT06jOYk3m11eJX4wrNSY+PQixdawRllo48wEfDHs/UVIhBE+sSk07zlNP8rCGBRWhVqWZgrPCtHpcu6bpjnndQESI4KSyt2d6zlBnqus3PKdWapyovGC6FWsrd7crVirMlqGtOJPIXQPPpso0t/zjP/7Ar7++oe9ntjk1oJaUG96CGYWyQNaR3FaynllLIQk4D3fHLYi2loZrjq7Yzdm9rIS84qylSGRJB+ZakeEzxj0gNWFaR1Y4l4jzT1Qczo4412HwiKtYUawTJvU4dlTroLui1Gc6b/ASMT6Ru5FPs6GkA14sO5NQtvrh5dTTq8HXhT4ojkTTytVe2K2eOXna2kipoSIY4ynqWLLSecPYeYJsrvGisOTKmsuWZyieqQQu2bA/wlEyRSvFuO3iQqWpYkrl/DDR4kTLK6WLmP091h0pc4aYETafvA0wmI50KZRctgssFWO3CmNcKkEdw3DkcN14exvoSiVkoa7KJXtOGc6SWOczdskE7Xi5HFjiwIKy1GsagS/NhbFrdAZM8fSm43y+ZjkPqOtQ+18PlT1LXjdyfHQs84hNgkwRvQhRvuAffrcwXDlK6TmfevLZ0J0z7uPE+980rkVeA7WWUhuxNYxRIKHy+jvndtACuSybV8I5kq40sbTSiMuMyg7jA9Y9YFzkykP1wq1/i6SMpAWvgi4j8Vzo30WkbKZPaQX8FT52JBkIB0/xI53Zk04jpttBrEzLyvM58P7N4TVx/ZGv7j9xd53p3URvPrKzWyMotAOUA1O85nfPI5Fbmu5R51lVsZK5HSyVF87e8zJv0SunhWaEi1rW1tGMoVBBLNl1LEUYy0gqFSdCdjsiDYsQvSerwXPh4BvMM5oLpoFvwrpuP09BFG8bbie0EHiehEsVNFdabZhgSbUi0ki1bnTUtkFfRDYmSrDQ6wLqUbul6FNJ1G4lqMeqobOWXBqwBRAXaVxCQzLMxhG1bZc4fZXfWE8zkKQgztH5gEexpuJrZJpmao7g4TnOm/SrLRhWvIAzG50wamWtC/Q9ohbrwPtG3xlSuoBmsEpLE1EaCcFYsN7yfNm0vrY1goCzilgYbCAYOA5bpqJo2+pwAt75TejVHLvXGnOxA9H2NCydbXx99OBemPeW1TtOdSHJlsfXtCHXpSmSLvRdBVfJROZ6JgK1OWqG57K5Gfo202rgMgmtjbhW4WXiRg3iCzlYXjRyzo2GYLtuO+/EUW1gasrzshBjYTaWPHREUdp84i5lvBuohyO5NB6XSBPHp08TdinMeSKmCLViAG0GkzsGTdy0RPFKrUI8Z1x2BN8joqy1kXpH0opzjirCrAV7DLRmiE3o0sroMu/dmclO/NgMj25AOoOLf+HJQE0Fn3oaK6Xmbe8jYTsMO0vOA9ZU+uAwzaBNSNrwGEQXJC7s7bpZptSBWkbxGFMJodtWDDZyMy503bLV0roEJOgU71fe74Sb/coXuzPBXZBakCqkvFU/huED+2FPxw7kwHPymDrxtGaepkKMF7IVFj/S9p7dbua2K5SYSLnQnCMVQyzPqGRUKqZFILI/zrhdQ5shlEJPYm4Lpjms7Yl14OOnN/z008Qvf3XGWCVGiMkSS8Co4/7NLfWjMP0YadXiuo5eHJ0Exhuw+kLJlRy3HWxLEalKZzb0p0krXma0nYlN2R0yJjRM9miFJoWlRLyZEaOM3S1td2Q+B9BM1q0m89PDirMeg0OoHCbDz4YBaBQ/8uE08lh7xPeYJEiXscFzPBQ0PWGjwZxXuqmxI6NeGYdK6Qs/LoUYezo7ImIopYFYdmPg6/uAOZ2QnJnyyFy3Dm3NK0UNqxjOzZOjh59Wfn4wW4vFKuIMGceaMtI5XnLBVkAbyzKj7QHRFSbQdcZ6R+48WTO5bAtJ9SA+oMYChVwLywXa2ePHrSqmdeZNP5KmFa0rl9b44bLy08uFGDO/uf45Jo88psB82lYN/XjgajCMYU+eI/s97IcRZuGod3x1e0t9KTRTuRmVXx52/PSwMj/NGO8ZR+F68OSlsdRNWVz1yJ+/qZTV0hqMNdNFWB8b31a4e3PF9TjgbCT4gjjZzo6mpJZQ8VgVrDhsdVsMoyYmmenCjm7J7JNwto1qFB90c4aURGuPeI4bKtsarFi6aiBtQSeoGG1ozaQ4I/EtvR+2/0cThgGuWkJl4rgT3mlhL8/s7A84+Ymb/QtfypmqC62uiCTIDp+P+HWPdweu3AENX/B5WujliBCYyDhZuU0T3k6c644Xf6AkhRZpEvgke/7QhAsbaKyIZ2lKdAGnO5I41E4kq2Qscxk5O4+xQmgXoj9jBgUqi3GodYDZ1DkCgYJqRTqlVYEIu04IwWKdY0rCmhq7cSC4fruoa6akCW83vbWkTC1bC6oLr3mVuNIr9FjsZWXo9zQMzjQWozz6QFFDsZ6lVmiRfT/SSWU5n+iOB4INlAZalGrYLj3nC73vuB320HmiOB4vC7YmegfeCsFYVOA8V5LvNlpiyxijlHTGSKRzjdoiDcWNhtwyszYG5+hfP8sxwvXesbNKCNvLPyp01tK5ghpHU6GwJfGtMVxJ4MsS2OVGNpYXe+BZ9kxZGSXyrldSLNs5YpXIholOpbLESHRKzRVomM5QTWauhWwW5jkiIdA7zxzPCAs7WfD9Fd3SMdfKfHpmrAWcJ3izTYvEbyj74AGD7xxNDQbHtCae40RVD8OOrIUpXnBtppdMnp6RPKM24HDbWrVV5jXzkiJZKoiiNeGcpbPCNZ4+eWzXsRjLc/O0wWNsI4hAsmRthGHPBtn3OGc20ZSBjNLbiiuZg77QkbiWA3/iLbn2zO0vTCBE6kYUtAaq22AvtsdKpb+7wo4jaOLGjfyiBmQ6owF8y1yZwldD4cYlnmvhabHYGtiFhhvg9us7HuaVX1298PP7n/DyQr+vjPczSRNZYdwJh9FQLw9QLlTSpnZF6fuVuy8uyBceJ3tId9SkLGlAujOHIZNjYUrwcD5j6srBG97eruhyQaoiWLRaUjZb/1RXvBdMqfSdMh5XXL+S6oCqxRnoDfRpIdkLgqDmlt//8Ynruzv2+8JaLc/rlm/YoErKF+GeuDxR1cEw4H2PqyOhyyyLkLDMubKWgorfOOfB0aJiNGOLYuKF2AzyhcHuDGZpdLbhTcK6hLMV7xVjF9z7A+fvDDnBJInrK8d5WnhcI4YOWxI384rYgFjPN3+wfP+Do8qADwMr4A4Lb75auXmf2I0G+W6G54ZJhr5Unk3F2CuWllnUYf2I6LYnNN6Ti5CKQYtF1viKrR14ypmlZJTKjPJQLM/R0eh4fDB8cwHTdpQo2+v1dYrUIpip0BnP0PcbDXDtKevCXrbKUBcGHkvhzz+dQUYGZ/AKNhe8H6i1kuJKqw5RS5uUi8Lz+cRyW3jTbWjbx6Xw5wf48WUkGcGPI51anorHWM/YC35wLC3yz//c+O4PPV+8Xfm3fzUyLhl5ydwFCF9d8Zxe+M2XjV/sH/nZl4nuezgpfPnOcXc98DJFPrVKKQanhtuL0C2g9Qw5EZrBnhdOs+GH9zdcvb3B9B/xwUDb3BE1NfqxoxUDaeMVeB8w1tIwiG+0ltAl0mFYm/CcGvd9QNKmKU52x8tq6IcjtXpO1VG8I62CpiuCPW9OMK3kXLHGI32H2ELLK1fmI//z+56df8HrhDaLDDPX+88g39LqjzQiTQA2SJM1PcYYtPWk5okext3Ml0NPiJmyXLgbGoeuED4+UlLPNG0/s0H81pAolaujR08FYwweJWlDywIEGoFChxHPXFbWBiU2HA3vPUYHFqlcHyKUjbCXrMXuAlYEI45WX/f91sMQGLqOK2/wsoW24pzwZWsV2FopMYJXjIVaMt57fLfbBDnrtvZ0ztGZjuvuACsMaYV4QoyhemXpPRH/uv/eHlmqC6Mb2TcYXAAgV9n28s5hLGgC4zpCF2jI5oShUWNG2Cp53giuFmiFXR9wric6S8pCS5HgIFA3I6LItmYDxLrt+yeWKx+Y147aZHtptwI0vLFUrXjTMLTNL4DFiNt8FWLx1WBTZhCPbZaL6Vhrzxorrlzwt9CotJKpr/Ad0K0u2QWWWrjUBWctO2OZJSIOTvHMmhPB9MRoyfGEmoh1EacOI8Jprog1OGt4pnDlDaMRuro5F1ozFARjBW/BN8XlSJ4KqzhOi2E2hqSVO604B6YHtOKMI9iOYA1jKzjnyMlSvUMMdH5Pigs2OE7NUYaAbVurpVVLeEXDO1+288hsTgMfBG1bDqCq0GKmIpsSnkZnCtLOHJgINjPZHdn+pauFLW8veunIrceFji4cMHHB+h3+OFDqmS+GzK088rJOVK10NvNuNHw17Kgv8PRkeYwbkcp7y6VXhsPEV1cf+e3dT+z9N+TyhHGZzmQGp/T99qdmz5wnjII1EWsb3laER0QKxvXALY0KNTD2juxOVFbMfuUokf0diOuBwI4ONzlKGrFmxojjuRrEDTTrEQrSBFcTnZvYHc7kcEtZBlKb0brQa8dCAl2pueeHH6/4x3+e+dVfF2JuPJ4S83IGKRgS17oxspMvJC8swVFSokhiqq9ubPZUIuo8uAHpOmzZoXlF58qQEiF2GA3c3jpqWail0tqMMwvWK9hGbQvL8RP2F7f02rjpC9p9YJjPnONK0QPWDXyxU4bBEh+E+cOArwOYQm6FxRvGv9rxs7+dsfoEJ0HLCBhwK7YJbRU+RuHlsdF1HcZYJBUE3VoDPvDxWfk8DdxpR4egAsl6zmVbs0QsS+1pdU/DMcHWSjGGqoYSDUrEdx01G0xTnGbkXAhi2LkBXQUTn7Fl+/l4rMLnNIIbCbVhS9l2nznh+t0ruQusKtoEZovJnpe68otr4a4PaDeS6wIukEvh01lwzVB1x27v8TtFbWGuiqaRl/yeh2XEMvFzM7H+tHCefmC9vuHwpmcXXnD6Pa5VxsM1USrDoNhux/7gOVxBWxvdtF2k7/W8HUy24+P5kZu+0Jcdn3/Xsfz2yHD8sI0Sl0JVBWPRVtGaaNHT9weMKkYuYJWsUK1C3wEHUhuY657O7umIJDVMWF52B2rrOKcDT0VYgtIZ5U8/9eyDcHPzkXHvkLgQ2wV/6BFgrDO/OE7c7gw2XqhrZm574kadB51pUskVCrweKCAOsD3tEmitJztPcRNWK0GFMayoecGWC10I4N5xEcuL9/gZdmyp9NY8zvWMotx6uDKBaVIek2KMYY4VDRtl0qM4B9q2fbiKoRbDMHhyKbRWKGw119ggChjZLp8xGs5YvEJKEW8r3ncbpbIamt3S40ns9rtZPa0YluboXsEy1Tuq8cQC6hyhKFYqw7VwLZVeMxc1PFtHkbpd8hoYFC3Qtbwdpgq2VS6xktSyxA2W04kg+Fe2R6Y15RyfkVZRsa+hwIInY30AAplAS227wImls5XRWrIqa0yIbRgR1DiSGrx4gqm0PJFxr82GhMEQdgemNVIqeG+xDZZYNqiOFIZ+oIllySudydjhmoLl45y5rBVbG5IEUuASE6cAyXUs60LGULSyrHH7vDdKKiuXmImlUU1lHDuoEKwhIxuICKHUiLieahRcYLWGF2mUsuDKwt7uaC2Qi+LGYcvS1IRfz9wapW8GLDTX09RCa3Te0eoKYjBOSFSqgDMNp5mUIjhDbA1rLa1kvLVoa8xUKpnB+I20abYmSBYhaqWgOPGvqyZBtVJywb+ufY3xtNg294ZpKJWg8bXFIOT2F4YOxZLpzY6FAFaICmlWlgLFRPoxsg+ZIE8Yf+b91YwLSmcFl2EXt9djN/QgEUqlrZmuG3g3GobjmSM/UsplEwBpwTpD31ucnDBY5smQo8Fa2Xbh7YKTitQMpiJksB7lSExKrhe0a8ypYbsJ618QE5HqCP4WLTfU0mGawZpALhaNBuOVoQdrLHGSjdPNyvV+wduCSqCZjtIuZDcydUrWbSE4txv+5fuKGx65ubqgedlY0lU5X848L58pp4X5dCI5qPsjbTxQZeUyfSb0Djcc6WtgmjKXGNGk7F3PqAGpE2Fd2EVPfC4Mv8mYcN46yi3SyrL5FbBEXdjdnAg3ldEsGHmm6DPD+JliM9Zf07u37Ow9tl0zfQCbGm+6whQbZ+1oTQljw/YLLBmtgu4bqhUtK3kJNHPHD4890R6oBkQT1WwJ6maUpoGXuuOfvjvxP/zyClkSpQlZLHMzRBxVDNYVhnpmmSrNNLzx1AbOdtSsVA2ktWIFrPEMWBqJqEpKiSCGYAdsSVgazXpc50iaWUWxrkMq1KwYyYRgCFKwtSAqtLLtdtuqrB+eue87OgLExlAttVbislCwDEEwdmVNjaUW5lZJ2jGIR8sNf/ixYgaDf7igZ+Xp8YLJla/vCsuaKFE5yELoKzcjNFPpr4+8bZVwKcyflfq8sLs8UOcz3f0N/Vee2yHw6Vl4ODnyfGQD4xaMUWLaDgpBKaWQKzx/Kux3txzGSpWVRZVkLNmNlLzn1AJzPbDMPT9LLwzlxJVLWJfxfeB3c8GrY+wCNlVOT5aPw885+j0/H74H85HcrfT1ib5uRE+jmRYFkwKUEeVAbZBTxHcZVaXqBs2y9pX8F5SSCs5sFFCxGSWS20wvm+CrppWiMzYo9jYgaSDrALahZkWxrBIo3Y6at6aDF8dgO5zpMdaiWjEu4LVR4or3YWsUSKPUhsPTScC6EWMbtIoxm7cjqaU2xzUdoWxti75lQjszeIfNyuwDT8aAJJxVsu2pNEJncd7RaiXVTBcsnZjXm5DHGs8Fi9fGcSf0BqQldqZHUsdDyqQct9S9WLxzmJZR01Flw9zuXCBWQ2xgX+t61QWGLtDbRtaVyyob+8A4Cnl7zVvPaB2lWlo1GOPo7Fb/DQJ9aMSiLKXSctkOdh+IOHw30HD4UKkqFG2vkKWGxkTOjVg37fsyL6i1r8bGTBd6LikhWTGd4FIhmUqjR8Xh1GPmBRNhWYSPU2XSQtWGuoZzysF5ioHH0yNaI9Eo1zbQ41lQig1czitGLGoc/NeuyKH0AAEAAElEQVT+P4m+35OMoXnDWjNiK30oHE3j5xzIGngohdVsF6tQE84Ychg4FaUYIWVFquL9tkrQVjZsdFW0ZfbGcGfhJekGa3OWRkVU0LwSnIOqG9JZXldwohs5E2FRoZmGMZsKuZaCoZHrFtLsh91GTbSWQN1+bqzBy6ZBllcQ0l/0MmB6y/PjzJpmDEJQA6ViD4mrq8Tf/DJj4kQthmXdRt3O5m3UwUCtjRS3tPllOpFXhW5PDh0Pn878+rrRqpLUUyiodZS60rWC9Zv9q9YOF/qN+sa6HeK2gvGoBuJ6x2l+T+NLNDW6TmmiLAmOuw7YfpmNcaRcNwRpU0yD1iync+HTx5V71zjuM2IKBEMIBk2RXhL3e8fVsIe8kksl01F8T1ZPs9s30Zdb/vDNmX/924CIeQ35CJIL63SiRUdrnlqU6QwTyvG9x7SCWQswM4wd/VHZu0KeFw525rA4yI6fTo1AJU6viEtdt050AXldDxn1ODOwpAzyhPMTWj8Bn2l6obWKlXVL+Sbl9CERH2/Zt267UVoh0vDeILqiaYbaUN+o42dEDC1CzCMPD5Zoj2gYKWJwwaAoRS1FYZWNG34yAy8OGo1LVaK2jXvfVjqbOAwzd28Nzuz5/cdEMZbUGlk3vLCUlb6H7rAjtcbw+MK1r1y8ZVLojXBVtzrScH3DT1OiSeB0Wfjm8yO1jVgfaFpQLvTauApbl7eWBq1x++YaTROOSDCXLTh7dFyXGXWGNU0MnfBm2NO7ivcjpXV8fJkxoeFTYl8TbX7hcx7YTQFdE9Ybnr6fmH61cjwCoXG7j9ReyXnlp/nMIDPduOcai7sEVrtwYIFa+PSysH//hi5EdsVhZsfl4jdIiVGMCGKVqkoXPLUqrvfML471wXK1G+lsQ0SJRinSqDbjcyGsjRYn3t4bwkuDtHKrM9UYPjw1Dt2OO7fnOigfcuQP3Q2qgaaGpB2nVOiD0GFprZIuic7u8XpFyZbV9kRN3G7mI4x9lfW8XhiNbeAqrSTEbIsDKwlrK8YErBGkNrCWUi3VgfjM9RA4xcDkKsXuSM3w3AYWdRjnudRMkcIlNXJpWHpqg3VuNK3EksiUDf4ihqwVKRE5ZXrxBGfoSBzLihElhEJwhWM6caUXOm2ocZwucC7C58tEZ3tG3WRUdrlsJjwXiCZTfUNbQlqhpLoBd8xrMCwIBiXUSC9nTJgxWnHB4XWAojRekQUieMzrC70Ri2BKQ0xmNHYLmaGIdeTSaLbDS0PzSs3KskSM3yh3z2XjlDhrWXKkGE/Kwpwb0gytetJ6JreIUb89mlJhzRVMIGSDLZEWdYMkWUtUwyCBshaWpETTaE03YJN3+NCRY+Lz+cRIjzTDwfbEJsw1osWBeEzNsJxp8xNOetR1G1Hw9XW/pmesrqhUrG/UICSFpUSujGJyI1tPs9ta1xoD4mmtEJzHZaGKAVUa2+GegWQrrq2k6LYLRBSshQ6/fQaqRcyG28dkrNn4EKs48EJ6lXz1prGTSq+R5LbkSa7t9WdcsMbiWoOm2wz1tfW1pojverI31OopidcHad3qq7Eh3tDaVhTIbNVPg2DUbCwH122kw1bw1vxlLwO7oyNYw/JSqLNu31xvkQq7Dmp+oMSMd3dos6gVmjxhJFPoebl0rC8dvg2IBAiNVRrzaeX5H2eOd47j6FG1VBHERPo+EjqPdz0pKSoDtWSw+bVa47cvrEBqd3z3L1/z9//8lu7uwL/+256u+xGRjGboug5lJOUO1G1u9Xpib98TS0dqyilbHpfM+37g+GoiEyvb60AFUxs3d46jHUnpmnUVghwIZcDJjrk1cLBYS63XXJJlcAtd2LSp+lxoxaDG4YeeWgutOi7nxjpVhrc9rhVySZS6ctxFbkalxYpeEiaPRLkmt4CYSlo3Br4KtBwoxWHcERcMuQiljtTWqHpCOWM00dkGxqEuINKRUtwqg6kDkwm20PJ2i64Gisl4u1BLQmkUs1DdjB97SjvylAc+pwE1AauNYLYxVrWNUjOpFLyHIGdSWfj244RTgzq4Gyf+zdvIu37iyj0zuDM7J6zTHcPDnosrXL2xdF1jHAcGhBQr5xb5NM28sZ/4SidOeuAfLxAnT988pnWYXLm3K//NXzvaMvHHD5HHpUEXcCZxGBs7f+YwRKwRWvMMzvH+vlHnzzRJVBFy2WN1D9XQ1HGaXygshHJgz1vKlJDWuPQvYFa8rtQ6cXaOpVlOU8aWSmyFhKIkzDhTXUU1khPEZpFmWcrEEAp7e8NgB1y7YPMZaoV4xXc/Tnz9N3f014G3uWFWh2l7VFacNFoRsELKCayn8wHTe+IFnO0RVmgJ6oWuPOPNCb+zLPKGlYFgJuTxMxoVhkg7RHI3kENC9AMHLcDA93Pgdif0nXI5WaZ2YMbTG4O3I0ttPNcRwpFhvCK3wGn5ni/x9H6bXvzX/XOpUAqIrXhfsH6AZcC1TOggN6EVC3UTHBkXUAYwHm8ao1HW8PqC8nuMGu4lE0xh0JlWFDWGOAzs9gWNC2XdSHi+33gMa2msuVIVRue5dyM1bSu7u/uOv7u7ZqwTwaw4Xhhjwp6fUYGP7Z5zOtDsiFSlGoutjV6F26GjlspTNjwopFrJrRBaxcSyGZ+CJ2LIFQIJLye6YWEo0zbdtIcNi6uGkjPatpbA4IT9ceS8ZhatWNmAP8VkshaKepoainVIEVoplDVji+EqXG1AILs9lopUWvMYs8noUlHEeKz23LTKz6hcoudFK4smivMszpJrB8tGLRS1WyjZBawdGa3lfn9g+f4jcb1AMGSjzNOCxASq7F2gs4EiDvodIj0me3bFkFPBmA1KJm37b6vtyQJVlVy28KORghgwxlLEbiKnCquxdIOnYFEs1XQYZ4gaKXkCYwkiZFXaawBRRFAsWQ0TjoXNJEhuBNvjXKC2RKbiug7fD0jOzEvhYiy4hjGOBY9Ryy50OLPSSoZtQQkiNBpWhKr1FVJnqJq3y3VWojSWsm7/hm5ip2a2FUczG9tHxWK8pTYlGMto3dbYa1sFsqqAbtA0yvqXvQzUs+Fq9PztFze8/PDCwxr54eWENyPf/nBGcsE2x7u7K5pkltSYSiQXx8PHPU8PgUPouTUe04+kPPP9OZG0h7Ljn36f+frLRPAXbu4q465tHw42gFoMgZoGWvWIZqrJqHNUVYQdP/1wz7d//BmfT2+pUdhfrfzVLzYOvPMGZx3rrLS2hZV8G1lzZbGex3TNKW4vpt1t5vrqhLWGYB0URfCb+jJtzOrYZewwMuwstcKNDZTFkJbtK7qajLGepwuMV5adhb5rnHLBmgNm6CjxTHh98VixzBcQ01FoVF8Ru1LaBScWazwFR7G3fF4HzliKdZwuhZfpFr+zzGXg8TNUPOKUm9sdh2C57RMpf0PWR2qb8buemg1OHNostQZaDjjxNJ/BX1jjnhN7TqanDRe6YWKJn7bbsBXcqLScmJ8tD/HAg9tRS+OWSCgrbXAIM57IfSccJdHXmWsXObSK3zmiLHzx7szfXT0Szj8idYZaIFrWp4yJP+f91zf88uuVzr+w7zt8Cjx+rvy4jlz3e37pYXx64rtL4F29JbsdrSqxdJweM/1t5aubR7rjmbtuJVIQXwjDwvVxRcoz83rZCIIucNyPdPkDRZ4pnUHtgRQ7NCbStCX0+2FBrfL0kPjzv3xi7K/52dtbjqVSz5/xomhfue8s38WJHyrk7InVUtxmQUSfEdeTL3njKCDUsklMnmPkEjvuaex1c82XVtB14uWz5/d38K9/c8cX9oI1b0gPt7hDxEpk1zUyjVr0lYan7Hxm2PXEs2BYoEVM2/SypixYfWZnnzAyw/MMGpArQVrg++8y9rDn6zfw1is2nah/PJPwOO1JaSZX+HRu5HpGS+a6v2Xc3fG8KC0pu1ixIsQ2Ms09u2FLkmzIvNcPILP52ls+4Y3Q6ogpCZM3doDpBmqGajIujJh6hU3XBCNcuRf2pmF3lv1N4K9TonFG2xOlPIIY+vCOlhPr+XsYV4wtuHFETcccA/NqmWIhVUMRh6hnNkqLQlcyQ04c8oznBPKCTZs7oiBM1THbnlIyLhd667Ch42AdB7dRFdco2BixxtHsNuLtTE+nAWMG9taxc3A/KO93hpuusMyOn152rBwZtGdvFSGjCA5hRPHALvQsraM0v72KrOJpWONYdWOxsK4cBoMdbrGhY10T1my0TGP37KQwekNTQxWhE4fajoDwtzv4ZYO0DGRnNpufgR+j5btFmKQQrGUIPQuOwqaw7oDDcM2v3uy4WidmXXhcX4h2EzXVkpEqqNoNHT9lFrUssaK6oysgbmM5tG6kmN0WrjObsM73A1Iu+K4j14hYizHbmiIZ4cla4pw3HLFxOPF4FYKCLXYLhNIoeaXR6L3QyoYjLnV7dTcx0DLBeUDBOdbaMK9rlpEtW3bRF15KZRHBNNmcBWKx2z2CVfx2WbUerUrVjLSKYZO5Wd0mZb3ZuBhNIapiXomUaMOkSqqRZjdeQ3COkA2+wSDCYc4brjorJW/Th94LkxbcX3oy0A+Bm+ORL7644ldvr3g+n/DfPvMPjxP//vczHx4DX79/yyKG+91IKyvfTMJ//H3i6fwFoTty0wu/2AUO5kTtR57XgakKpa10pxP/07/dcX/1jHMzxcXtYMxKqxmNHW3eYaVhZUF0AWtppiPHn/Hhmzs03bBznhnP08uyyUJcQZwHCagGKJlmtpt03+2AA5f1QDKW4+3ClzcLh1tFn1ekWLT1aE2IeHwQ8IZowDpFO+H24Ll2SjdV5PMZbAM5cTteeHcz8eVNY7Qe7wT1Qq5g1ZBapdZK8OAkkosh1UI3XDgcDE1mrFOc2SGLQ90NP8xf8NHuWHeFnIVV9/ynHxT1A6cXYVoNtsDBKA/fO64Gw83ecPP2LbvrC0XnjaktPWjGiGMpDq3KwXiqbZgAn2Ljk+2YXGC3V0K/oLaS0oIKYEF0x5yumPWOs44k2zNedfTFEWXmtnvh724XvrSV7pypDwstNaiBWAzSKfX5E/244CPbZatT0tI4nwW1Pb/8+cgXbyKtGPK68PnjmXXuOVcL2uhapBdDwvCUDaWseDVE8Szq8CXT0iNpjbgSwCy4Vhic4FioOiOacWJwZsE1R/54RqXiOkNrjp++Wcm1Y+97jHPEPHN1f8Of1sbHFfauEOYzd7nidSCnmW1DlxmqINZykYGoQqnbKK+miO8PDINjigXvuo2LYCEmw5o69slw1XkIAQkVlxspWn76rnH/S8Xc7Xg+3/DyT3/N1c92XL/5EeOfkFqx6un9iBML3UpcK5++vcZ1ldt7Q2/PGBrW7YErmh3JxbHMhWG/h/MHLuuB7z4O/N3/PPLGf0A+PKMpMucdP02R+9PI7U1gf5e56mamdeIcPUknmgns3AE5wz6u5Jw4VeV06vnyXcBZobVt7N22xxHW8IoVLiCB2npqWqiBrcfdC7lWLJUgDmccg03EdWVGmLVjV0/s84y0zxR9JPszpmvcupWy3HNJDdUJa4XWGk9J+fAEU/Qb1VJ6cuugbU74ZBy6nvBG2ftEKIW1QTU9bhg3z/zF8EWrUBaaVs7Nc2Kr/Ko21O3ZV8/NWplKZDEQbYdRYeeEXx5XfnE1c9MvBCY8F1Qbk7lG5ciP64G+dOxywhhDflUs25o4x2ea71A6soyIP1DKTC5nxAuxKUPoGf1A0ETDYv2R3jd6MdhW0ZzpbWV3UFrL0CC/wqp2ItyFmb0GovOUsIHjirE82x6aI5LBb5I6i7xW6ZQQPM1bdrc7XBs5zxdM6Plw/kwqlcH1vNn37E2gd3uelspSwLpxI9SSQYSZjuYb57qRaX0tOKuAJbWwjZVq2VTDxqB2u2yuCqU1tOUNwITFNEM1hrVtrBPrDJ0T5rRloWBb+xQsqVRUF6x4ct0ycs5s2uqcHUE3nkpP5q0xtNdpl3ObU+bKew7OoUkopqeKvE4pGi1XVDK1JqpA0619oTogpbJrsq39uoGWC12rHPuO4HqyFNZWoCk7K+yNYykVgsIat7WvZioW8Phg6fUv3CYYr3pCGDlHxQVLL5Wff7Hjn1Pl0yVz7fd8GzOTTqxpQtMjf3q64Q/LX5P0ltvWQYyIVMZ02JC4mnHDwrF74d/9NnPjPzHYCWhIVbCOFBtOO+o8IAXoEmJ0U3dWh+o7vv/2nsvl3VYvccqat1eHDfqKB+62naHb0fIFrZlqGsYEajEY0zMeLL/5eeMYCrI8gFpKhbUI3vZI83TdZrFqxuGPjm6w7A+Jw5j5hVz4N798wcoZ4Yngn/FyZucCx3BDXUGPOz6/rFjfGHYFzZFBPYUBdpbDteWwv6C+bPs7FZbq6fxbzvnA9y+G6IXh1uKKMEblz9/+RH9Q7oYdv3pn+esv33JdOz7+MPF8rpznxunk+dW/uaHrPmAkby7sFvG2EMyMC4nhpiNph5wd8qB888NK3FX++18nxn2mlortNv536Dxz3PNyORDNEfV7nAyIFToyX46NXxwTv7lbOf/wmdNZ6egxfrMwnpKj7nqC63C6YmSHNkeOzzgZQK4wXQcmoaYyJeXjxwV7Vhh6arpgJ4cdHGbtmFriqWVcG6kIswhzady7jLMzpQmpbkS6vFjOZ+XtL6/o3MxuhK63dMEiemCNPaZbkTSRcsfL+S2ZHfbY41UpzfD9D4kfn2HpR1IVTCwc+gOmpC0gllfyUliLYG2HdJZWDGIt3npsdcSXM8YfEITcFNsNNBRaQIJnbJXhxmPyAH3CXBpptTxfPD9+nHi3a7g1UPJv+cA1/U4YDyc6E1DTUWIj5k8Yjuz7X/PpcsN66bCDcncUCoHHhwPn8w3V7PlwMriL47+5u2J9Knx7PvCYLMZGpDxAajAXynDD+XTk979LfPG257h/5ovdR/xtYV73PF8cRhZyNtha2MXKmgxT6inriHWeVjeUd6mgrzyUrYNfaWzCG+WwURVz5BQbRrbKXGc3WZOlsreZZhvNB6IPNOtYJWPMSmkJcYnOVSQ+Erih1sBUHFYMl6x8exI+xRsye1Y6sB2IoFJZKSjKwe7Zd89ca8MtgrF7PmbHnB1pMVjpGHKkLmdQRU1HE0cZDYsMPC+VJU182Z/pXeI5B2q5YtcPHK4yv73/wFv/HaZdttxPbZS6o+rA0B8o68CiI81vXoOmCWHBh+1imVvCiMdhsHQ4EYIXXDActNIbg5SGc0dKs1jVV0BRw/j2GtRccS2DGLwzeAJC414KOzJitgR8lLJRk1qg1YCVnhAUfbVpers1bqBhXNjAS2bb4w6+xxrPbn8gScUZw9FURisEG2i10OHxFLreY82Oa2e4CpUpnXnvAnXaLIilZKZp2VYg9YSREdMKFktrm3xKmiGljLOelrYaZXUw01B1qAU18tpMsxgrSN2aIFGgGYcYg8PQ4RiK8EUXEGvIEuh7w64VQoPiRibNrHGltMgw7tGc8G1rOq3Z0Mz2aNmpY5SGpJXgAt4LzgqhWa5e82XJKvVwZMLSmsO2gKExOovzPR9PT+RWOVqHpIWpWIodqKYhXnBWMbpua90m6H/9JftLXQaMtaxFeXo+cfbCjVxwLjO4zY1908HXNwPl9IGdXjiOwk8foVs2/C3aaCGxxGnzNtfP/E//58a/+vXC9f6J4/4J8pkSN42o8yNaErF6pktHm3uQjAuC9SMpWz4+X/EPf7ji++/vCeaeHZZWoRSltorYjLGVy6Xw8Tly2HuG7oYcF2r1tHpNSSMa9vR+YR8u5PWBVhfEOlKplNKwwTAOgXHnGbqe5zXS1pmr68bYPdPJE84vhOEE7TPOR8RE0lKweo20kc453r4pXLlIYOOArxfDj58rj4uyCoguGCbimrebrt0kLC+XyvP5ia/fV66OK70qzy8XfjP2qFHS+pG744GD7+ho6IPhNnQk3TgHS77i+eHMl+8tzkVMt2B0xokBJ0ip2GwI45eUyfHl1cjxJ1gqtGlFtdA0b35vL1jreJn3TPIlpzqQSyX4hpWJX39R+Zu3lRs7kdfPTM6h+xvWi+JZaSjNWrCBbjcg7gJWkCDYamgY1uI4PSe+/faBpifWdN6Y7uMFO3zkTQ/GXtF3X0K7w1+Ectp2sNbvWGvP0iI3h8JVf2Y+NaoMqN1h5Mj54vn+u8/85jc9uxGcqYi3lFNjlh1GAmVpPJ06prpjvLpHdpZqGiVbPj+eKEZxtw6Te6IaJrOyv7+mTh3kTc6ynGF8d40mz7ooS3lGm+CkwxpDrgVj9qg2ciuUZohFGezK1VgwlwW5Npj7G65fHPJwRZxG8iI4LdRp2ayW6z3zl7fsjwe0TrRa6Vz/2kWGUgsvsSP7Wz6eO1qIfPoAP/w44swtb+/uKEtimuEfzhEzX/EsO1YRpofP3N/Z7Qnvdryox4a3fJ7gf/2D8vX7ifc3K0N4pqXGecrs2sTbMTBmwRplvVi+Wypx9rTSEKvI5pRCt1wgAjhnmUvFiCeVgVwNNiz0Tkl5wnc7nDFo2V6lRiPjYGihYg+V210DnZH8QjOR5CJC3fI0Gaq5JtfAZUpEvyO2a+iPoBuESk0kJUPTkaKKdRlh3YikpeBQWoWnGpjmHb44Um2kUnA4rClYcdTFU/uKDzN/tY/87VeJX7xTXO349//B8h//twvdFVwfKrdjxrfKBvJ34ARtgWocU25UcazaKFQ6q/gyYXWiWwtbesIyWctjvnBOFSsWMYUslc5Udr0B9Syrp5oeaVDy+r+3r6zJOBpTWUlGmasQvOHaZw4hIiwkvbAqvNCzlkBaLPNiSZdINQLWYFxB64VcMlUMz3XHRe0W1tVGr4otW3q+vgqNYomIrdAlrjRw1wrUSMagvqNMmcvctvR9D4N1NGOYqFgfcNbj3UAvmbevYiOpulEJJZBLxXeeJoY5TqS6srjM83rh5fSMiGclItZQa6WzntqUZgU1gBEaBi2Gr8Ybft4b5pR4ZgsUupYZnUEcvAqxadViuwFjDUFgbIZhXtm3yt0Q6HzAXgpNesSumMGSfE9cLWGxjAWiz8y61bJL3ibI3lqSNlpOdOIQ2HgIzVLoQfZU72nmkWALQ56x5YyRHqN/4WrhTVfYh8p46LjfG3ZuzyXO/O7xCfezxP/9XzkO/kK5ygTNmAJ/82XPpwp/+PzIaBf+zV813u5nWD/z9ubCv/ttIrgfsGahlq2PKnior0n+DJKuSZcRdMHuHUs78C9/3PHh+cDvH+745uGezy+WYHfs3UhdToQy8bWN/Hf/44Elr/zLnxPff9ihvufqZmQt8PlkOF0C4vYMQfirN4lffvGIk4rKjloFxOLMirSI1sxa4PTwQK2e443lalwZho8EHujcgsiE67dKZCoKYqF6tAjOV5y9oG4mzTue5o7nB+HDU+GhZYajwWiizJWqDmMs3XDEyBVr9Lx5mzkOPyC2oRmObsHtHNIyeYk4WWnpgDLQ5Hr7cJXM9XiEprTokSaIbhKR2iJVoO+uEN+hdsf8U0d9cNgY2euJePJ8+4+Rv/m7kf7gCV4R16gEPsw7fqoHonf86leB62Hh4B/51dtKLwstTZRO8P2O09U18UEJpXAuiaaOpIofPNIL9ZKo84QfGzVXHl/OrOcbnl9Gbr94pgvPSHvCuYjYCAq78ReY6Zb5pWJXwy2GuS4k22+Titq4v430/TP9+5FjtVzOwk9/XkjZUuaeee42Ktmh3wZ4ZYZh5LRAbAc+no98jA55mumXynHfQbJMZQcm43aGWBpLq3yoF6bza8ZlGDhHszUnWkcNjoZgEqg5Y41Dy7bbz6ky3PbkegGUo+u572CQFXlRSB4kIaND2GNGwxdvArfphVYjHy7K59OOy8OO23cHBMU6i6rBuRHRKz5PRz5fRlJ3zTN7fv/5mVYHmhp+9faK0RgGLZh64WCeuP95QFLh+VOmqwoxoXXZVm37G/QcWPOZaY4MxiHNoe6e1pQb6/nyix6nC+vTB9rZUNI9L6dK/ijUtse7j9vkzr6m4wErQOvJ9Y51OlLsHrpGsAlMZbc/kstEa7AWS7IeGQuuLRxGh9rGsj5j6zNBJ4xEnMkUzUy249KtPInhWQ/MKZFyx4Sh2AWTL/xmV7m/Dvxv30fO+R1GEvd+5Tdjw68JP/QMvkFqLC+Nx9KxMwFcpoTXUJhuVT0V4Tdvzvzbv5m5Pa548wymcDrd8vL0jq7eYS6VN96wCw2tHWIHbG20GunHDtMHfjYmhuVHvogRTKaTJ273GVszJu5hPYB2XIry5/PKi7NgHNlu++6Rii4nonRbD1/PjK7yNix4OzH6giuFUQrX3qA6MLlA8Zl9OHPXP3LdT3R2QcXxzeM133y6op47zLryhbE8lkoxQo0LwYGnsFZDWhPVDpvFMUWQSpOyeQxQal44qMFZz1IqXb6w12Hr3athUmgSWe0WOC1sdck+bFME9Y21NZaiIEJpm0rZGc+cKtmk7XAsbC21VQliOVjPu+MV59SzlExbn0EcHqGvmea3C0B9ZSjUpog37I/dVkFGsdaxVsWGjmoMHZkgWx5tsR05Ka6B2nVTkBvFaSPEE3sMSSqrcTQXUOc518A5G6q1YAveVNo6vTIoKl3ot/WGKh5LJ3CFsNPKjOVReqJasnQ002HzC7ta2KVCtA07x7/sZWAvlSuT6SRCnrmsn7Eu8H/5zTXP05749BPFrSBwOTkePytmN7Ib9uzdhZvuhf/2l4l3xxPjUPD2TBgWptOCMUotlVxW9qND6mt4owyY3FPKVs17esh8Xm/4X/7DG759vuesB3Y3bxnvOtIl8cO5UdJAqDOnP1/46997tD3y+enA2f2Cl7zn9IPhUns+PBVMf4Pmyr48c7OrJFUUISXFu4Ch4hBEA9O655//EHleK2O3x3vl3a3D2gWtJ8QoJRdKzPQyYHSHE8HIkVoGqlnwHrqh4+m54/ffOtY8EquyqPL+1uNaJF0yqRiKs6w5E0bH2Hc4kxC3ST+Qws4VphT5/LLn09MtWnb0/Q1p6pF6IC6F9byhRO1+BNlStaVtf0exBnGGJAa44/HpZ3z49pZ9PWJq4r4XWjE8v4z8+fsjP/91j/iVcfCIeUNe3vPxY6Y7Ft4clNvxE6N/ZLDQctn+1ELLCyFU7O7A6BwtTjznSibju4aMDbmvuFGRnIEesTumKDz90wv9buXf/O1Eap+AilgwKhyC8vztmcv3Abtafknjs8J3UTnVxDgWrvsL1ARNsV3H+mQ5TQNzydRqeHncczo1dovj7TvPqo9M+sikjjl7LquyqtCKEheY44Irwnmp5F5pVAqVq2DhbPg8CZOpFFM5RcO0OLQpQiW1tuVDnMfTaC1iwp50OnHVd3ReaDYS5zN34w3MBnxE9wLqyW6HVccXTrh1E0OcKVII3pEeI0+PI2/TlyQGYunIUXBiEfMF355/zVP7gjUaRBqBgYMUBpvZ50Sdn7hNgusbg0nMMRGLw/czx3d7ODtoPaX2XLQSxg/87c++5f/6PzzR+QdOL5WnHwYOVwdu72+RJqS1UP1bzN2IWTpu8sA/f/rAy+ma21tLrpVUoTVwFjCG0m74nO+Z2dG7wPA64s51YbpAqtvOds6GWjJHnxi6zdT3j//5gftf/Zyj7TG1YMxMVUjeb5f4cuKSnlguHVkCxjd+dT3zi3fClXvitk/Y0nhb95ww7OzI267BvDJFg32zo8UX+r7jLvWczwZjhY5G7RZMWklzYLl4Qhf5t7/NvLv+BmqEXKFWnl9u+ZefBqT16OWR6RQYd1e0vHFUamnUpsRcCfbM2+FHvrx/pusiVlacPCOcicsVH7//DS9LT8mGEIRf3HacVvMqrSqMO8O9scQT5AB6Veh3E1f9B6x+ZAgru67hi3JTDtjHK77/QTjFhf7rgLubwX7EuweCzUDHPB/59rMnaEegUbSwsz2l22rITRrdYU9tliUL3u0xS6EfAiqJ1UIfDpQ1sWaHtQbnDcE0vJ6RErGidHiqeFYcC4JoIzWPuI1cuLMd0lbEOFZjCICVjHiL+oB3PdosaKBkxcvMz97fM5hIBzhdt0NTRr6a9vz+s+FUMk1fkLLgh22lsZQKtt/qelROrXIxlpNCyo3r3nFEuS4rYiODF/558TxXj25GGExrVLtNQnyKBG+odsunROtRHJNYslOKMTStyCvot+sGom3wWhXFCIIQ4srRGrqmZBuoxbGKpVmLmADNIHWT//kC9i9tLTxNj2hu2HZmGFaoK+dp5hSVVIWf5p61Dkzq+ebHwOdzh+0G9rsOlbbtsfRElYVYG+NuxxIjz5NAVxmGkTCOxGZZlo7Tc8Zbh3UD0Vvml5WX557/8MOX/Mc/vyd1X2BtR/6cGdyFQxiYWkabJTKyrrf8f/6LZbj5a3ISMp5PMzzEjs8nMOHI6HcYLfSmEXXiwzOQMp7KMDa8KLXeME9X/MufvuQfPn7Nws/JF8Mf/n8fqP2ZX/98wvrEspRXsGqkacW7PcvasZRrTPPbCsNA046npefp2WwELOuAxmHIGGZoQm92xNaIU2ZZnvjy7ZGmhuZH0jpjCnhpCEf+07d/xf/772+IDNxfBfb1wth3PJUBXTuORjjmTLi3LDnQYsVbYRz7DRySrvnpxy/54+9+zjq/4wsrXKefuHYzpW9UM/L4ccev/+Y9IejGK1862mnkjT2iteHLwriD/XCgC47zwxPrsmCcoTZYYsYEQ+4tV9dHTIO5zIzhmcYKYUI6qNOBDx8GPr2MrOGGaY18+vCZ7l9VxC5bBe31H2cWzjQ+N8+imaoXHI3R3eJ7w5u3cNVHaA68RduZKpYwfIFfOnK5YNrI8lT5/Lmy1J6h81SzYMeAMxY/KDeDI3T9xtRoQl2F5RJxXePmxnIzOr7aeV5+WjitllMZeYqBl8vC2DdcjXSmEUaHN4mbq8R+F2huprULO29YLmd2dyu1XTjuHMGdoM3osUcvhVgHvllHkrniq7c9fv9IK8+kWHhKjaI9p7Pw43Mhy4BWT/AHsgrrOrKkjr67sHeJzp8JZG530J8bx7kiZWIYjrx8uvAghtq/Jcme+2slLCd4UKR6GHq+3H3mb//HzN3tH1CeeD5Znk+Jw3Gmv/E8/2Hi4/N7/vQNtPUKrwZfM6l27Ia3nB4cd3ceQ92gOLrlBkSFmAaeHnoK26tqcJ68OmIO1OIpzXE8BPaHjp0MBITTy8KfPjk+yjVLvOKX+8bg/4wfAkttRHW4oHw1ZH41Zh7WgtkVdveB2/vCaB/h82d4Keg50n2+5nFZ6d5/RW2J6eXMTy+Nj+vMv/pVILjEzbgdFHNbuDle+PkXiteB//j3wqeLMDgoy0o7GDDHzRw3eB5/51nKboPu0PHNHxPX94Xf/maHkYLSmE4wr5H91TP74RPB/IAwUVtEiDhXqAZM9fz40FOkx7rMTmd0rVwWJew7QlqwLdKliNNGbxu3h++5O/4jjgcMK1oLBs+1+xnTZeHh41se5j3z05lf/XeOuzdC11V617D+wOMnR17vKFZxpm4IbFGsrxsm1/W0Ijhjueo6pCmh76AKS97aX0X9hp92HWKFJBVtkeAGYnxhcJah76gmbDZXv00CivFk57Z1nk4YMfh1xdfK0Sg3nRBaJubGjOOSE04m3jnFzhPlR8tFIVqD0wtOE9Z5Bgzvfc/b8ZaOjmASoQ8YY1hyI5mBJSomJrI3NNfjVbgG/vtby9t0YmgT6ha+9IH4MLJOPVXrVrGuBTUbXjig2BIJXWCNwos6SgXcFqgW2saLUFCEtTaK9ThrsVq3lUtL2JKQnFGUpTkurZHZcnRG68YekY7qhRLMhrH+S14G/vgRjBbKqFzf7Cgp8OHR8dPD1u88dD3L08KHZJnaHU2uqEVwq6UT5f7NHT/GyJ/+kCjV8v5dx7try/NTpIVKGD3G3vAv/xL43Z8HTmVgf+O56SOeAqnSypH/+MeRpF9g7PVmaatQSiO6LdcgxlG1Y6kjHy5KqRVcoLTKnA3n5BE/UlGM+Ffjm+Mf/6UQ7Iis10gt3L4dGAfDtOz5pz8e+Ic/3HHSn4MckFrpUoG/f+ar9wFDIteMsQknK8YMxOx5Wr7ix48/4zx3WC58/Qa+7AWtgqNQq5Bfe6SGig0GzZutzNMRipJL4byeOC0L1oCXyr4P5Br4/PIl//4fv+Qfn34JYc9zm/m7dws3+4p5KVySxXpHWxcezh13i2PX9fRdY78fKHXg9PCO7393yxq/wISR0DJ9rYhGVqtMfmB5CaR15DAeQTynx57zY8GlSFoLl2Pm/dcB1xtSipSSUWOJTbkUw/fnwGW2uE5594Xj6jBzz4Q3hSUJTRu4kT9/uuY//8OOH+M7XrIlKoz7nuAqJjiWUhAMopUSX7Ddl2As0ipeFW1bHbPieH9fCDqhpSGcwAg+3JKksdaEkUadJ2yc8a3n06cL918CYWu9UQKXtfDw8D2q0DnPrh8J0qMt04njy+sd17tKx8T+KJhd4OD2HGPji/XCm13i6BRvLIvu+P5xYXARloihYALcHfd8yA0xForF2Q7DAK6i+xM5NP7LN3v+v3/qCPdv+PGnCz88Cr/d3xK48OOpsmqAVUl6R8qJoMIgkavdLdOUCOEbfvsbwxe/GHHrE0Y6zhfLj3+fqM1yODiubjLhxvDw50QelC+vKiMRPUe0OqiZ6VKwV5a74yOSf2Btgaqew7Vy/XaB8YFTG/j28xv++NNAf7hDxFLWhWAKzjxi3S0Oj5gIqhgHxm95AbME9lHwYc/TTy98Ch2lHDgcj3T9nhRnrnpwecYNOz6cHN+eBn7IIy/dyDRZfnY30h9GKDMaLS2feHs7cCgrj999S5eusf0RiQ0xBsTSxNPmSEuGGo788CHwU2m83QVob/lpFubnR/q3ytv+J8hP7Izw5v0Nv/71QAiPfPfHC8/Tgf64YxgGvn+wPM5CjoneC3dvPbXeY7sjuQVag9wq//D7C+NouLttXM6P5NXzxVfv6IYz08u3SIDaLK16qoLgcMbgnHIqjt3tO7ysnM/PzJfI87KnLIbw/oCphpwDcfGkacZ65X6/0NqKUKmpvaKTDe4wUPcjz/EtDy+Jh3//zL/7H6752RfPONfx/Nzz3ccrPtYrjHGQNvlV1UaOF/COeVVQjxOQunJ92OO6judLpfo9xQayDog1GGnEVshlRSksAsF5lIaWBkbQ6ljUA4FUwWpHS42ghsEJ17cjV6Zy74U7F+lNYS6VH5cTi1GGELlV5fPF8Vx2SO/xPm5kPg0b6t06rvoNTb0rgqmClkITYTCOxcDgAxaP+oZzlr23/Kpr/Ew+0ZdnbC0UVgYbue+v+eO8tR0KBus9YiwO8EmwtaBlpdo9F7ZMj+YVz2aNTC0h6LYqqdshP1bBt4pqotSCYv53T8NLTcyvNUbTKhrPaCusxjFbYfKKL3/hy8Dndst66Xh6Gak/Vaa5ksuBJD3FQLckDrpHa2Nw2+2+iqfpJr14WSr/5YeV5VmAPd+dHF++ddTnG2YxzP0Vp6drvvlu4GzfMfuB9OOZXlb2svDVlaGTwENmk5NowrSyhdtEOaeZ1TTwgZiVKBaVHtEdMRqsD4hpYNJ2oDQlzk9o6XkqCaOOl995jt0t+y4QVkMyR/78Yc/nh1saXyEMG1nMJtTOJHPNJX4k14UQPN5sWt5aD3z/4zv+/X+54/v1DU9LoNMDvz5V/h+/qlzJidILUXd8PiXOKfL7P2Xe39zj9BlrFDFmU6jiSEsippHTafvQHPYHHj46/vN/vuLT53uGcE/1O+bk+O4hcd1Xbgd4544bvtL3lPyJ6+M9+2GlcxNWGpdz4+PvGyGN3O6E82mlSaGp0ovBG0WkklfDOgtxMJRLx8vzDbSAlQXNwscfEu9/OyDmR2q5oGx2M9zWD/7pQfj8vHVuv3+Z+b/9t5Gd/4FWzyStVLPjdz/c8P/8f93x6elLbP+GliqBM/c3DecScW1gDMZ2OJSA4PvCcazonEiAdz1WoDMr729hOFi0buuKl/WW//V3Hf/0Lwu+H3l/1TGlBSeJSgZ1rLkS28i3f3b89NGwFI+qYIxlxvD/Z+1Pli3J0vRKbO1em9Pcxjp3D/eIyAwkEiBQRQpLWBQOOeCEU74oJywRTjgrCEAQgkQmMiOj9cbczK7d5rSquvvNgVrNcxAPYG5ifs7RvfX/v2+t0AqO9XuWs+Gf/zCx2cJXtyBzoBLRQrDfBLrNATk/kE+VkC2ntseMr9ntX9Me3tOUphnB5u2WnC0pPa88guCJotJTUP1EUgN/etzw08selhnEwB9lx6ftA38zNPTOYYvhSuHlFJGysR8yv36zpZ5Xney7d5nuJoB+RqhIiYbzWXPV33KKiqCeyfo95j7x3WhAH3jlXvPyh0yNPY2OY0i8zxs+hVeEHxt//cs3KHOily9oWzC9gbIQ0sDh0FPNnqIsomm8VFyLR1eNTz1xMsQCRbOaJAUoLaiy4UTlco48nUYmQMgb6suV+1tN31U0Cd00L2d4P41g9wwajmnC2Jlt/0ysH0jtgDKKu16y2w5c3m/4808dhzCSP6+979sPhf/5/7hBlAuaPUpBk40gLI9PhsdpS8irJ07nxnl54dYFpG6oEliun0D2lNh4+OBI+TVGa1SbKdER9XfMSfFwNfzxcMG0kaH3nAKkYqktcY2GH/4U+PTQuH9zx2AvLOczYZnYbu7IKSCUpeQ1ayJVwjmNdB0vF8lZFcZOocqW0CpTc1yi4fJTZKM2CNmTa0+OJ+SPF/72VxsUE5q1D7+cBSFprNG8GhX/dIZcNoQL/PMfPU2/ZpCBZfqKOX/HoW7WynWT1JgxuqKqwoQJVxdUWn0uSigEV4ZdRWiYUJAbOnlkaqvMR3hGZvYmcttLdmOHbYCzJO3YzJnKQqkZoQQpnzFKstHrM+5moxhtZawZF15wYuJ2I3ndLzjjGaVkme7w52948MMKRDIKZKCVsOZ7EGhVIHt0L6kRapWkIkGtgidhFYYBUsMYGIfKt/ZAfz4gY1hDhgpKzjhVcCykCrE1tJNoUclxhWbxpUpYRGYuFSE1ggokprQwt0RVilglsQY6DPc14MrEpTaCHlhkwppK/sLF0BWsBDldUDlQvmQsatUUoajtL8wZqMERAkxVgunwOSPlgGDtfGdpKapgZUYQCaVSWqZIh6iS4/HK7eipFIyE53PkmKGUV5yr4KVsmR53iLRnUZalNXITZGWoLfLz0xXVrlTZQTxSW0c1IGQjlUQSDSmB4mmiBzcyZ9B2u9Lb4wwi0YShybUiV1tceQVc8WRCcjxnyY3ZoGbN47PjKd5h9D07obG5IADfFFc05xh5ePaE1FObRlgH+ms+v7zid3/+FX96/oaL2FOFpldw+uEZVWZ+s9vw5wV+elFcQo9Ulo//eOHcFv7H3zR6PSOFWhXAtWde9vz4yXCZR4JSRLXj/feGx4c7qryna5WUI7Fpnk4dD13mrW2IljDSIGql6zSlrtpea9w6gZg195s3nJrl4/UTz7MjCom9GclPCz4HPBPFWl4uHd32FWUZeD72tJahzpTScfws+e3fH/l3/35Gyuc1+SoMz0fD7/4keP8+07SgCMeHH048fTXx5tszrb5QpeFp2fOf/+6Oz8e/AfOahqW0hZIlJZ6J6Yo068OH7DF6+FL9PKI3li53pCgpVSNqR1kWToeF/NZQq+V43fAff/s1/59/eMdJ3tEXwycl+e8vlbHN7IbE243jcu75r38e+O2ftixyhzKFXl25venY7EY+Hl+YTolUBtpLhxYDJV75t79u/JtfbnhzG9gMMxsricdC8JGWJD5kim6MTqLylZAzIWuuQaKawvQDpoX14IwvzMzIrqJlIukdL+c9pWyRUYHWJKH48UmwuQm4OqO7r/DNoVvmfqv55SuLIuOnwjE27LWyCxK/NLav36LqhBWwfVWR+sLXb15o9RnkgrtLWOPIHzOHh3sCBq0dp2vHB3Y82oHDH/Z89dUN/eaAbIGQJzruScsbPvx4z3npcW5PyZUmI62dUbpijacfNyC3UCeWJaCVQAhISSDpGNyWKUiy1GhtEdIQypaHa2NXGuBxrme+7jgeeuww8XY38+t3V968u7KxP7NcnxFqoVIZbt+w+JH/9g+O8/It1nWYVLl6+PCp8P554a0yGKuhBLa95H4UPM2anx4jehwwsnFvOxRr8aBpRQ0epS0fPs8IYRDqFZ1tGBVQ7UyaZ2SWtDZyOu/46cVhTObbX46Mx4kQJ4y+8q9/0bGrA3//+4kQFv53f+tWgJvasBsLoZyY5gNCa5TKdJ2FrHk+ac7plutlxOTCqDoGpZDOEbLk+VxRasS41dKoquHHz5+Zl56NqcQsyNlQ1cilGPYmMw4T0t2TS0/RmU8vhu7xDa2eaKXHK4OQHuO2xNIRS6IpjS7wqx7eaY+cGzVJus4iusbmjUD1YdWrx8B1Wg9PQWQwkZ3zbGxCFo1MO2TpCFQmnRm/9sj2hNILQmQKjZQanXH0WqHFaoLslGAjZrTyFAJWTFi5cD1Jkhk5TfDhCHGuaJFRUtLbDb1oaF0JWrBU+OgFOY9MXhKbQSiNVIqcBdprZDZ0u8LXryf6bxVSJJTMSCWoutFsT50kXVqoWn/BnktKbeQWybIREcRWSDWsxMGYsKqhRKXUjMdTYmUYHd+Ijj2Wb6qn1cwSFZ+F4ig0M40qHae8vhzLWhE5o5AUoVDKoenpcIz6LzwZEF5AakilUbpjv4XoK7IqrO0pLaAaOKuRTSCEpAhJE5Ime1LuMfIG03tss7S5cLgIXuQ9x6KIyaKEZbPr0WicFSi1QZLWm6fUlPmF+2FBcaCUHdWMFKXWv1c0FAIxamJpRDFSYoKa0KaiyoQ0jUWte8pUAqI1Rq3QMiBKIUlDdY5jsMhZEWtH7RwyNlxa2DRFropFOOammGPldJqR0pLEljmP/Hm644eHNzyfv6Z2XyOawdYKAi4x8c8PM9dL5vm058INpetprRHjwH/6/RF51zPNJ3wU9P2GJgfOy5bnF4lPljlnjteAyh0agc0eXRVZrvQ5RM+fPl0Jd1vwgc5WejfxbS9YqiCFHp8UneqJ7HmMWx6vmkc/8sPF4aeJ513l372+ZzqfiFLxECMvfxgZPt8RFo1sdwTfEFFyulqu3vLpv39i2Fr+6juLyJLf/9Tzhx9uOcZvkNsbmug4vGRq2vL+04m/+UVCisg1jvz2/Tv++P41vblDIyk5kktcb9IyE/MJrT3WAsIAjtgMF9FxdI3TKHhuAz+eRj5cBJeo+F//W+UcN+gMH953/OPjdzymr6hmwF8Kh6US6nfQ3jJer7yrKzL0Dz8OzPIVte9odUE3yRIj4xyYlw2fXjp83lKwSNmh6y0P//jMw7nwf/m3nr/+1nPxjUZH1FtySTSpkVLQauDFCzK3xCiY20h4trx9J3GmIHVCofEpMTWNKFu+/2Hk5eyQUqOEIitJaCDNFt1FTE1csiZbw+sN/Oabnq55iJogKqHbUueKelZ8PsGfz1vu31xx9szb14Ldq4wqPzOdP5NqoOmMlltC+Wtm8R1B3lLiYX1gioFZrYHakBzNZ7QxKGEIi+bzw57H41swA1J5uvbEd19FNuIDVWViPpKv37MMAaklUitSLNQGzWiUNyQkQ99RXlZKodOFhmKpliWOHA6Nl9ORjwdFVYbt/sBvvn1ht7+g7ISfnhF6nab1O4k2kedPgs8vrxnNHb2ckDohOsdzE1ynwKvXCzFnhE8Y0fjlG8uH6vgpZIyUq/K6JvwsiHcDUi+I7kKWkodnRW2aXu/49ivFfqsJQSJUR2+2HC4W51+RSsLnieHG829+c6RXFywTW215+LAhBkeZLc5ZaInpOOFvC9tbvToZkqAzGi0htm/46edbqn0FcseSZ0rzbDqFywvFN5T7K7xwXJOnUwqH43jteHkxbN+BFCvER6lK11uMsNxJxfBy5ZpHUhvI/o75g+f2qzdkrzleTvTWMoUJLzsurSMu0AvHr8PML7cGhyTnFb07LxmXAr/47hnRvkfLM1KuiFzZCpWCFIVWG9W/I50sLUqyc8i7wK77R4z6icpMFZkm19+VqHtkcrTiUWr1WMhyBTLFJ0gzUkuUe8eSDUFYmt4TiqC4HUoKiAuKI9oagmu8nxNzGMnFUnBrHi2v6OgtHUaN1Co5HBaOl8BXQvA/vdsgeV513MYyV8dPHyMhDTQqyui14t5WRPRCBiWYSiGVgGwK2SoSsEbQSmAuHkPh/PKJm43im7s7hvMZRMaUzMskuaqRz9JhhAAFISb6qLF01KppAoqxRKPWyeu/zFP0L78MtLzuDoVwlFxxxpBEBFFoZT10pcyUNCGFowqBtG7l79dKZkPMaYVe5IKNBas2+GSJjIgm6KxAlRnddRinoSUQhVwKUWuacdzuI683mbZkrrni1cgiKoJG1QKcZhAria8XA6XOKD3Tu4rqIzfScMEScofLkV/2DhsWHg5XpmFHcJpLDMjW6AdNSgudEGxlz65pirMsDS5fapBhFrRWKc7yHDd8fB45v2gG2xNS/BIGYQ2UCEtg4BoEQjk6bSkoWs6Uojn5LX/4LHn2jXPq8GnAt46cBwQdslRyDqTU2IrIttOYy4U5NM5KMStQNXMMhUus9HpA6YZRHvNecbGviIuA1rOEkev1np/+3OPLnql2ZAZajbw8PXJSM0YPLM3wTMfjsefpZ0MTI1b12EWioiGHtu7Epp7rfy78n5JmmiO/+/kXfJx+STM32FQxrRGbI+fEhxfH5+sGpSz/9E+Gf/zREd0buiwRMSBQCCkoLaNNXT3kxiPVQMx7LpevOJ6+4RzfcC4d74Pk95+3/HzeUd0NVRfmw4k///2Etpp4Uvh6T7Ob9TLbMrkkgnBkt+cidvz4+UhNlcG8QurNl4cPJNHxPM9kJJPf4tUr0DcosepHC5qX3PF3HzSqnbhcZjAdSxq5vCS0AjcM5NwRfx4oYWHsfo3VCdd1a5XVXrB3idIaJ9/zdN1znA2Ph573n1/xsWwQ0qDR+FTx0pKFwt1/ha6J85PiEq7M0wX8hfUbpxj3PXbZEh8O2EWS556//2y4CVve7if08sC/GiIbFSh1NX821gfXi298WqClTAiN1EHeTrjtiX/1bUSJ0+q4MJUSNfOiuJ4dfQYtLmz3gd/86sI3N+8R5cd1LDpfSeqINIUmMr2uq8VOiRWF2xzKWhqR3VZRhYOacLJSVSXnxOOHCxwXcla8+qvMd19N7DdPaB1IcV4FNk0i7BafX7DhSAmBvlosCisdLU2YvKBxnCY43SvOacFUSyo9ITVS85yroFw9vlP0XeHx08Jub/jFL87c6JlYPVlDrpZ0bJTrDmdvCRWuseP7D43TVEmyoNDEqMjhwm74hOUFURsp3fH5aY/PI32LCLFKllosTLNiuFslNDEKSnII8Zo//enXvD9/g9A9Y68pZmTxgc4JdIyokIiq0Gy3UvGUQOKI4RU/fRr47quBxrK+FKVIjuc1qCZv+WZ7xz9OI5dsQApSmjm8VExS69i7Zu5HixsU1+b4w6eAz5UQBMtLhFjJBS7nRGAkz5H7V42v3kw4+wnZZmpM5NRIRWKcAbWjVENeLCJXhJsZukQnf0aIB0LJxLxa+3wUDE5xZ0ZKSyDPSOFp7UrJFS07hPAIAU1XTrEiOwPPFSUGXKpoBb3U2BjRVSOr4jQ1stquwcYqkcKidaO1sLJBZCPLhlQOn+74u+8P/OZNR28bTRgeveO3hy0/LHfMZqSUjJMJJ2GJFzyBKa9V3Go1WkjulERIR+csqiWaKcTrwikVfGtMh8/89c0tTRbQldd3PV+fJT+dF4wbMVaQSsJWgUmeXq6tlaoaSWpKW91B5PyXvQxIlZDagZC02oBGPwzkxdP1ClUNI6uwwse0YjtxNAmZSFY9n04Tr7uRTTpTl4RwhWqh5EpXJQbotEDIRKgBaQQ+R7LI+Jy/yCoSth3pXCHmxrmsVY1OGyYZqbKglaSXK/41ahBSU2WPDBeESQhRoFVaS8gcuJGJ6DRXoYnCUbWmM5qN21AvaXV6Lws3ZsNEZdAdg96T9SuK+AVaH6j2FTndkBfFvRvQquAphJJA9SxNMBdPExZjLSUIdKtIUUgyk9sqYooRKo4sBxY5kOpAZ0dEhSYytSi01Yjq6WTm1S388HlG6xsGJTCtoLRDmw66btX35j3/8GnhYxgxRlOa5eBvORw2tOUe6kCram18CMtRwj+EBdkGSqv4annKWyIbhO7RwH4c1jqoKEghacJyuUT43c9Iq3l//ZpHf0uNAx2FGyPJpRBKx0f/iv/1d5FPjwvvn++4codWklfCMuiOqhQpLZTQuMaOS/iKKS6cpnveP37DP/3+ns/Xt3i1IzXBEkYeTju82KCbwaiCz4J2NEQqxlgGq6nxSidHqAopFZJASvFLn3egtQknZ2QtlBxIOlJlQKjEdA34YFcxSg5Y1ZFrWjvWZuBa7vl4+Mgv7kbORfBp0Vzrd6sxLSpk3eOXPX5JGHvkV78AmS+odMSdLbvdW35+CPz+8w1/vN7y8Ww4e0tig9gYujRgmgLpWKpGpsh/f6iUHJmiYD8q7vYVJVl3hDWhXCV8vLJphloSKleKr0zXwv5XAmMV07ww7NwK+tJiXcfULcer4/OpsOt7lhQJeuJ//PeNX//ymcG9ENJncrqQWsa6jhwkfV/49a8KSl7phzOb7pG0fMLImVDPFB1RBmICY1f7nhSsStZcOMwee7MnXwY2QySmzHRZqKKAvGJFpObKfjfgikMIwWUxoPYMeVk/T/Sq9JWNYVBok9E6oGQgXM6YXUXlBYHAV/i7D5r//LwjLbdYeYuoI0u94TlsMEaCVXgtaVYw2m6tgt5oaou08EyqZ5rYIJTl+FI4z4opb3j/uXE8WqwdUaqR54VWGzFkXK9I3nC5aD7+PPBPfx6Z2VLnF15ertg205LifNLYm5HzJJmXPZ+fNjw83fHw9Cuu8i1Sa5ouyN5g1B5fzrzrNL/YGLwSPEa/ciCEpDS4tI5/+njH3avbNQxrRpTsydGRrlts2ZCV4VXn8L5QWlxtpXQkueE5aUQZ6aPgq0Hzq1c9Tg4cj4+8HkA3jYyVFD2pwp+Omc8PA388Lvzf/q/v+Ne/PGEIKz+/dLQmCDnRuw3zUXB6mBi7jvHVFs0TVU4rXrfKVYQmVrV8KZLcOqiGKhKtLUBGIKBesSYgnUYlx/W0oWaDRSHQOFHppKArC53QvMwzk71HqB2tOlrTWGlWmmSpiFLQqqGrx6jt6j5pA49+z89RYMXCFAQ/PvV8P90wdTsmofElIWNF5kAskVYjICBElAo0GSgUjLWoc/sy5l/QshC1YhLQtcxzeMaJ9UW7M5Jf3jhGbZByANWYiqYguRdXttd5JRhqvU7eqOhakTn9ZS8DSlUolSIKTRliKYjaUKIxdgJdJTpV3DhCpxAxU1VmcB05NVxvmFpApYSTA7UGUi0gKrJlpF+VnkVqak6kklFCUqjrAxtJrWv60sqIk0fQioAjiYaqjabimtIlEy2kJijKgRQELBSDxnMtkSi/0LEqUGaQloRlKVCERVtDbgWkQOqKswlbBLEZpExoa5nqyKm8wbVKWAzTAoNWqGVGNkMSjarsSpcTepUrKcVpnpBoUp0oKLJeMaqd1oji0EUgq8bIAcVIDRVTVwWm0B2hNpQoGC6MVjGOO/Kwo2hFVwNOJq5+wXWGOQsyjcA9U404tbCEyGnRCNHTOUVZCqk1RNMooCK50hPjGQgI5WhV4Jyi5EArkdQWZDYY0RFKowhNYcRPji5FNg4+L56KQ0hFiomSGiHDYen4+z9uOC631OEtHkPJFaEFSzPMIbBQkHLDHx9uOcWvaUie53f89tMbPs/fUMwtrTScKSyxUeT+i9Eu0kqmcx01R3KdaaqRZMG4SshPbNUNlMrWaUiGKDTGCXJt7HY9NSQEmlYzkbxqZKUgFolRCtMMJIVQFm0cl5gIdEylZ04XYswYe09KA5eiUK2jni0KC8KyLJ70BXebquF3pz2Xacfvf+/5VN/woF9xrYJYoQJI2Aw9tjladpynRF069IdVdrLZwq/uPTd3hqk4RLUEXzgcn7HhLW/MHVG8UErDNk2nI29fFXI6U5Y1fQwrJVTiCGGkLhu2naG2BWF6avO8fntmd/Oe+fq8EilZ6XYxQ8sJqxbM4ElLISwz3WBIagcyE2oGOZJqQUqBqJ5W0voAzx5y4vgc+fzywmazI/uKqkdk8YSqWWQk14bQlqNY/799+HPi8scbrO5593rif/73jrtdolNnSl1todYKzA2EMeFLIeeFHsEpZ16U5PfTKx5PK0OBtkG3DS0bak7rGN1JYr5yd1P49qsOnzWlOEqViJppRfJycrx8Gjg83+KFJRjNlBxNdahaoS5IATknXg6Vj6cdp2vPn38Y+eHnW67iDdIIrL9yjHC3afhYWc5bPv9gOAXNp6eBp8sdp2VkdN9AWkPQSWUuOZGIaB8xLWK6DU4qaGtYLZeAqomK54N/xX99/BuMmjFar8IboxhKzxt7x1wlN0PBuIXGQqcrvp75r48wp1/i9B0Bzc9PM1OpdH3Hzc0GqQI+dGvVUFuMkAzbDdOz5vmT4/q/nPh//N+/5tdfWVrytNbj+og0C0YKXo6V7z8ZbG94YwXf7CzVmBVOJTNGK2gruvs6J54muN/fQnteHTVlnYBol1A6Uo3heNnz4WHPy4ukUw6tQMqMlRJbC5021FQJvtBrTVlYzYW1UvKKzldkZM3oYmhVIZujSo2vPZ8uMxTL8QSntifqkSoElIAWCSE8hQXfMkUJlJRkBSFcsNojdCUuJyQaUSrGKErKSKsoUpF143MOKKEIiyD4yKE8c60GUc44M5KVoqRIr0DVSCyRjKW0jDIF2cpfXmHcO83pArRKoTFYi20SWzROr0lKodZEdrQKNzh615FSZNg4jGmUvCXlTBSN4RY6bdjYgUE7bMo0YchGUogowarlLInOdsS8uthlEXQCDAHRFbzMBApGWJzWNFmoQtBkRHQdSiliWhWOTQhiWGjWUUqk1oJE0nJCNkOujSIUUhmkNuTYSFlgEUhdkCJgWkV8CaFk9jyEPbt+lWwsRTDKwq53xFZYWsLY/ksdTiC6Dj02ZKuoKpBV0ZAkoaB19F8+kFFBkpYlV6SCceiwta3Bri+kvaHMDD7gWqSzlbOpxFoRrWG1ZL/ZcM2RXMU6DqyaJYzkkjHWsnM3lChwbUZ1e86po0iNlJVOS4QtpKDpdVnfzHpFqnF1Ooj1xuycRtPR8oqR1WJgtLeM+kJthZvBcoiNzjpMyAydWh0E2dIN36JsRUiDk5qoGlNKNONYpGaumZY7zvUtw1JBaKp6i7c3zPkGaXp6uaJ7h9tMjRPnJWN6Q2ctsl3QohAFNCNJKqBkw5pKjU+ULEHt6ZRBK8lUIkIZlhRRulJrRKkCubKkQBUQi6HWKxWNEI0YFDlUfGnQPFcJT/M6Xr3MgVNxPC6Sek0M1XC3MdAELXfEeUKoysU3Hp4kf1wcaRmprsN2hq4KhMho2bGURFQLAFJZOj0SRKbfWkYlCOEJxAPH+ULJllZGSpY8nl5oS0a4nsvJ8nRexV95CoTrZ2r9gGyRksu6GikLJQnq1HFje8LWcFzgZfEkDSlGZp/x1SGkRemOWGayn1aVKgElK0Lu8OfEMnaIW73WCNMVgaXUSi0ZYbdQFqCuu18aUsHjqfB+urBRidddQyTJFCSfw8BPL45YO7bOAIYPz5VoHOPOMD7/wC+/tezHn1CioI2iZUEolk+fBh6uA1IOaAdQEZsOGQXJS6YlozXQMoiJXius8tzut2s1MsCv3hn2d5Z02dDUgqBhnaFc7/jdD9/w5+9foze/IBlFbgLdd8gaIC4IH5mT4hwz//whcfgvlg+PHXP4Bp+2CGFxIuHo+Y9/WNjcCFIxnOY9V7Ffa3Vqj0iOwTlK/gwl0CrkFBHWrN/jTiLthvNLx3PKpKZwdeY330rutw0hdxwOHadzz2EOdDvHxhbud+C6xlI8tVm0KOxtRuuJ2xsoaH4+CQ6nDDVSaNQs+XyqiJAIIZK0JYhX3EoD6YoRGacb204S0obH6Sv+4feR21uDLSeMEmxvOhCrc+PhOvK76TXhrNkfM/8qDfz7/+GGzf6AVmlF06eG1DuU2HO4bulGyThsafWKVJZaFppONKtJ5jv+8Xdf8fH5Hsw9nbZEWdZKnvCgPKWuO1xXInf6hBZrqNIoS2f1SoNUjX3NOCKhKRAj3mckkpIkslqGvuN6rWhV6IxAoWk4REuELNdJCIJSV1thLZ6sEmVZyLKuPgygBbG+YJSCMw5hFA/zmSDuWPxI0pVJpPVioQWdrFglyfOMcQNNQJOGUhVFrAZJwZfq9l/yMgAaZaBrmmQ06suBIFulxIpfPAVJQXJNdfUuLzO2k9QWVvUjldAUFxRCG2TXoXNDhoL7UkEqbQUmSKEgZ4ywFLH2kUUVGKHQFTQNZb5wpJUlF0HXBFrb9Y1crPv4REHolWJFbaAsUqmVEV08RhaMligkBo2UFjDEvB7gUkEr6+54ptBMRhWPFhqhHM+h45Q7ml2paLcic9MX5pR4roUiKqZblZlxiSS90gVzimgJCrk+DIWkt46CxyrDKAxj7ohVYYBeWXIolCKhalpTWNVBWjCmIZrH6gFSoUoYRkecEkssKKPWN0zVk9IVUyKdKEhtEMsZaSSt64loagPIlCqx9hbDI+hGKGv4pbVIAopUyK4S00IWw0rHKpKGolVWAYq4Yq2lloCWFVMWNq5RbEdQCtcphKhgNR5B0IXaCk2xJnjLgOp6er1BiYbuRrangYdTo4qI0A1ZF0ZVKHvBXCTadhgnUHn1hQutyKZRZSWz8tFdXykZYthQTVw94SWQvlQia1vxrzHPZBKteOZakLXRWkFwgWqQrScuAms7BlewWtObLb28sukccdY8Lgat7hl3W1IVkEHUjrRktOu4nBdCcTQ7IGuixQvtEjHjlkpFtC9qUyNRVJSGmipm21FdZfGeUhJCZKaUyQlqLCtjfdgwMfIPnwrl47iOqXOindY31a6TK7e8ZWjrGDUuASUkQxcZ7JVjGPDaMefCNEX0OVOFQX+ZFLaqyGW1sVmx0MTM4XTgeEi8P3ne/ivLV68kNQtinCnSovqern9NWB7I4bTa4KQmYfl4VTyWwrC1PNaCFD1/niQ/HAYuvIHuhveLhxyxHTRbOUtFansOl5nLY16//9sBpTtOy3f8l9/+kge+Qw97/CBw+kLfJVya+S4IXt05QsgkZqyRfLPb8Wq7pyTBFBQl3aLFM9FozEZ+OQwFfnH86afX/P79O07tNSS9fvdlwamZTidM8TQVUFtNP0rOiyA+9SzVUnRBy5lBeVzzjDajleNaej5fF4q0RKXwrTFIx1/dW/7dXcaykFtBO43PMM2VG7NDVklut/yHSyOFRhOJm33hl28m3uwloTRu3AbpHHfzlsPyTClPtCYo1VK1Qds9y7mQZAOlONTGFBM5R/73//qWDz8HQpnWiUQtpLCjZssZy5OBlwwtwliutBRXW18VxOo4+4HLHBj1PSFfeItHFMnhs+XheccprFmu46nyNHmG3T3/9t/+iNMCKQ1GSUqzDO6el3zL+VrpO4PV0GpG6LxqjO3X/PM//2v+7h9/QXJvkcKtdVDTViy4rmw17HSHUx117Ki1Eie9Vj5rwYpGK4YQFr7sbkFUFlkQYkaJSArrGqAKgRUOmSd8XIhaIVpCpAuxrSuT1BqprATZjdtQwvO6qmwSBKTiwZgvDAZJqREh1+fs4mcQmiYFBcFUG86u09RGxihNo30J668NqqVqliQpUhD5CxMIl2xXPrQRCCmgplU00gIpFYQSFBS+lFWfmGacapQpI60GoYktg2kswpLlKnfQIYGvCLulaUFsmV4bnNEgM9Uo5pxoQiCyxNk9hEKVM8rq1ThVNVY6ZEsoU9HCUVJlIbGw4GTBtDNSBYrRKxu+VJyWKx2sCXpVcSVQ0xWtLUpafF5DglIovFjhEUJGjIKuVJIQZNcTSr8CVL6kO9fDvaKURahGlRmhG7Y1UvEseKysSNNotVABWVa2QMJRgFgrMc9IBAhBLZ4SwHuLsI1GJTWDlwXjHJ0EKwWuX0OTRRZcr7hxiiVNxLaQsmcwBp0zpmWkbCiZKXJmkRcuJeNEh5UClRZcy2itCW19sFDWrEWqBSEcQqzM8SbXH1mWjRMKXxReNGpX8MsR1br1EC0LRkmCsVyFxAlwLaFKoxRFlYFmClmtlr8SR64h4VWBOlNlRhPY7/Y01aOKx2ZQ8cpowPSOJBO+FhAFrCXLzFIytEZtCZk9urcorShZoKSi1sBuMFixdt1zNSgjsM1S8ORsIXve7Ubuhg0UuWpRhacVR54ztXToMrBVkVtnUE4hhePnrGi77RoCy5naBM4aPD1O7KiioAdBqusFxBpW9LSMNKHxacYnGNTtamEj4FRHFhWpO6YSkaWQcyS2TK0R0RolR4xWSNfz8+PCzks6pxEtk6NgCRU3CmRbpyC5eFpTlLolLh1i06HvRoJXYDU1grOKbW8JKZBbJedCbXXd50q1giTKxHw1PJ0rJ9nx0z8k/qf/w45dFwhLwy+Km75QeVj/jBBoo0HdcL3ccpk6rtZwXgoPkydHwccHwzneITYDpgmEc3SbESMLWSayrNRocdKgU0dWd6uwjJ7f//HXfDj+hkm84rII0qJJKbHpA3/zTccvt6CNBSxP54VcE4Yjp8tCnAdeQkBIh/agTOPNq56fP38kB8nh/Jo//PQabd5wYztyzQydZuMcuk2IdKYZT9pmhEyoahBeUNJxzee4E29u7vl2t2dIGdUix7zwJG84BE2vFXM80o8jnW386iZzwyein2hCkltDKwGTxo2FXjp+/5xYxFuETDQ/o0Vh0xWW64LsNyjteHp5IbfEptMMWtOLC6LMJCPZvXvD92fL94dGdW94eF6YMjgJv1GBd18NaJUQKvEcEi+z4HwCIQpCeGq+cnO3Qc4S00aOTTBfPEVKTCeJtZK9ZmP3CBTLYnj/4ZbjoWNvBZh1mng6bfjDjxv+9m+2ICdaEahqV8Rw7Mhhy/Nj5NXtiOk0UGgSivia3/7+b/gP//FXlO5vUXZcA3imImVlrZQ3nLVYNWP1wksceHxU1LRBkLjZWCgzKmVqhYBAKUtWliwWql0dGc9V05kdIUYmUVjnXBoj5Lp6E9CqQHbDWsVWhlYTRhV2zjLnghcSVQWD3nDyC4ySlCtCaKoCnYAU1pBvtAhnaW0lCQipyDnhyGQhyFpSpQJtmbIgKknVlSzKX/Yy0Fqg04qyKQQRkTmTYmC6vBDiieI01Uiiv5JyZDvaVfQjHK1z9Hf3DIKVVd81klDkBN2SUEIjhKBLq82uMxrtBEUkkgJVBSk2rOwQKuBlwKiGVD2D2kHe04qk6AmlErZJRJV0DaqA1mZKVTi1Q8pKaxEpGlYKmlIUIZGtshGCXUnIUDEhUTGUJhEZliZIKGqq6C+j9NIyuWmkvl0/ZJloXKhtIcdIPyoGa8naEsuE0+D0SqOSrYEsICtaG3QWGBEpxa8i15z49PMj49Ah/ZmbznI9lrVmNg4YCWIz8Nwim03FibKmRqMnkxHN0XWWUSt2cqXuKTVAXnCywwnIstJt70nZsk0KYyWKRC8Ktsx07cJ20xFUz6clMpdIaRUtGqFVZCnYZpCtckaQFJzQOAZSgyISxlSEn+m1oa8TSTSmuuWaDVVWTIPqMykHhr1mypHVo6YxWnL2hYOpVKkooRJapoiA0YJOVZxQjLJHSsUwrnpRVwouB4y7JzdoqqKVpKWFTm8xqrDfj/hiMF3PfhiRrhBI+FY4B49QGiUlRVRca5RYGIaGai+osjLHhWhI3dHUgBV7XFPUIGlSkFthOS3c6nuW6FEi01oiN02SkmsUXFMGackpYZ0ipJkg12lXLpCqRGi7riTyCav2GGVwQqGKICdJlQOpWY6LJbY9pR1oOUNbk9XZa/xi6aWh+isFyNXxcNqS+4XOjMi2I0bJNPXM/obFv6JX3/GxWh6kJFhLrgulvaBkwupMLStLuH4Bp9AsFUOrkSgKizIsDByvhf/6J804aJZZI4Xl13Lh7ibQ8hNajJRaifkdL6c7ZN8xmEKtZ6QJDLsNb3c7vn8YOOuRrEEpAULic10njLZiqkKlGRkD2gpgz/lyy+fH18Q6oKzEVEEsmsiek58YX21R7YJfFqK/kHLi4iMxGy5hx5IdS4PeVFrd8PjnF75rjrvNV4ST4PpYieJuPYz8zNAJvnm7YWNgOUcinkRAWsnVR0KJ3N5tsanSaqDfNAb7SJdnxqxoJaBr48+fJz4ky199O/KLuz0BQQqFXmpaNMRgqL1mChNBG4qy7PaOrRmIJ81nv+DcjhQCgy04ZzhdzwghGbc9lj2XJaKojAqUyCAKpT2T5E/I8d8Qjztm3/BN41VjLpLfP3q+3mzQ0VPyM0uu1DJwM+65kQtfjxkjFcPokMlSZg+2Y5Y9Hw8vaGdoxuFDTy8kMQ/8+fsz3/+xwzaHIpPLmUENLMHx448dMbyl0y9ooUBaRBp5+Cx5+hTAQYkdZi+oTZDNN/zTP/2a//f/656y+Wtk3+PjiSoXbgYLwhOWywoxqgKZn9j1jt996vntxx3C3uI0HIrnr0aFyRcOyTArg9Edc1lIQn4hL0p+jgolDVo5LqHhm0FI6Mn4uqrum5SU1BDSrZyKljE18QsrWNqRz3HlA9yMAzfdjocYuSJpQtOU45ASJlfuR4mskqoNEYkVgv04Eq8XMoUrmtANZDVBSNjWEKKtYf9/YbfwXz4ZOP2EqB3TBD8dG6FYRMuYNqNlwWdP61cErNCFa1ZkNdC8J1wty8MJScWIwNBX7AimBjq/sOs32LRHlq9ozeKFxWmLoFF84eXjZ4L3dFbQbywv/hOqXPE4ho1CWcEyT2gi0/VENoomFXFaLzBTCGQh2HSKsYMoDQWDqeCMw6iBeK3s2kg/Ojamx18z5yJIWmOaQaI4hUxths71ONGwPuHL2rUOqeCaIM4zXkSmMCP3X9OkoBWFqhZjYFCVkmakMVymC8Y6clpIjwsHvxDCM6E2Zr1WBXM6MuQrsRlyVlRxR0xbrlVwvEzUHNjZI69vtrzbfE31mdpgEj23b+8oOaCKX2tLtbCkgo8dxUjmnNHRYVLP2809xRWMC+z0zFgStkEzjWp7blxPiDO1RKpeQ03LBQY3oM3Az/PMJMt6OGgJKeBUZa86Bl14pxvvbkbmHMhLYNYjezUypoAdFSpNiC9BIdkaggS1cVoM79NEdoprllxCh68NmzzGOWrqybUjYWl1i4/rZ2XYMgeotlun4KUgsqJWw/V8wLiKtmoNPAmFrBUjG74maKtCujaPEB4l04o6zomQK6YajABBpZSAGnpaWcM7VShy0MRJIcKGN/0O221wTrLUxJXGuUakKETfKGmt9M3+iNGNqRSCsBQSsbFii0ulE4VaBTFIfAiorIltXXGB4eBv+Ps/X2nFY9KCNRui3vH5UbPf3FBKJRVPbpmQNP/06YbyFNFa05pj9nsOx4GlbJhSz7IYSjKI6rDGYqrj6Hvu6j2LD1wmseqvVSZOkdYMsrxjWV7xPg8sw0ipjkuO/O7wCnG1+OagVj4uHxlGz95llMykYvj46Z6fj2+4RMU1T7Q2YdTCKAKvBs1Bd7zMI9lYmoWmDNZaivCMnWGMBVU8OTeqsSzzht//ccPDcUMQek10S02sniAscdF8/+nM3p3x14kQA91u5N1Xr3l4qHyeDeeaUaYnS81z8cwXyR+vV+7uLSaPbJHkGpDuTCNgjGbJglbXy6AcNixz48nDS33NVAX6KjApIGpFL4lRN3xn+coOxCj4mDIveWApisfDxGCg1ESaZvJuRwiNmjXeF6IWTD7TO0tqjblknqfC4gXZVHrdI8qV0/lCbInBtlXz7XrmJdM87KVhdIpcAqVVLssjV39Pazdsx579sOfxMjHPEV0uKGGI8QnfCtcmsDer8VL7iUOKFNmTDjO6jZRzzzIJqtS44ZaXsPDb90+UKvnlzYbvNhb/6DBtRCuLyCt7QCiFtCPPhxt+/GnHv/3NQJOZlhXkDcePhpIsWSSq6JDa0cTX/NPvvuP/+b98w/PyV+zGHb31vLsNvH3XsRkzl+nAeJ0gZ76923ErRxahuPxz4loHBjlQFPjWsC6jcoYqmLXC2UZeFmzN7HXDlwVfCwsOJS2pVgyVjRZY07jMmdb3ZBqlVloTICylJlTNfHt7Ty5buJ7JeO4Y+Hqzx0bPh3jG54qriiZWqZuPV+5vviIERRYWVGEJ8/rfVh3vi+JWWLaiQ6Yzuzax0YlTNaw0vr/gZUDXwDwHYjO0uEGIjkZCdpJUryBW69xudPgwszSBLw0rNWlpJCxZgSwKvyzszdr1BokOniQraRG8XEcuDwW6sup6QyXPkdoa2SUeJ481F2xZ+DwXPp5mxHjP9HxFKShxwcjGbrMnes/5FJBV4JTAqpljSRydge0eUyNp1rjW+Pw+c62KKs68uzGUmDgVRdtqtmZk093z44dnslbsrOZORPZkavEsgyaVBZGv9HmiatBUTtPMx+MjS1QY5Rm7SrKNZX5EjprZFzIGIxPdIlFLRbsRQSHJynZrGYyhnQtVNoo0dPsBuh0+CebLEdHvuUpPSZ74+BNDhlo70uEZ4SLHy5FOJEYZaLXyw4PEty0RQS2wUZWNyOwXieorqRzR3+5ZpooqoAyMt5XRQi8r1BWeGU8g2rpzo0peb0aGeuHddrs6ARjBWHSJjFkwIhhUYGyV3A3YOhBmw2ArbkjYbuAwTSgJtjZkFYwYlIVmd1yOE6enK9oUHGYFeqSINAOlgfcLUlgGo9i2yJBPRN0xNbm2IUqDkkEkcpPkHJCmAzIxNTqlkIJ1BSQlc03kFlHKYIVjMJZOZsYBRFpZGcoOZKmINa24WN24Co1MHYeL5FgFiYCKFSUMW1HYjAZXKsl7lBkR45ZLPUEOvLvf8HT1/OkSSEJQWiWWgmmVnWm87gTXwwvWDphOolnBMb1SfH2/RxhPGXb4HHlOms9px+N05f9869iNiuwbqiykrtG6keOiEK3Hjt/yu0fJ09Wg7YhPidA0TSiE0pQY2ZSef/x4xyf/Gz4+Fq6xo9hM1ydk89QsSH7Hch7RamR0hiIrsTOczgrX7/Gs9D5C4OenZ67jO8JSuFwUj4cbTtkSpOdSAknCVq44aItnPwb0PCGEACQpF6pZ11zbPPFaX1AtslRLbiM/H+/588sdi7qhu9ljlCO0hrxklkuhyQ3H6civ31jEppCyILXIzajwruNEJstMKJ4lntC2QAdN7ZlbIYWJiy6kPGPKFWugqY4yLVjZIWVPvBR8u+Ufz43HpEApnBKoFkhZo1RB+8wP58C/33XEWXFSFVyH0RAznK4T0hRCXjgcPR2JVixNa1KO+FRxtiPJPT9/OvLpCTbDN2B2+HmClKktoUwjl4Xz5Jl8I/pAJxNDb1E0RM3QLKW0lQ44NJqNhHxlO8z8u686vt4qXH3i6fTCYxp5SAO66wgHj5cdXu74eBW8XCO2BN7ZV3TWUHQk5R1/Op4pd/+GWCRuqpxfrpRppjOStt1zOWZCrVQhyEaSxVv+9P6Jv/1ug7JHcnWcXwytOVCVVAUx70j5DS9PHf/hP33LQ/w3sP2Kawm82ifub84Y9cJyman5zP0g2Y4jpnimaeHs4Xb3S9xlzRMZbVYioBTEVsFkSk3IlnjXZW6M5/WNIcvGHw6RPxwaVTdG7dhQGOpCSQ2nB65VILRk44Amv2iPO7rW2BsLemKyHZciGFSH04YbHEvRnOMJmRpVGprVLMDiz4zdLUvM1NpIOZFqBdPz5BMPMfKNrNjk6eqFjfHMuqf+y7YE//LLQCcgUFCtw+qOjKEJRdYQmyKLhEqZQUk2veXTUgiiYboOJyqTL0jTQ+koOEI803eZlmfQUClUFaiyx6otoYT1S9xWfHHNILJjvnrKXqP1QC3w4l8wojBIycUHapXU1DjPT0griS1xozpummcTPVNyRHXLtewwUtB8RITMLDRVdRgal+tC360jmcukmFLFtEdSbgSlUEnh2oV9vzCkSGiS2ATObRko6PTEMPY8HCbCJ1B2QJhGXmZmOyFc5HQ4MgwbJIWwLDi1IwhJihZlCqSZvETmmBhcz5wi0VqaESQxUbVC2ZlaVmiHLwNXJMZW0jVQpOBP3z/TbQamnLnGgg6VsgwsQhGsRQjBpu+RwnKaF+aTZ6HyYZnJ2VGiQrXE3e6Rb98WKAFRe5ap8fACkQ3SFYSxRBVZzg9sf33D7A9stntqaZRWsLpnqpJUKtY0OqWxL5GaBNo2aoiI2thojS8RB9jicS2RW+aff/eJ42NERDBWr5OHJpG6chq3kOB0zYR+YOw63u0l2/RElbck946FhBKCy/VAdY2sGtN04tW7LVIYUg7kGnFCoLWg1sL1EpBC0ncjrmpsTqSXT7QbvRLMJFDXKl790uLQplGM49O55xAcEz1ag5QB0zK2VapPKGMp2iFFpoiIcxqBgZR4vRuZteJcBEU6mjDUBkOdeL2JvO0Tt5txVZSWidwitip2GuZJcDGWGBqeLU/hHi+2nAnc3QT8ZSLVGbRkmeE0OaK0qFyYMdjtFmsNsmuYthogliCJwbCEmZeyZ5l7PiXHNW84+ismZW5HgVGVNgyEptgohRELVSWa7ZGlY7vb0dE4+oLuRpyzlCwQosf0PTu1ZTITd2LAR8u1KRbvqc6CMogW6QaYyShtSWmilitbo7hvR77qjggiMzt+erjl//vHDR/DbjXs5ZkmM7EIrrOkiIGKYpnO1PlCCZ8RgNGO508fefhoqNXhtAUlsBvH7dZx8Wt4ULcrqVSW7FGdoIgGFFpbqHp1y/upMh8DxUpENyKVwRiHEwltwDUoMqOMYDpVnqeCySuqVtmGUWvNrMaCTwu1wnWe8YNA9gMHLH86FK5F8VrvmUrP9+9n3idJN0KJnlI94GnC4/OFljrmpHhZYHAd9+OIEM+EEFHS0sqew2nkdBUkUdEyo1iwYuLNGBnKZ0TKDC2hpWQqmvnpgKuKw5IxztLkBtVFeqW42XTs7ZYPnw8oN9Dsa46xMOfAri48YLh6xZwKs19orcO6HpEcVXbMSfM8bRBlRJQjaS4ooel0QfnEkiSPLxuav2eZX5HSV8ihXyVHsnLxCzdlZo3ZVYRSKLmeKc/PicvngtcGRODdq0wOgc3YoZUi0qjjBqEqu5L4dpR8bUAnD/mR4kZs7hGtW5HDuaBapdeNc7pQ6gZttms1XUpaK2irWVKGVJBtzf+MvWauiuJ6gnZUpRGx0JmMUA2hJVWCl4VT8AzywqAtF7+gpURqw7VMFAO+gugKkBFiZQFJI9DxL4wjttKgpUAbS2tt7RhLTRIV1d/QVKBdKr2FUUqi1HxcIGuFlAIjMr5FhHSQJbFYUi04K0gUalU0NJ025FpRxuGrAtEYB01rYsXtNoi8kE1CaYsI6w80u8q2U0xTRDRHLuu+V46WnezYnCtdDehB8ywEp+rwMWM7h4qgui0hKEop+FypeaK/c9RqEblhZCYDdtMhq6BViR40MhVu9nsOH48clGbj3hCnxPS8UKpkNCNFG5QSjL1BWsexPIHqqa3Rb3ouE0TbU3RCClBKI8XC3kpqg2otc5brjtsaimpUmRi3I/M5UGqGNjI3jQOaEdQGfq7s7wZwHV5vuM5nkrI411NzJlPRRqO15ng5EhJk2RPmLzRCJems4pAfUccHhFBMiyRMFaMHmhuZqiakskKUxJ4//Tyx0YJDzBynCV+h1xVRJIbC23vDvFw5PM3ovuNzilBXHkA0kiONZBqmXujaKj7hJDC5Q5qR1ArKWmKLxBipoUOgOeqFqgW5ej6+XBBiYT5cOegzyWyJAZZp5iLkivXVkTn8wHZ/w/l4RrvK1794jW6Vpz/+wCE2FjEgzIjOgiFBWgpmd+a7X98TvKcuF16//prHF891uqB1w9lIumzYDO8QCHL2mL5xMY3zT4/YTYf96mu86slxAiEhK4zYcAoBUTNO3yF9JlVJqqA6wRIFHyKMxrHEiX1nsDnRqHTbHU9L4RwGDkhCmlFqJJctKUx89pUmG667RbUbkIJjuOLFBh9WO9zQ3+GDpdVGP2xxQpOypKjG0iIoh7t9xflyYsqQ1YBUHc016uhWNsRFk5sm5ghDv17wpaPrOoQY1vS0WOgtaBVXnoNQa/CXM85IOinpZcfFtxV9oC1zVmAEt3cNZxvOCoIHoys3KvOqXNi7THF7vp9u+P/98YY/zW8o455SE7oK8rL+G0ynKLmQK9Qk8FNCKTBKoaQi1XVtosyemjJFFkxvibC+2fcdKjcae5oYyaICmVI8VWmS2hDaHm8g9xJtGvthC9XgY0C1RmcMtRaaEtTVR87sj+yEReQEVqLtQPIztQVKmVBaEFTjKCrzEvi57Phz2JCk43zo+LEUDmnPUUnE9RmlDMjGlsxJGK5qw8fHwqdDpir49q1FecVxgc68IeSew3nk4eDQtuPmzpJVwOdMS/AwNS4lMzhLMo7jyfJ8EKixZ7zZoZNgOgWsztzp1ST6cvxA7m5ZcsEoxahvqctE13f4WfDbz2daGEjWwbYjFsUSC9lqQlSE4rjEjpBGZBM0FEjBcLteGGJWnNOOcsiU5Li/7RiWjM8LUklezo3bW4OUgVAiWlSs0Pji+HhWXHJPJBLEme2rkdYcNU+QA9FWfEy8XF+gVbb7PRtXSTmwlMrTufIURzA7NJHReO5M41f3O358THw4CZpTCNXIoiGEIJWFxCpVK4zU1hG14FrgUgdE6plKI8aZTvVUkYkoPIWiLb4oOj8x3sAlHVBKY82GhsGnRCyAKmhnqcXitcIjKf/CY/5fHiAUCiMMSnd0257rDE1qmuwoLZErGLul5RkhMqPusEoh3Q4tJVsXCcczSsnVFtUKTVb0OBBjREwS6S1aCJyOZCmJuSB0R9EaqwwxJKTaknUiW49Cs0VyBlKv2WjJ3bjl9Ow5XSva9qtFqjqUBKRC9ZL7fuB4zTjZU5eJoUlaWXBqoDRHEpCvcO8kd9ZSO9aUJpJFWZIqJLnFiz21LvTqDbIZQm30b96gnyWHxz8jNgKTHMYOtLiQy4yRjnkyYCRWrfte0PhkkcYQ6yriEVUjSmCrBT6xjsCVo1bFJVWaEgglePXqhuOpAJYWKkGCNgPRr8as6TzTO8tlKqhxT5srOiVuayEKibOS2S+47UjxjRDX7q3WjgikAje7W5yJXOeJTE81imYdzRlabrQkKG21Svr5xO51hzAjSRhmBCk5KI1xsLxozefLCyk3uiWgcqYzilJX8iNCIXQh17UimX3gZvOKcK5I01FEIxsDTtLmC0EoQCPHzepnUHCJnmgMMSVE018OJYVUAwhJFQ2lDNfrTGovpFyppcDDJzYhc3tJ3OiBk+54zj1NWIqsiOFbrnXi73+KpAS6JY6nz6TZ4qUlGUUsAqcVXTzSiYoRjZcAMVfiJcDzM69i5FwzUVRiiegW2OjGdrBc5wteZarSSKkRUtBiIk6Fg7hFdY68zIxGMdQdUSzc3nWsDahKFZVmbslppGRBL7d88j0fpzM5RYyxoAWxqbUG2hRSalwRxFCpzcGiVqY6MPlAKJUsFD9fel4eF6Y5Ik0hKhBiw+fjQpwyKawHZqcNL1dFpyU5SrRyxNYhmwEv8cpx5itS/ZKbyJJUNDmL1R8dFeLao9PIMTsWJYl6gzcCOrmuXrrCqBt98xjTcxV3/JcPjb97L/k5vmZSHXVaaXNbK9DWQJF03R7bOhYV0X1h0R7dMqlWVNvx+NJzDnuusSMKhc+BbApuMISkuB4uDE4Qg6E1yGTsuL69vkRNSoYiHVVGjDAMLSNqWCd7sSKlomRD7zqE9ChZKNYzqSutQnMOOYAxkVYzsoNBDjQtWbTgY4VL6XlMA9L1dMIgc1uRuU6ztYZYBD5GpEq0fc8fDidepszZO6rukCrzkirXpSDKPanB89Wx5FuEGbD1SlrO5Oa55hUudfhU6NH0uiExnC6Cyp4pWS7njMkwYmh+wuhVshZk4fN0oqgOpRVCFDZ2IOtIFo15vKE6h69lzXMMHdk4mt3j6gjnyqXseF7eYbUnt1suyz3Py8iHNnIymsdrYY4K0QrWVm5Gw8/nSKGjtpH3B8/XI6SWEEXTiuV86ni8dBTjQAiKbLQ2kQHMsOZ1SmUKmUOR6AYvsdJpQVFbPl06fjqNnMSOKldwE7qBa1yWK9te8zUDz9JwrZGsKjUnmigoWVCs8DWpdix5Aaep2jAvnpYybwTstKIqxZNYW3qtCWZdmfOC92eiKbRWKBgQEofEolBUrNEoZSlmDa/jzF/2MlBFRSpDKhUxFpxRTFNAaUMV/5uQQTFnSW9WU55UinmpSG3JrbDb74g+I0i0kgi1cC2NyQdu9R1dt6VVsN168OdUkXUNC1UhyTVhO7C3b1jKEV8yzkqEX/vVXkl2G8u+CsIyUROoIpFbB6Un5UCRCfQq/Kil8Pr2FjUdaenMboDztO7SmwA/rRztqMDtBpwbOZ8yYnRcZsePzwXdeh6OZzb9K0ybWGKg1IrgFtBkEYkh8WpUCD/REnTGoXcWpytVFNyQ1g/UbPFBsbQAKhNbJuRMj6RLmrK4leWu4Ro8969fIYug1DOdikhpcTqRSyK2HtkUfazoMLELBblX7G62HH9+z9v7DfOSoAXCfkOYK1JKxsEwzwulFYyyLHFm2I7cjXdcp3UHaXS/7vZKWul4u4Gnx4mGppSO6Rx59Y3BjZanT0e0tbRa0L3hPFdOwdJbhVUSIyNOAa0gtwNxutKaQsk96IGpPNL1CjlaBIZUAs0YsjHIW2gpoppA5EathSVkeuuYuw2n8wXMgLAd6LLCbppEKkkVhUpPkYraVUKJPJ0uaFF5qzUlgxokCxGcJAePqI3WIEhB1JZMIpZMLaAYSXUAVSlqpmpHzBdKLZQkyQW6zYZ4DsTH7/F54mQk/X6HapEcIucieLhe8eqAERKHRQpJqgItYXPX0ZQgUzk8H3Ff33EKlelUcPIWIyu9jAR/RKZlrYCaDUsamGaNNI5B9yA8DY+VZn1ANo2/HimxYu1X5OJWIZRpdLZgWiO0zD9/eEEk0KLHIsl5gXlBxAAvT/R2vfy+275jjDPt8oIebsn6HXMIxOfPsMy8rw/8J6Fw+zcs04JoEqMcshUOh5nH5wut3ZKL5JNJtI1GucJcPDFdEbmhWiZox2kJfK6KOXRcveXY7giup1kFNZNF5iVOWKDXHeFSKW3FhE9J8HzpcXmHNHBNI6fPhsj6xiZEpus1U0nUZ4+2ilgrfjkT0pVaMtvREU4XcqlQHSE7Zj8xbCqyLNTOcfr0xJWCtQNKOl5EZb/pMLKsyPXrmSl5OpXJ8xWROqyzqNS4mg5U5XLN/JAlFY3SDrREyYIVhZIzVfRkmblez5QmMN0e7XqibYTs8DRMb1euRM20WBG2I6L4dJo5LqtMq9cFbRw4yAkQFklb22JOcpwuWAS7uwGTdjwwkJCUkogist/3iJbQEopsZFPoui+a6tEQidQIufZ8aBtQniA8rWlkv6OoNQw81YroNlx5zT88VXzsuESHzwNV7Ml2JMnGi18IZUA0yZQrtW/oashCk4vA1y1Pl4UlF1Ju5GrJ8ZbYdyS1apxRbn0Z0IWER1hDQpD6HabbQK38nNO60vCOuWryfmA0liYrpSW0sMRUuSaFkpVulJhc2ZgeVEMrScyBebmydeuLmNAGbRVdp2lKg1gwWbKtkqF6pDFUqbmEhVArRmuKMYRyRetVrV3qKnuy2mCTRJWKUgUrFQaDFgYl3V/2MpBbQBgFsYOaMZ3EikxYlnXUJwrSaU4YjKiMG4t5uqyjHSHXrr2IaLciVGWDkgpROC7nK/u+0XXrQaRVh9V2RYkKh64aERNGLriucooJqQ2XsNCkxCgIKeGtYDYw+Sesk9ToIRemAubma14+RppaU9gLC0pnpB2Jc0bWwL7ryPHEZSoMxqEQhFQJ1jAvE7sikb6hnGRaIiabdXev4ZhmbnrBMi+0omhqy/WamIWnohA4Win0RmLzAVVG/PmKYpVZCNWz0Q5ZHEkKMoI5BLIRlDRx23XUoClBoZrivreoFJHVIKYLb3aCvhlk8DyFK67/hpfnE0p6xo0BY3i6HPn6F684x4oPC3EJ2KeGuU2o4RYPRD/zaueYrwuxFiqZzjVqmEmPz+xufoEde+ZYWEqlqIAxF2y5ovOIXAotz1y6z7ib1+w6hyAR08Tx+URGI/SGhsAHjx17KpnN4DgsF/ZidSfk1KAEjHGgoTMNmdea1oKgas15ufLd/Zbp04SIlZ3UJDJFw4uSBDViq0CJipGZa3jCdAM+Jqq1ZBqzlwzbHdNSMLLneXnknXGIkonGcZ0zJa6+gs71dG41cV6ywXU31HSlHwOpKaIodIPAiIZrhZIu2K3AbCyn68zz8TOD09QyI+crY6/p5iu2REScMZ3FKcu5VVzO3NgNw909n54OpGXm5ceJKgb6fkDOnscfDyw0ZHtiMyiUrZRSWI4H/AIxC8rwmm60xHiiGMWybCFG0vnAuDXU3NDGMV8+E2Pg1dtfobmlKIiuII1F1sp0/ECyhq9f/zUtQ44Tx88/8ertO9LnT9THz+z3t9zfdezOT9jzE2I54O7ukL/4G/7L9xPtcELnzFIa/+Ww8O3/0DP7E6oJ3r35hpeHmf/+3z4xe4OxV5ytiEHRRGbYjPTfvgO/sDM7pMhMlxOXhyuvdl9h3Z6xU8h4pZIga2QLKAI+Xelsh40ewZmQFUlkmkhYV3FiYCqN46VH6h4nFb2o1LLQq0y8zIQIU5epZkaUCynFtRqMhbigqyQ2x4JDIphPC840zgv4uIK0BBUp/Cqv+fTMIUSEdSiZyXlBtIaUktwGjOypJWKcI9REDqvorBlDzSesErzqe25MgwpT2vB0KbQmEboxX080DD+JiGLGOYcSYW2fiEQ9BJwqLMEzJ0lWGmlmWp24nBI5BWpbKYvaghUBUwMqRXoyrT/hxsSQb7i2Ad0ElJlAJUZPnRNqWVP0dttjtj3PZSXi5VBIcWK0IwVJqhqaxIlEC2C0JhUJIlKk5g+HO6ZFEYtn2A5oPRJywwhJ1o5F7WhNcELjhUQLs6LVq8QYgWgS0Qy1su7xm0DXlWcShCC0zOAcuQkUq0zPGUNIgmtOSC2pSnOMmZgrulNoPLCglIJS2WlBV2bkF4qtEZV9Z7mUtnpwYkK1zEZrNqYRxVrBTcrQ2orcb62ua2IEshlqU7TWkCgUguwXlIJa1gttrgUt18yTlgKjDUiDUA0rHE4ZXNN0f+nLQL0mVF8xKMq1onLivjfIQVNaZVaZc6xcqmJjRrqacNOZvhPYe8fD+YwxktICWkhe3e94OcHzpTLPjss1Y7sL1ISPM8mYlURYK4PtaAlMK+y6wlIC12JIwgINJ6GWTBOR5+lEiQu3xjGQ6MqMKIlz7knuhqYNviT84rnbO7wPHA4Te+M4PC7cv3uHmI+YknEa5iLIujEtM670UARpvmKQNARLi6uPvSwMasPTOfJmO1Jrw5+u5LbuPq9hoTMDKTXEItcJhehJMdPrwOBG+nRFRfA0sIrx5hUlnaj+Src3XF9OtAqaYWUk5AuhSeY5IkdQ8YBIkcEoZjWRwglfCse0opBTCcQWUJsd19mTYmWkMX3+jL7NDNVQjweEULzbbnj2nohiugbk/CM6nXHpSC8dw27LyxQJzdPmE19tNXoSiOAZKKjLTL/JIDKwUtTSPNFveqpMhDmvveKWUFtLSmd2aUJMI9VKisz4dqQzkng88MZt6IRE2oFnLdYqa5qgNmwfqMuFsSl09dRi8E3SjMEqS72cMDXxrjMY4eG+sQBzcsxBU+bCRkuEbTi1fm/C1ZN8h+h2tLDKmKRojN2IIHIImQj0/Q1/87pjofKnwwv94PjlV19xejry3CTzkrjZ7GllRjRBlqt62W0dvYGczwgyvQKnGrOVfDzNWK0QcUGLyqYrzNczRhSSAKTEbCUJT5pOaDTXS6IbB8btnjYauBGcfzziesf2ZstQK+eY+TQvpGtExEYmUZKnc5YUKilLrocTo45cwwuzNAQUqnmsvdJlS/w58DzBKWZQwM+J+61hGRbUMOKs5fr+mSF7uiZZzhPzzx84T4JONJTW5CaYKvz8sGBcIfgj1+sJNXXkRaHNBqUtzSWG3RpeTC0Qf/wdOmayHFaV9xwIS2Z5eeJ22/Pm7T0fP5yRZsfOdGgOnA8nTqFh7npszriYud2/ZlAO0xnq+UonM9NBUo8jw/Y1W2co8UJnLdefjtTZM5TCzb1gUVdmEdm7Dd4v3N/ckCXs5cinKIkxslcLG7XmO2KNxOjp3IZv3r3m/PTMbrBc2oVTWIiLX6d1SyRNE11naS2x2QtSnGjLdf286ZA50qJn9heKFhh9R7vOvH2z41md+fxwpDeanbM0GiGNJA0ojxGNUWlEK1zmMzVNSBQ3IvNWGlqdyNFzWOBaDcoUelsYa0KVgsoBTca2Bv6MLWe22vOb13/Lf/7xBybRf/HwBISA0WhMkcii0L6yxDMv2YPbYLVA1COyk9SsSUEjhaCoK0b2eJ8RQuM4E/yRP/xpxhkDWjLNEzUvSGVxtufTsnJtmlI0IUEUfDBUDBKPVRNShDU/ZjoiEakUolaaUCuoiMyi1+llrhmtFUIa1iVupYlGlRKRKk5KyHFttciGznINHF8jWdZVL06jOkVrC6oIYs7r+ut/EweJSNQFoSwSgcprZk5nhfhiHM2iI5VGKRWbC0olCpWYZtBp/e0KSa3rRHhKF5J0NNsTlcOLjlotumjy7P+ylwElHbVAald8spA69DSx0Z6SFtzdPb72hKSISZOnCXvuIcLm25F4a3l8eCCkyM1uj7YbaJVlWcNOtTqmVOikwilFKgVCpCCY/QkRMiln3Kby1a/v+P7zzCIUtWaUUAgSvTQspwujG8knDzXTyYQtJy7PP9Lt7nmeBaVZtuaWXbdhOj/ThGZOlZSgPB2421m2KIQqHA4TRVpsd4e/NpzSa5c+s2qWJaAapXp8PKNb4+oTY7dFzQmVEoNbw3OtNEIo7PobgvB41dF0h68LLTVaPSBLQrKh1EZsfjWesQZrrBDUaihZEFOjmUrroVhQxqFNIed1zKedQWpF5zaImtYfl2rr/mMYibNCyEY1lXiRXD/OjK6nnhvNVqQ+Y2LEuh0vT0dMjSA0x9MTp3Rk/PoNi2ycpgAl8qrfoNyF6/OZcTPQUmZ+/sRu2FGLIBwv6BiwGvrecZjOaKVYUiBYTc0v3DpDPGWWOKGMpK1ZU1rJq8Y6BVoNiO4Onz3daEhdQW87tL3+/1n7s145ki3NElwyi6qa2RlIp/udMqIis6vQQwHd6J/fv6ITharKjMyMjDu4O53kGcxMVWXe/aBEPkcD94EAXwiSOGaqInt/31rouvMkiVYTf/zwR16nwPvbndoqzh+EQac6phdOprKpym9VcZaFh0eQSbjdj8CoVoNed0qPYCZMdJSUka5I14SSQBLN69b482tjOimCdlQF/+Vf/4q+d2QYjAu8XTtp15juaKPxohpOMjMWtyy83ys/hDP0QXSOicR5Wph95PP1Sgwz5/NHHs4PvKSdZjQuBNwYOLWQcqeZyK/3wv72wuW00P1g/t0nckvcb1e0WsmrYGVBycxyjhRV6dXRiwIsw1XW1nj+GLj9KpziiVYHIg43eYJKXOxGVpYXo2jOsKvGr7cNHS+8j8pD2Q/yZgwUuTO2SsuKoCN2sgR9jNpbv7I2g1UNZWaGc+TsEDEMHFngFANPP31C3t/YamIxmk8RdOmMEFBiOcWAlkwZlfv7O85WtvpCMY7RrhjT0FZxenrk2893Og6TK48fT2xjRbY78/nE5WFimxa225UH8xHljgyPjyfC5Hlff+X0IRCVR+eVYHd623h/+xlvFE0BaaZk4RZBTYO23vHK8GFaSNrSm+Xp/Dts31jVjjtZTmNChid6xcPlCUXBlM6PfkbdwLvM5zQ4nS98fvmGC5F5PnHbX3h/+QVQrFfNeX7md8uEGoK+rei68fzB8vN2pemGNR7v63HDXBPzoqjrRkDz4Dv/9LszP/z47/j//O//QqqCCyeiaqj1K+r6ysXPlO2KUZnJJuaoiVVh18AnBZ9f34jTwp7f0KJQZkYYWKP4w4ff8+16/Jm9bjgNy6xo+99QOqD2itMQvKZXi2mO0YVoMobEPVeKaqAUygycF5wp1C2Ra2JYjUJjjVDTCyd3olVwTkDqYfBVnl4VVmWc0dTWaDIwcKyTjP5uz0xUDViPURq00NWgiEbXTu8NjWKoI4skdGR0QBhO07yHrmhqpTlPK8IYIKrjdGe2MOmG0hw/k6GP0GBV6KEYxrJqg+6dOgbr6GAabdwY7IhOqKi/A4U6xg5O55ntduftfqfpM7sObMw465hU5aT/zm4CayeGORZJxkRGPwx3teyHNvG68dPT73ltd2LVyCoHia5qtpcV8+SZvafWRKmZ9+s7SMNLYQkfiGam6cJVN4IXZjMzva54o6A1tDYYdeH27YUPf2r8/vePfPuvP1ObPoKE7oHzCEwUar/SVUHZ8/cf187Udlr5iu0nrPL0unF7vaHkMFnVMShDaFvj4+PEx6cTW3pF7ZUt7aADYYpI65ihMF4RHzX7tWB6J9B4mhfy+ztlu9JuK746Pjj46cNEve+knDDRY6ym3K/4S6SLBz+T64be75zDisodO51QCBrNcAtNayTdiNodAKJ8oG2Hamgjx62/D4YoFIq83QnWoVpFU1FDuMwTOb2SRke5yKQa81jJ+oG9HyhpnMV4jw4G1yHYA0FchqGGQKqNkXfG9pX5w49kDt/MlhNFBrvOcP4dW7oTvGH+cGK9JbruDK+pStD9RjjBft/wYSFGy3ZXuGioFMSAnwJ5KAwG4yN73TEmoKyl0dAMehN++/KGDRqvBqegWOZIWivTDGUWXk1gLMsBoRoK0zae9IlPT5HUK1/ef0FKwe6B82nmumpu7oydNCpnojNUHGUDpQzX98woEUNn9sLHx0cuw1B1Je87FGH0jhkVZzVxPqqpVjuKKcxLRKocLY5wdPgZE0mfKHnlOZ75px89ble4YZAmrM3iyyOlLJgQqN93x94HwvRI//aGFpApwOhc+47j+MxoO9OVsDw/8rp+IdgHWgVFAK0x83GedfqEtIERYd80Dw8/shdhNpY0Gs4vIBvaC8oPcAeyG6VoOKIXfFR0b6iT8Pj0gdZONG7stWEQEM1eO0VDnGbiObArjVJn1tbZdcH/MCN6YqcwouH9XkhV0fSEjieenz+gRubl81eUUjgf8dLxDJx12HFlCZYPn57Yb7/heuN+vTOGB3vh/LvfU9Lgl9RR2uH1A3XLpHrlrRqaDvxcNhydqKCFQbptoDS6NMTvjF5pvXOOgSyNsd9xPvEn/zvu+8p967Qm2FZ4MJpHFdmk8fX6hefwSE9X9rJSe+YSOms6ZFFJFLNtXBgseT/Mf6dHumsUq3h4OrG1xBItHx4/MJmGyZ2cVrQy+AFlFIwvLBfL+UmoIXArA2pCmYb1gQlDH4Wnh8glTpyswSnPb3+9suSZx9op+473jaA1c4y4tvN4akwuc46D08mQ08r9t//KT/4T3oDUndLb0YO3O+LAB0fd/4rOG/o+mKUTJsVUFMFpjPb84VmhR2YwqA5Q5rgxUzBktLVsRVGGRRsIHnov1Dp4jA4bHd5XZlOZfUWrRM8DKZB2oRKRKbDmSmmVwVEdFAVaCyIDpTOl7gxrcMFjnEJEg7I0OtGdcL3S9xvOOaYlHmyJtGMY9D6w3mKsYQxDUZ4qCm0VGMOeV7w5JGq9y4FBF00pncdh6HSkD9pQR2PKCUMS11oo1iJuItXOmgevBXIBjKEPx9u3GyVXcnbsudNeKm+S2dxO1oP+9xYVFQS6PnaMFHo/diulD3qwB1BI3bk8GMbtyvjuce42sF03slo5nwylC4qNsq08nc48uEj+0ritO/aHZza1k1VBlcEiDi+RHgz7aIf1rFru98TDQ+Df/fjA+3UcD8ZF42Tw/MNHHqYf+Zd//heub3fMKfLtPTFfFk4nBffC5YOn7I5qH+gmkvKdVApdq2MkenZ8+N2P+H3mozpz/bYjSvC28KenT6R7YQzBzuoIWo1BWq/UsfOHP57423/+hbae0doyG4gqUdMNaZUiGacgWEA0vXlyh1phFs+iD030W81UOaqaRhQhHerXvN9Q4YKN/tjZZ4WSxtre0NbglaBNp71uTJyZo6HnxFBCSp14NhgRxGx8+jB4ShvrTRD5QOpC8IGqhbeSSX2HtmP84K3dCKeJkqCjkK6QtAINbQWtNPu+UufBt/KGcobUE/L+GekdvThSaWivySSGHzQnBK1ou8Jy7PK7AW1ATw07GiWlI0zlMj5Y/LxwGyu+N5QebHsHHKkl7IOm9cKOJr19JfVAdI80MZymB/atIrExtCbnRrcK5RxanSGcePvtDVvO/K00rEw47XjohddcEGXQBtrooA+n+HQJqLqRjGLPkTouOCCGRLCNx4eFt+tGkUqIjvD4TNeKi3tgXzfO5wvrPfG+vmCHZTJnvH/m4hstN5xdiKNRROPOnreWD12wDpSkqO6wJ3YixiSct4zWsHqCfoYB2li2nvh81xAuSDHYEMhjID7gJoUWhWEmFsEoRWsVHTKneWGsgiUT08ZsLSNrnLa4IihrUMNgfUD5QovCb9Ix80QtA4UiLhOjJ06qUKWhv2ODT9GgEVLVpAZBDFrDcjG0fjjFnGQe/MKkHbdWKWXwl7crZ6/w/gHbNXU0VE/0dgNjEW1xNvByeyMGiHHi90+BL9+uLLMw0mecnmipEBycbOPD2VOKIn9+oagT3tbjVgnokZnMwTzx2nF6fOZVCrf3K1Z7nFiUDohqfHiw/K9+JrXbkUcScEaYJuHHnmmyHi0TXzBesLoRzM/IA6BP5FLRqqD1wbR/GYoigaeoyf1XPsyWPVe24vEO/Eh4F9ilY/zKMguPwRNsQPXO4jLL2ZKSRUYnhs4ydThpRrVMGpzb8N4xhQu//uULP9nEj/NA1IYJA+MMIRgkZ062MUXDaVJMi2JUwQyDcZldFr68ZFo1dALDaOYlMllBtZWUG8M4QOH0oXMWGtZ2TssMKlBEcFbjfcepTDRAt7ThyHJhrxFtQbMho6CtRfmAc+a4tOQroyRK2tEy02vg9W4oXLiLg9hZvKMIBG3xzrPmG0E1TF9pY8L6ww2DcmAc03w5bvZjEGpGceHWBqJBO0W1joHifFqwzrGvCZh4ey+gBrM7pp9rVehuqaumi0O5mVyFlhraqCNoXhtrFWrf+HjaGf0L9yEkrZkuJ27rHVGGW+687hvzNKNLPay31aGyRjcY6wbs1HmnLYYy/s4EQqxjyHHLOC8g2zvnELinzD6g9obqg+fzEyoL45ooGpIeXCnkLGAgRsV2fSfvldxuXHzg06c/8p/ulXVvJGOYpgkzCxJulPs7zBd2BiKD07QwVEZPhp8+PWCM4l4aLlYciTW/Y4zlw+9nTIC8T7xbS/aO008RfvvM67d/5hweOJkfqKMSZiFTOX/8CFXAFK7txq9fvqBq42k6jFKMjQlFHNtxuuuW4coxhgyd7jthCkRr2JtDTxoZO20/GhQDQZymqAqSqN/eEPWIeINCM4ULbf+KpMaH389wXvi4POLanV//2//JeYl8HeuhzG0n1powXmOUxkwaG2d+WJ653V7Qtw2116O7/TRzG52SjvCmt4KZ4BwdD8bz6YPi688rIgvWKIxReG1BR5ABpdL7dJyydUS0Rg1Led1wWuHsTMdijUdCp6kJJRZVO3VXsB9QE4xhKIsLjn17I5wmdBfS2hDrqHZw+vRw2PPSC04LP358ZIzGs2s8ugoU1nVgHFhWuAT+em0oInZ0Mgu7cfQ+HSna9A6bxjiP3hNKbehFI96TS4F8cC6u/ZVW3o+ErghdW4rVRL3yDyfPUBuSb5wfzmQzEO243n/GVEVXAamJyYGsG8okxqhkufFsPFHeUVrRu0MGMBSmd+6fN4ZElmkiGI3Vg29fr5wnjzSh9oJN9+MG7BZa3Y8bmFSsnSAnZCTO7JzPMFpirwnvJ3odqG7QzaB0YtyFgMFMgyFXaj3qlYyG9A7yjcfFoEYjb5klOMx45zFmYig8nhWzdaTUmDaDNyfehiBMaKXwdGIXRtnxbTCFnecH4YPr/PFjZJlP/OXtxvv7RmuNVK40Zclj5poqUhLqsiO8EaffU0phCTunsGImw1o61wROTXw4XYifLryumSEaJxPpntDG0BoEA+cwcZoWYBwGx2ePM4eCt7REG4MlDB78zmmq7En4n//B87puOFeJUdPaoNVC68LbVSNWoVQiPp95ejjhYqT0BESMVMZU+PFjpOZ+hB9b47LA09yZvWNyE4ZDf34clY6XHcohHUYx9GFp2iBmYi9nuo60kY9VjRUGjtIdrQuMHWMclYWExRqN1QCGPgyLnzHSMePOZBrBg/eWVjyuBbyuiCSU1zgv/O6TZ+QBqrPLhY4AGhtO0C8EU/FeYe0hyrHKo8RjtOOWNPPjQqqGXDSnhxOKxMlVvCoMZfnLWyV3g7YTt9udGGcKms8ySA3qaCwM/rAYenvlfRPWpGjqzN++DsSeUa5i+sAOwxQVD4+Rk5nIpbJoQ7w88vl2p/QTWS5cfeD1bVBkIanMQ4iIGrQOVhk+v3zjaYlYGWAWeoPcFHV4ttxwJvN0eWDkDHvFTZ6fX4ThzgfnZFha1wd/Z8C+B1AGYxa8r2id6EMx9BmLhWLAWvCO2jtOayZvKMOwirAqQWEY+42Ld2AFsZqXXrlqxZ4SOMVsPEEbEMNQAa0NsQq6bRhnMcphnKbTSOXvfRiYB7VqcnV8/MOZ6fXKePnK5BzvvfOqMqUAqjOfYVt3Svdsk+VmGl0E3wSj+sGfb51ed7RKxHlhmR+4JUHHidEKW/7Gx3PjfruRs4YwIaMhypKH4z/+b3/lw9OMYuF6X3GpMbnEabaMljDGkK+DtxvIcmEbN75svxF+sMTzGfZCef0Vy8yuBjIHfn39BYdQBH75emVSnovaeZAr16ulJ9jPlf/pdws5r6xZWBxok1hzIksAfygylRk01VBWcV+3QztsJnKHLpWLs6hWSfuGlsej3qIae9sQUxjN8HT6kbw1ugSMuSAY4kOkdk1Hocph91Myk9JCb571lhjvDa8/UK2grOXyNNHbnaIGenJMQxADeZ1YuyMoxUdXKD0xCLQkhzzHHFpbi8W5zte+MpSlYghd43CMkVEOqhSaEaRDVg2FxnTNGJrL6cQ1FazSGG3I3w8Hxg2cruxFaERqdSg10WvF6Uec82zboRb+eImYKKT7imqDBxexaNqAUARrJvacqMqS6mHtU9KxQ/AlQf6KqjvOVpQ2XNPB9z+ZhrE7wRdO08YchMvpxHYrrGXl5Bt/+uEDp0UhuWLnRBqZ6NtRw7KaX759Jiu4zJHZFKao6SOT2zsld7TWPF6eqF349vXO49MD91vibROua8KeTzw+nWgjk/bK8/Mzn3/5zBQN0Qbm2dLSShPh4fEjb9cvuKiP1PsMblSMroyW0dpgg6L1SpdBLyvS7zinmIIjOs+gY0zA20qvx8TqcrFYs5HTxm0bOPcDbc+HKlxp1j3T5SASphZJamIrEWVO6NDwZmC+74gvvjHpnWgrwTQWU8hjgyfHv3ueMGrDOoX2mSGZXNWh0dUOURGjHD0FRhNGT6AF5Swdh2AxsjL6lfFhojU5bJqcEb1wqxprJlzrjPzdAaAb1oGPQggRlJDljtMZqwSUoOcAzfNYDhGNtpXRx8FSMYGtn9h2zbrt4CeyUTQlBy7aOpQknk8BqyzWGFCJYIWe70DGnCLGQjAVo4WOgqFwdFLOlJrxJwNlkNtAxBJPJ+7dU2qkJUMjoTHMpydSU3TZcVozzx8wTX0XvglGB4zyGANWV2a7EFVjCqDV4OVbZl5mTkExpOCCRjvw5czbe+fPb53/+JedUS3OKk6nB7bbjS47MTqUPrJPcTmhMNyvCYYGMdy3Th6auHS8tzwvgcUK27Xwr18VV63ZdKdUw2KO79De77hlJmrF4zK47ZVvf1n57XbmnYhohR0zMcCgYUw4bLWq479otKqMkvFUHpbI1y8PhzvHNoytjPZd9e0t13uj90IbBW8Ho1TuxYAerJIYPnJPC7k7hgpYrfjt9sazU/imSO+NtX2khgtpVLw3B6Oig5YGrhCiQtfM0/nEljSdQmcczy3vwE2810P6FqzCq2ParaPHqILTmgNiMfBaaEaRlFBHo7ZEdBaNQoumN9DGsZjASXZ0qRjz4X+si82w9FL/vocBrQtNbTBF9t75h6fIfX0ht8EyHQeeJo1ab0ynE9Pvzly/dDajaOZI/bcxSH1jcYFpmTHpihoblDfOpwtfqlC70LKmmoXiVz7+6cz1bbDuK0E6NmXaPlO6J+WO5YZ1wmgF4wL1fccH+Pwm7PsF6yJ9lMNT3yde3htmg2cJqL3j3XETvl6v2IdP1Bx52Qt6aDbpGF85xROX6Uj8Jjnzz193RE+McTyYbCsEf2aVwdfrjeXjfJDdTGBYTRkdazwtaxSBXBPNW6wdOFUpqSImspVB1BNKa9rWSF/uhDgh9oTxP2HdYb56NI77ezlu+M5RRuNWBJ0ypilctYfy0wY21blXoelBdxUF2Mmy5sobhtIXWrvz9DxTh6ZbeP36Su2VoS0aqLmAKpyCgjYoxqClH8HJ3tjzlagj0ygM1Q7pCRMyFKNXUt8wvbIYjRR9nMJVZwrCeXHEHy58/vZKl0iuK4qCVgPdA702JrcRm0HJ4PTo6bIx24zu73g/oWsh5Ssnb1CUI2BjDuuhE7APhTi94ZzHGUV0wsMy46XD/7UfqxjVcdrwuAhWXelloDBYJZxOG95ovHI0NWhaMfo3lBr0LrR/VAwljPpOzhsDQBuaDMrwgEXrN8Dw07ki4xfkYqkqkIrBuYbSb4g4qrJgrvx+Uqi+Ihq8FcLHg7qn9Qs//GTRpuPchjFXtLoTXUPrjjUapV9RNIwW1NjRUlGikDHQ2iEqMJQ5AChGYa2lld8YcucUG5fTxD17xC3QhVagKA1uOsabt6PjPC8eVEGbju6NIQLOHiyE0Y9AoD1uuhuer9uRlZ5i4xQcp1CZ3J3QBnuxlKFRdsEOQ6/Qv996cJ6uBK0D2i5oMeQ9I99ZCCE6rF3YR8SIhW6OVPijpSkFkghxMEd7vDjMeiCmVUTryF9eLS/ZcrlEHInoLdd9PQ7c+sy3EvkillUKwU04o/AnR1rfeH58ZI6eNSeKApFxBKOr8HJv1E1j8cRNIb0zzzPSDNIVagwuXmPUTuMQb/UKwxiGNpQy0/AUJTTjcQCjcy8L73tDu8h5DpRVs67Cno9I2+QSD76Q8speduaouCyOUzwaHW9fCv+qAkMMuXFMHhDediGlQXEzL/dHNEJIirVCyYIYR+8KbRR1BVsMmUrLmqgX8ibk2lDessmClE41gyct3LbCfQh3YyhG44Ogx4ZTwuV8JjVFUbAi/PVb5tuXQJ0+0Kxn1BunKCzWsO79gIQFA8axNkMToQ+HUZZf3xtRLaAiSjeGvDJPmtEyW3eUHuimMmxC0Th5YVYCqnEvKy16xFVQgneBUTu5JO7taByNrhn7O1OcsO4Iute8sZwWZue5b+uhuV8quM7bywvDeURpYlBYdu73nY5GOUXrnb3VQ8Utg3t5J9hGpHDNO/PiaUrz7e3KLW0oNai6HRAjd0ZEHx6iklm8hl6PPMVxliDXyqT+zpMBoyyiNEUpZNsxk6C7QQkIHeigPfd7oqCYlEE7QA28iux5o6hO8J5SM1YPnJ9RXci5keuOMkda2yhLlxMvveAXxeWTp10bbat4J3idmLSm5Uhq7cgTiCF0xTIvxBDZrvt3CIygEU7ziZEGiKFiqd0c/5auKfoQGel8xe2ZUYSuPYLj1if2+6BQsHGh2Eipgdw3hi5YEzDykZEa2e58SzsPNmAXx76vmKFAVVK+w5joRVBas6eOGzuXy4LJV/baUXS03Qi2MwLk/gVTI/qucfWQNaEE1d95soNw2uD7jZhRjj69N+SasWFBhmbXB6bVM3DuRtvfsc7yNFnaPbPljnPt8MNXwU2O54eK6IExCq0KjJ3RE8ZP7AW08xjbuSyK2XegQFsJTjPPkffrytozXTJ//EEzmW9cZsd63xEUPni6aHqpRyo5NnKL5NrZt40QA3100MI8e5yOjLZiDPjJEPxH1FBIb1TZETORciUYwxgRNJyXCCPTW6b1hlIdWLFyhPdKvmKUh9Hx+tiba61Q/XZw9m1BDc0Qwxiw12OsN7Q7oCTt8KKXVtHG07CIzJRxAFjQCvxMY0GUO1YPzR5WQ90Z2lPQZC1ED344ehOU8khWGNyhyG6WtchxGNERqzRaGaxTSNho5sJQK9iG052uOzLuBJtwoSN1R3XL6If4qyOUoSnVoW3E2AeMOuBPfQjOCso+8bL/iZfyzHodSPNUNLnA2wbKnnk8XfCuEENHy0Cro189bGAHuhSCa0xhEGcIXWH3jfPsiLNGSWUKYHVmKKHLmToe2fNHhA+U2qllp4rQmVnvg6oc3k4YhK3vdHWMli/zQsoa0RNbzUwhYrXgo2eyiqgvKHUcaBHB+ojiTPRnrjchi8PNAa0LwTjmaeJWG9dN8+XaeckF+/SAmiZke2G0G2U7tMwxwnq/cl0H9+thTn364cS6N7ZUUSJczgufX1d614Qtst4Pap2VjYdZ4f1E7YFchL0M8tDkrVDKNy7nC2JAhiFXTeqDot5JPbNME4ZEy5meB8MapB3q8ZMptBpp/kKdLENnooWwVeL4wLfk2AY0NBqHmIq1Hu8NYx8MVgIFnaH3HW1W9prYh0KJ4LOi1pk3NyB4slnpWztw6daDFtb9FfKNx3M8AoKzI6EZXiO64FRjHY1SBMuJVg33ofn2qlDhE7uC1DYmM0BWIGHMimjBh/OBakfoNaGcptad0TJ/eH7k7WuhSmGoyj4S/mPkljJv6QuiFF0y0SiiNhgOloalcW8bMhvWL994Dp8I1pP6wX25K8Hrzsl1jBZar2ijMM5Ab0fcexwyvkaiSKdqw5YL0xwxTuEBLYXaKsvpfDS6QkfZzu36Cmaw5TtDCjYENjW47ok6GkJnUFlrwRiDN8LsZ5SOmL1QU8F4h9gOQyHaciyG/85ugi6VPjqiGr11BkcSXpRgTWCZPHsXRmvsSWitE5h5JNLEcA7z9y+bJzSL0w1rB8FZZMv01ohRM0pntIM+KLpwdYPqOiYqlLIM6aAcF9uR3lBKEaXRykq0jnTrtHslqomOINKYvWM2Fesqt31lpBVpDc0Z5c4MXbiYxh8/PZDXK/vW+Pp1YO2FaI4QkR2anjNGv+O0cPLHqVh9/1A06qGOJSP5ysXf+XjyIIWgdtS4obTjWk7cZDpeSuNKMDc+nS683q4MCpdzZ/LC/PhII6H6HaKglo6xE0ODHjcu/hBgoAfWDKIVluCxBow+YXWklIZow3nyaG3pMuG8wliFqIqMo5drLShl2e4bp8kzR4fVR5hp9A3noYtHK0XadowbWHdIhzQVay33tSBqwpqO8wNRb9T+Gc8LWt2RPhijovB4F0E8edegZkRVUPN3yAbUth9NiXGEIZVqIILSR+5E61d6FbruGNUYNuPdkWMQcXQGvd8YRZDhaO077KR/f2ntjtZmWpXjZmwMvRtKhZQj1zXTitB6R6nwvSblmaLDzBN76WxbRduJVAfoiSYK6xpd3rG6HeAte0Z7z7CJ3DK9RpQxBG2pWJoSjNOspfBkP2G8ZQCVxlAFFzxbOmQl6IZxBu8151NAmY6xz6S+kccNM2lOs2Fdv2LtgnOFGCuqTtAaMjoyoOkF1WZ0n8B6xhQxqvLx9JHb9TPQ2NKZl/QHvuwBe5kYOIYIo0DRO9pH3p1iMQvYdoiljEG3AX3jvHQu9splWYm68/D0yNvbirFXfnx2GPmKqJWhCk2Xgz6HEJTll6+FtW/cs+HrrZO64n3NpDExeU8cK1Z3rBo4s2KD4e2b4cvbwKidaMDqlSk6fNAEo3FamMLxWS35xjCV6eL4+aXy5bfGNVfMVNnUnZM/zHnvq+G3N8e3bXCvOyr/jHFw5sbt9o0sR4BXFJTS8G7Ba8MULdffvrGvGYtC2uC23tgSoE6gE6UZkIbrlb98PV4UQxnWZBAVcF7TSsdby7VVVGnM9sxaOtVZkjSqAi876/r+3TMC1jhae8OGSvdQbxv3e0PxERMtq1REK9b9yusI5B4ZFZZ5Zq93nuaZvd/RNFCF1/rCcjKgN3qtSFSs251FG7oeTGGiZ6iuU9QNHwezO7OXK23sZLvD9sYtRLqqpObpxqGGhp4Ra8k14RRc5om//PJOOD1j/ETuGrGe3BrWdJoeNDtIUmCO3NI3dNeMqtFWwxCUSggr6FeGdGoFsYmmd27pTnaWvdxR2iIMWis86gimU8WQtObeCzVdjwPr9RceJBBK56Y1MhlMgHAy7GVlq4KIA1WxupLf3zCaY0W6OKoJ2Iuhbzfe6s7H8wVdCsqAUhqsAqOpLROlsyiFBEfqCspgoEkpH9mwqDGqI1pQ2sDo1JTovqCV42QNXh9KcWXHMZXUgnSF1n/nw4Bq36EL9gBF/HZL9GEYVEqpNG0YQ+FMoIxBJVPGhqYTZ4dCoYfQ0u0ItXRBB0O8zES38G27Y+srFxs4P8xs1yvzyWDp1LIxuYnUd7oS7nfBz4rnODjbzmJXJF1x9sLbtbLVb/zw+EwuYN0gTpo5CrOD5lfcs6CM5n4r3EvmwTREVy5mZcSKnSf+4XGg1I7VG6dJ0Yfl7X3nT7+7MMeOUo7UAAaagtaF8xKJxpPvb8Sp8eMP/nvQpnCaDcFrXl4V//nP8Pz074mh4V1HW0dKld7aMXJ1A+cmRu+MlvEGWklobQja0PbjC2L9hDZCGR1tj55tbwWlDEhBKvTvieahDKXs+CkgRpFHodYNZfQRsBLF/LAweiaVndkPrN5ROtFapymDoBCvGGowRqOUgTGWMoQqHukL6Igen7HqFyb1ihpvjL4DoJXCaoOMTs0D4YktPXHL83dBjMZ5TxfLGJpWKko2WgNlNNYb1PE9oHVFFc86HF/fCtINs/WcQiS1Sh+a4DRBCl55nFVgFcNF7ilQUmT0FaXHwXbH0YisTfNWdoyf6c0wzMAZSxsdPRTmTdH7Qtf2qNg6iziLNo7cd7qevqNsB8pfcKZhRkK7TguWdTSqP1NLYZkMwWqkdsQ9sgL3WqjG0/oETdOt8OAthTvFNu6S2WrFKo3vA6cLxgkqGK7lqGG5MOO8p8kNYyLOWpyZeL9a3tOZ25jxUyREw+gZF+Gv7xvXF7jfN8R8YK0PoDRVeVJTXKYTX16/sWXNs38k1ztnq9nWQhkQjeZhEs7xC5+mn3lYXukq480PdJmO25x1iMrc188odXxmu4XSB1plpG/M6hFvPJI7+As2fOCvqnKTxh+fznyYFN5tOHXnFAzeev77F89aMlo1fvr4gNWGn3/9yp+/3Xg6P5HzDWMKwRu+vN3Y2FBusL5tYCeaGkz3wZNtWDV4/W8vfHn3FHtCeQfmQP66dKwN0/2NZAI3gTQ6xrrvByLLty1htEZJ52Ge8FZTWuX+NRO85fx4BAJH3fBGYa1B9JHv2OqG0RMKizOaJo1eG9IyXgekKaoIeBDVSOnOPf+GcUcdLWjHxjttDLbSiaeFWyrEOuHsmTYaIRgyd2yI7LcNJWc0g2XyjLEzT4Y93UmyktVKHpmc7zgbMFWjtQBHndlNAzM0SQm3feOkYTEB0YVUE2IHuSU2MRhvDuOhC2ivUF3Tyo7uBaWEIDsfJsPeE4Sj8196pdaCDY4uiYZQZFAlcx8J5yyij9S+VgN0Y14GX7f/BsvMWhVJV7QUeu8HmMiOAxrZBW8Ma0msRrETeW2D3Qa6BmTwyVT+wcNqLP9py2j/QFIZ6za2UdiHxgQPY0d5+MMHB/edoAxrdfz8lshw2BI5qvhBMtMcWbumtsLYE0/nyFlbyth5KwWlDUMp7jlTpH8PbHcm5w4DYtCUXHnwC6EZGMKsj4CvrBvOgrL2WAEpzVD/tnf8/x844p0hoMfxQ3qlo83GZHei3QjO4OU4EMjQOOO4zIZUr3QtGB1pZSChYpWwTNPhKBgveG/5x98pLvsr2li8E5aPcrjujSLlldO8E72liiIuliVsTGbH6IyzmSl6DIq3d49w4Xy+ID2xBDAmoVXFiqKXAVpo6sJ//u8K5xfmh5lUNrRxWPMMrfIwNYxecS4yVKb0xrbDZW506dTRmMJEaztabseN2G702mjPCq0hmr+hzIrWBW8yTnV+ePgd6+N/OEIxzqKtpfUdbyvKDdBHmh0a6vvv11IxQ6GlokdDsNShkHEgTstwoCcY+hAtNaE1dbxEj1gXXTQyPGbA0BUx8xEWHJreBijDXj1p60Q7MbFhR0Nbd4ybJFLLISkxooE7Lji6dXx+zdzuHS2WP350PC0VO1bOi2UOH8j5BaMF5wM+LKS88Pb6wNr+yG9vJ67FMpRFe3182K2ntX5UyPSOKOhKU7Mml6NK2tHsVZGqoQyHs5GzVbgl0HtGuqDioMk3vK4oo9AcUywzjhc5CsRYhhyjtNYNkh3GP1DaUS3SZtDkqEFttWHjRO6KYTTOWrydKa0xO4Ml0CnsXVBa8dAVsxn0Vg75k1IY8ZjhuLjBZCvawOQtsyuIM6BAm4ZSDpFIaZ2AYAmE88I2BpsMah9sOdOzZisHfZNS+OOHGYfl188v3O8d5yfm4HEY3m8zX24OmTUfnwL1/Z1e70x65m+/ruxNM80fMBKwTqP0QADvAh7P//L7P/G6XtEMPl2e+Bgrs/P0+kbnyjzD86kzWqP2wugKCKi2YGxBuUHrb/jgMFrTe8Fah/Oe0aFTmJeNUmeeH89c9IlrAl13/vjxhGsvOGXJ2zeSSqRs6Wnlr98U6yaEYPj5aybviS8vO/cKv60rQ45LiVKFVDvFFF7urweUK85oBjKEIIJG8ed3z1vxGLWxYFEouurc7u8sAoNEFcXehGo1o67kvnGyE60XlBwv82/rK9YJzgbe7jfmWZO+bjRRxwHfGPqAnHeGc0hwXLcbpx4IIdL6jgoOHQ1ZF8Lkud3vYAyKnU4+zHajoJ1mbzt5FPKecRp8PZ4Nwey0FQbq+B7VAuM3nGhyTsc+e/aUtpNlUHuhj0JNd4I5DvxKa2qt2CFUGk40qdzpY/ofWYUqwtvtN6w/fAZ1zyjT2FSjlMxLznQGOsNYbwQf0KNz8p7tuhGxiIaXnOjKEeyJ2js6g+hBKsckuO4ZGQdbRauGolNGAylEGfRamBchl8Y6BGvVQTjsR6NLiz+eM0q4qspfG+SeafFCbYmuLAHNLJknEk4HgrXsCrJ0nKwQPEU6aItWndvoPPoJPxWW1mFXpKuln2dcUDQR1BC8UsxamKXT1MAExawaizLMCj6XO9pBYyBqUKQxRsUMMAqcCww1sM5wmmbYFduaUHFCu+NAN4bQa2WYjlIKVeTvexj49DDz9HzipC1iB09T4cOyHOALJcxnjdIaCNw3MNYS3QEr2ssgZY+TM85o1Pe0vVIVqzLRCqoPRI6bfGl3DAkf+T6y6lhzw7pAHw0lV5xNDCkIQpeKEoPjguoLJV/w0og6QbljvRzEpg6MRu+Kqk4s+um4Xd42hMqwHW81fmS21xemqWFng1hBhuCMI6Vx9JmlQyt4zaFfHp1aOyIe5U7QB4gwRNBqPtLPqnLfL/z28hH8mVDkAOj4iKiOaENv8L6/IeLRdqKN5fi7ake1ilEKbT5Qx0CLRotGKUftx4e+yeB924HDzqeB6A9BiFEOoQEOEwyn00xXjlQOHOZ+HLKJKaGiI5gLXTSlHw2CrhReFgadWm947binQvWecso8zpqf/lAIvaBkJp48zmjYI0ZlwgTDLGz1J97VP/HGJ/JphjZwzjH0oIqQ+qCZgQqO1u6gYK+d911TuwFtKb0fI7YxwEyIWLIMQml41YheUSUT9UweiaGPg1HaFOsIVAKiNKIUo0Frg9IGIkKTyJ5hcgejXQ49DNHMDA0fncf4wPCOlhV/eD6j2jdUvTNMJTuHN4r/+1NB91+5r+84pzHBEeM/MJmNy9yxQbNJpLSB0xlvPC3dMW4cwbGhua4b++2VJS68fVv5Zat864qXNR0BN2XYcqIbx/Wa+OO3yv/86YlfPieGXih40BpJx4PTOcuf5sBzzITJs40TJTuUdXhnmU4nWir88GR4T4Nvt8J6r+AKv3u68P/4w4+s7c5FwdQ2TvGOme6UfEdZqH2gxdHbBWdmRAJpz1jtDva+q4gMctmoJeHdiVIHmgutKd7WnZ9fV7asSLKxd3ihcbsmZHvh80unST0qdlVRkuG6dbLSuE0QnbmnRMmGfeyU+gZqx5hKCIZ93A8GBhvOHYl73Q6L3LCHR2UvsCGonpGciSHQageB3ArnKZLTYIoBGCTpKG247zc4uF0Ih7429YJhoM5A7NjZ4a2m7ztFBItBagGtjgppGATZCVIxRvNeE9I0a9+gN+45Yc0EYfB++xVlKiKdkRpxPngHYQ6kvZBax0ZHDxvTEkm/VdCOYAJVNaJNxMkyq5X72zvmDPdeSD19D3ZvqGAO4dy64ryj9II3iltKNPGMdgjsGI0qBW86eVupHCvMRuVl7+QxuI5GTg09Bou11LoRjaMOWFsmuIIyCikNMQNnNT98iOT1nV06UgYPc6TnjVoT3YwDr26hjYLxkMtO7BWDw9Cow5AzTDHSSmO0xDQ8MU5s9ZVkKo1xOFBGxuFQStBD46g4FJbOZD03GXSOvytJooogzRKMIbdOqvrgfJjOw6w4VdhUQblAbpX3vPF758lvK2NUsB1RB+EzD6G1hg+GrDPdH6CgGGbSvoI6sMh9CL2DtwGGZbsWRE+UptnF0uxMuSswYPXhOnDj73wY+OMfLiTleJoND3Ph5DeCeWG0N7Rpx6jGHLeID49nerVIP8zNwUauQzGNQvAcJx896LrgZCXag/es1U5rGck3kiRMCIyR0aofCehhUVSUXmm9MgSUDkdnVwZ9FLR6QHGCHlFGo91h+Rsix8PfWEa35PqIPf9AJ1LHDmZQaCgb8XjwjhYgW0Nl0LQGB856rDY4Bp47ThW0WyjlCnSGcSj3QFsHSilGTdSSsGEit8jr6zPXbSKqQQyG0oTWB93a4+FgJ5INpCaoKgccwwvzCSiVPTXC6QHVO1RFdBN9WEQ2rOnMVnH59EwR6Bx0OdULQwqzjchopFx5ve2kzSFqomZ14Dxl5xIbz8Hg9EAZz7Uphj7CbOI1+Jl962QGu+KATTmPC/57oG0lhvk4oLiBthXrB5bBYKN2w/39mev7maQM3Rw7sFoF9Hd7mAyUgl41Tp2Pg5x0JHhMnBjf3eZDHS/xbg6q1z89Wv7DMzjd2dY7Su3oPkjjQm2O+9qo2qL8xKQV0VUUBxMA0eRWUFRO1mNSZvGF3K7suRPMwomZrnaelgeqaP5SGuHB8ah/YbZv0O/HVGN65vW+84dLIl3/I0+PBWUaOEP0O1F+RKTx9ga/vEf+9rUeB8wm7C3RlCK6M1+uf6X2gZHBk8/UurNZyEHTqRjvQA/2+0YvjtYHNcOXL39jiULrllEMmxjifCE6R0qZGAxdPmOI/PbzG/d9IEPjo+d+e+MUPaO8wCaoe+Bj+IBRDacS+37nX79+ITD4aZ55z43LVKllp1wLe9d0sYhMaArn4KntBWcD5/lGrle0s9QyUceJcX/i64tmTxYfFvDP/Hw1XB4+cl1fESsoGiHOaH3Cqs56T7ysgvEL3Vh2s2GjxYeJNW3se2I6R2iCEodyg94G93ZjuIRgyCPR9krLmfM0HdQ7gck6ohpYdbyYjNV4bzBdISaS6s5kFT46brWQW2NIpymhtgL9CJVO04SoxuAIWI7Q2dVGyTu6NqQXzKgYUfSu6WUhdYVVmlwrRevj/4djaEcbCW0K7qKo5pgAdN/x/hD7IAat1VGdHINUvq8KnOfWV+iKD+7EnAqvpfFZVZQqfHADYw09ddatUnzDBcXSr7T8jm6Xg3ugD7qh18JjjDgsUSt+P3uyMrx3zT40ta7Q+0GAVYoAWHPgFKKBvN1w1jM/nMits143lg9PTET2lNmlkm0/qnhSUdxwJ0vKjf2+8cNZc4oHzOelZNaeQFmMO5Dw2oPKg9Z2YnCYMo7vRWv44Cn3V570gl2Pl3yylr03JA56uSNiEatpDVwIFCpZGk11Kpqtb8ToKaVQWkGFBdUaQTpSNHVLhEtE6sYP8cSLVL7uhdYaIxr0WWPyDvtGPF9wLqBGY1szRRlCM9gYuMtKHgWGPi6KCmAwuhxZLxeow6JCwLAgXdP6ARIbWqELTKLIruL1v21P8G/HEVchGEeXI7jH6NR2p7V8jK1cOHq5xVOrxjSNGRplhKwtb9tMZ0b0kQCtqmEtIJ2RK2oUnEnUUcA62lCUZlh8AMkorRhasM4cSUnxWBspeaCUxWjLaDP7fWKtE9YuJD+INtCK5n0dvLxktH+iNUvrDj9ZhmpUgS03ah88uqOH2luha7DTTBVNV8JAsLqzBMXFbkxyw6mGsxpR5thLaei1olWkjwmjn9GmfIcmzez9R8J8IkbPoGKMI/VOGvq4YfTjVth6hbFhRkFo9LETMMeYPa2M0XBdU0uht4m3vXDPO9ooRB/SjhgtwQ6i61hTMcHgQiNGCDFwjo3SvrHmRu+VKWycp4RzhnvxNE5oWSi50YbB24jS+lCSWoszglaRAZTeWJTiEk9MYWC0EMJAyTteWywOGR/5/PNP3F9+ItpHznPATxYXLM5HcgXhUIFaq1lmT2v50A0rz61YtnzgWweCiKAmz9c9kyXxKS6cpPLLl2+MofDxkN582wprOgKL2itC15wWc+BmRyHEGaX7cSDSlQ/und/7Qlr/Quk3lJ2gLfzgfiBvCaMTYn/i8fKAqHfU/l9Y7IZ2Bjs/k8xgey+YXpndoPU7xmSUDaT8Soi/pw1DLh3VHacQ0JNjOV/49csrL+93StM8nh55rSu3konzxExkS6/ctzeGVtTmqa2xbxvKBjSFvG50NQ64EZ6yJlY5cbeVyc4Mgf/8t1dav5G2L9zvCu8MXlceL5phDb+sN96CxemZ1gpp+wu9F9Y1ks0TX1eNGfA3m4ij8A+/tyxeoatiTQfLXTlNiA/ckqKNC++rsP2meNue2NLh6cAtbC1SeKA3wDRkVNRwnFNl7Z1hMiW9U3PDlELddvYReCEy8s5JWVotiHSut3e2nhkCpSb28k5ub0wLaAql3/DRsO+FD48zqhsQRQiOaBzLMlHf74chVAZEj7UWrxRJDUrviI1sraFQWB2hb4ymaQq6KHpt3/fu7+hgKT3jusEajehKbp2SVrw3x4FAgQ0zKe0oPDGeQCpbvjJZwegTqTbW/cbl5Mkts7WKGCjtYHBITkTvUdKOz7E2HMV3dfTl806rlf/bxwf+lxD4778NfrsbkvM0X2juG84o9l5RTkh5ZdE7lxl2nWmlU7uANQRpnIbmOTp+fH7m8fzI2gr/6dsL//nXd4pUtFUEbVjixGI8qd1JdQctFLljzcxW3qlDU3rmt/dvTFrxtt1p1lGVBuNxxlFyxrpAGkIdGWmZfP9KapVm1fFsGB2tobUCxqKcZauVrVdQEWOOg9KQwTnMfNAVzYEh7tpya0LZNzQaXXfO9oJTmo4mG092jVILXe4EGj3did4RWqXJHYY+gqu10vYdNYFvK8+j8Ejk7iJJC2l9IVuDVTe8F/JIpGqQXHDiGcqx+IAZiafwzLUqtrIhYsFqhjqm56VWZEu45YS2HrLmw/zEzHGR7gi6Vs7OYYwh6r8zZ2DIUQv0jzNhmpHSUX1HqwPHOHSkm6PH7/WFs41IXRFT0P6BNiJ6fmbo+r2zfjie+zDUsWJ1A3d4tl08UYswGHR9oG47iVQyauxQ37HWo1uhJIXWEWMc6y3wui7cekerG3NI2DQo3XC7qgOSsmuUaMQotrEe+QE6pQ+kd3q7I2rH6IFoT6qCKHvsnK3gfEHbwbD9GM3U8B3RG0kps6dOxbDt69G0YAI8vQ1KshQGdbwyOUBl2vdpQsWgnWEKiegrs93xbuc0KYKbaEOR28AbQUuAFlm8Z3RF74opQZGJJAq84nKZWMIgesFagzEaGQ36nd4aD1b4w+UO/Z3aM0b3A/NJAXOMnbyL3NMjX/eZ0g8DoHWglVCqo49M8B605X278Xw+8ftPCW9mnPbk7YoiYtSdujf+9nXi//jnR/bxA/Np4nE64WeL9I37/c6tCGImasoYGte7obTObU0HAXMESqmI9d+12IPS3qn9aFZ8+/JK1Yk1NfJwJKXZe6Mmx54rHYix8RAUCPx4PiE40lZRKmFDJwThadrx9SutZ1BQyo2T8+i84dJK8J4erqSyM9LKYgSsw5hDCSylM1nHt+ufeZ4a1sBQCtEP/PnLiVuxx5RiCM+PihDBBIXYhlghRI+LAXGW7Uvlt/3Gl/uVf/j4yD/+u498HGfe9sptK2jj+eGH3/N2O1L2sbzw+Kgx0hCxqGkhvwghzFhjec83blXzljRbicxz5Pk0cYkKjeO+FV7eC4XMJWp+nByeV356OFG04b983al3YdhAHR2U5b/8+Ru//1Ghyo5Io1Y5tL1iEBOws+WbKL5uE208kJvQSiEwcd+PfbeWgeSCaYPFGIpcqXQKCRk7Kxk3ClYNTGhs9w3FCTEeZRWUjBYYUpjmmTY6fatMzjKbjtUGbyZyqzhleIgTwc/seXArFUFxGQZvZqKzPJhA09B6x4eZhrBcFi7esb6/cU9XlIWPpwfSlricFrayUvrO4/mBVDZEK7QTtvuV02Xmr7+9U+qRBxIa2hy3Oaygg6bXjkhjmMHwkNtG0wNUJOWMc4a1bhRVwAndNIw+goTaW+id8zSx7TdCWDhpTy7qqLlOhtf7F/T8kX//xyf+298a/1pWbmU9BDgGlFPctzvSK4Qj9T9GJniDHgbnLMtQaDXIPfFaMn/99QVtAnWcuDxEWq/Ms8MwGKnghUMfbGayFE54JrFEY9kQxBmqVEZtuMWTc8bFSK2JVgvKGkQNtDukcAZBe0cfK1vNqGniwOtoNEdmLfeGUt+po61D72hj0dIPZocupPsVFSNdFKUWRCsMGu8M23bj03SmtMSqPTc0SVeGbNArYjkU8L3T5VjtxnBBtYRVx2QzGjgzuBiFypUmjWYPLsskDXGBV2vY+vc1s3F01cl15yRH3XmRmTEEqQoXAsMca4LpbKApggm0qo415hgE0VgRtGScJKwoLObILP09DwMlDbpJfHtLTENjqj4odMrTRSjdULqmpUDwjiYKpTwdYS+JYuA+3hl1w9RD0oERcu1oXQlmpZTbkYRvN4oIQzpqbFidUGOnm46WHWsLRnmUisT5cvS724WNjyS7gFZYJxTRDOXJKPRkWebAqKBkIFroeiDm0FY8OAU98zRXvM5oq0mjUpQ90s7GHjt41fHO0ZpmbzOlaobSiHGYU+Xp8Y7hjlUH/tNqjdYWYy1Kz2AWZAheVax1R33NVIYkpigo9Q7yxjSD4BD7TB6e0i+8r1Cqo9WAUxNea/pQlDa4aMfWGnqa6MAyBXrdqLVQCWzdkvbjyxmU8DgPFntFu5XW71hrSdt2RB1ouL6ixxU/PhLEUYocSdoycA7U0OQklC2Ra2Y4w10U/+f7+/e+b2B0Tddntl2TN+Evvyru7YL1YPeEfUtMeqBIdA21C6XsWI4XQm073kbqsFQDi/N8l+oeP0NjeTYzqXmMc/wQOpMTLtPBRo9TpPYC/SuLrSAbj6eAtgY0zG6nq4r0yqDRtQE0MxVpnVwMt2zwcyQNh27toGCORt7f+N/+ZWfNlqfFESycZouzndZX9Gkmt8MNoGRB2id++fWJ//jLv+P/+GapxjIFw2WFcxBSWRnjRlMdYy3j7TfqWNF+otmF24CXzzu/7vDpIXC9vR0QEgzeT+y5kttKGJn7fiBhSxFS36nKc3GJ0wyZwpe3GwXDNVUeJ8UH8czLhZQNp4cPJPXAvguVgr8suOq49iN8h7H8X/7xR2zfeZw5uu72CaU2kJXb9QuC4PyMs0+8rJqsF5bW+bq9EYxm29+ZpjNDKoOK9Z1SNqZZoXpll6+IL5z0RLq+Yb1gpIFuaN1YjPBBWdI4RvJPjw90vzNHj94aSXVKb4honk9nlghjJIZ43vYdO2mij7Ta6MZRjWFvUF8TP9lIzpVv6cYuDRtm3koliQAvnJ3FDXBOY9EoGQzTqf1OajesNuRtZZoP54jplsd5ovfOqO94Z5F+TOZSzoyhGWUw+sZsF5wWbIgoqceOuBZO8YI6fWSYxjlcKH1lVwnRiqYaLhq0NKbZ4zQYcZh+tI6MMdR+gLAShq/7ilWdRz/zVYSsNFUPfNA4YyDLkfNRGu0DjIFSmsk71DigVs0M7l3x9m752maq8myjkVpndh7q4EP0xFNkchEfjpbAroStVNrQ3NKOG4V1f2XUY6JBPwBwrSZGgeW0HDmeUYk2on3Eh8ggoIcjiDCcwTpPrY3z6QE9GvjAXirpf/zbPVP0jFa/V6sHoYGziroXog+kVtFK4VxA60odO0UMWxfeG2TVaaOCdFIZFDOoozDEM2qlVMMwYCyUvjF0xWoOs6A7U7d3+hS4aiG4Qs8dZTRpW7mECxDRVKIy2J4xShGdpXbQXRN7ZHq8INL59vUrY2gUghV9rGWzRo/BKA1HIdqK6plaBPtvaxb+2w8DSY0DhtMd911jcTg7o83AGk3aG5IGi56xQzNGxQZYFocaG8HOxDaYfcUooaoTf3nt9BGwVtDxgdq+oOhHx93vWHfjfL6BrLS6E6wmxEYvCaMVop4o/Zm0f+Rez2xE4iUSEKZF0ySiZMLUceAb1WAKhn2/gTModfTtl5CPB3K6s8SJdY3c6nGQWFTgQXaUvCKm451h1hdaDzQP635nGAiT5TTdie5fCfqNnu9oM5DvqXClLCE8kMqF61tjeE+ug9Yz1hdi3DHqjTG+MVRmqzNa/4m9el42y14G6+65rYotG4J9wIribc3cysanDz/QVOOUM3UrxOh5vb3S4wzqgVo7Xhke/UJv76zrF+LjGy4WqiTsEa0ntYLqG2oojFrYauHn3165l86QRnQe9IEivedE3laM6cQpINdC1PB+K3TAYLE20obhbCL/zz/9kW9JI+agos0RZu8YGO41Eb0mugmjFKo1hsz4OOHCmdv9yod541HteGsw9oltF2quGPPIUIOnZwEqGE2uK6n8jedLpNV/RY1XZGS0cigWUvb0OjFGQcjHWsYt1Gq4KaGvnW954sv+yJevB8v8IXhOZqCvCVGVX9ZA8Q+8Vst27+y/dKxWh1JbN+b4kWWaac1wTYHX/Ylv5ZkVECmsDd5y4sMspPUYHTsPvd2wNLR0cl1BNJqJtXZ+u92p7R0/ChSoynHdC4ghyyG2uRVLUQExJ4ae0ALzSNz3K8oJzjmen35Av74RtWPfVv7rf39n6Ge6Gux10CUgcuL/+5eCtxP7doQMnTXoqaPLlW+3OxVwbuZ627FBo/SJXAsxTBix/PbaKXpjWo5x+6NdmZeKP010bYhbZ5cd7zVzONZjfpoIzvG76YnXv63c0hUfDUYFPA5JG89GuErh1jjMoV54K4OsG/cts49ORVhTOX6t98M3YDNaO75db+ytswPWLOQqpF45XxbQx+75vJxZLk+seSC50aTxXq9EP6H2xqQKy+TounJLO13XYyLiJmYurJtmS4XT7ElpZY6RQUfZGecDojXeBwaa4D0zExcX2W4rZj7Tx8CNymIt1lvupMMtEhfSnpAxmKcZ2QfRCF5rpBrOMR4B5hAOmVKDB+s4W8vaM9ELn54CP39eufWN0vfjRWA8pa4YFMoLRTte10KVjqMxhQAKFjcje2HLjauBPgmJztAasZq9Fz6/3pGSeJofj/CbroRlQSkHTXNxkU8ezPnhaAP0zD3vB4rZGeb5hMYSlOflyzeaCMN6JAnrWg68vTvaVqNzhPDyQEqhu05KlWHVQeQERoFeC9PsQHfMNI6WTgMtipOf6aOjFJgYSHnjNjJPMbC+JZqBoRrbdmc5H5e40TpFN9To9FGwPmJtpvaNbsDYMyoNakoEM8h154uCMM+UnNFD+BhPLG5GuTNzU6BB6gr7O6fN8vD0jPrQKEaotZLzjafTBwqNGM9Id7S1H1htY2l1EIfiUStm23kfHfg7ZwZK19Bg04P91aD1wvT0gFU7Vq2cY6abV5Q6RszGGKZlQfvKSRWMe+QSLozvmNuXm+cte9ATk/OYduXjQ2SJd7T6inGJYHdm9yuiN5Q2ByhCMi7KcRhoGnUP5GrZbxtOa1Tr5FHY9Xej1Xjh7HaCayju5PxCtA2lH2iyEF0lmjtSBW8C632m1t9xTWeaKjw8vPBpuTL6byiTeFge+PY14vyZ4TWny8Lrdmc+a373bPHSeH/7jPWCNpraDwSwMQHjGvtWKe589P1rBlUp48pirwz5guhMHo2cB153cimU9EKqvyHVsegZ5y5YGw7QzxhMIXK/ZZZ5Ru+WqDMnEufHR37bB1sfzGrw41yJ7Q01vfI0f+Ny3jD6yB8o5THKEXfDGIMYLaLAk/jd5YSxj6Aac5wO94LOoByGE8EoaJqRLcbsuD/9QMuNyc44tbCXFWMGYid+uRq8E5y+oUwn14x2hg8KhrTDZZ8bA+ja8bZu2FYY9Z3FbwT5DeNOfPlS+e1bRFTgNB/imS/vNx4+GHq/olVmip17fuP99TeQclQLtaKJ4u1meM8d7SN5T9TeyXInj8jokdcvjjWfuOoT78VQtcFYRaTix5GIuq+Cmz15VPpwdGUwCnrZ0dYQ+jP9PlOa47YZ8I+UoQlB0NbRRsdpxYPt/PCoUEaoMnhZd3LrzNOE1wZtNPf1AEilUgiLYfEe5eG6FWY38XZrjAEDaC2zj6MdY+yOlUbV7+R2hzmQEd62O2A4hRPBGV7vkLuhWQF/cEEYg9obrymTWyRKwPZB+fKOoZOrPZDVRhP0D+Q1seYbZv5A3RW+gbdPlGG4vq38tAiBL4gUtvs7BcueM1kJVnX2DdZ0Z9eCKMO3+Ree3IwOjtftilVHl3rxDmcCchdmM1HbirTBljfe6kYajXvZGbXSpNBHQhi4qmnSDytiO/gayhhsOL5nagxSj5TtN1qw1JRIFNatUPKR1vdBk9pK753LOWC9pdWNKo0QPUkS9/3OeV5IuVIErvtObyt+UvSesNrRUsVo8AWihbFvlHrnc1MMUQc7oFViOPgVpRmu2ytxckRn+TCd2ZvH9Y4zjtETDk1Csw9BjKIMxciVk4k8Os/Dw5n99k7eCyNnTgzqNPG67YeBUgtLdNjSmCbN9b4dnwkGPjqaMuAtL/vgAhiueIkkCShzZCC8MbQ9Q6vUnNF6xShN7nd029hrx9qAUYqgFFo10IqOIo+OKEWInnXP7OXOBDwvM+fwQHkXfjc/kP2J13oj9TsuWqb4QNGGt7SjXEfLO0Yn9l4pVnNLBawwucjkNKMmjDkU0892pq+N7o+QeefgtDQpXHvmSx4Ubb63qgrKCGN0Sjs4B1ofuuXoLZ+eH1C3N6I0tBoY3VmiwfWGtpqadvLwbF3RcmZWE4uNfFAz6z2yidBjRKfIfFd8MIHr5xc+/fsTb96zKpj9jDWR39ZXahtYNJO1DKsow6DcTFSNByMsrhKnGdn/ba/5f/thYAhOHDE45sXyOGf+cNmZYwL9jjFvONvoBO5J6ENhwwUdF7x95vZ54INguPBta3x+EYZ7wPmI8ccI37s3pvBOlxe02zD2jraFQaZz1AOtBWMUioo1HMGLkZlCx+j9QETahcpE6q/88XIj8heQb1iTae3Yjxl/QssPGDVI6UrD0OWR+OM/UeoEnzPn05VPp9/w+gt7+szDxdL2V5bwI9gBTrE8GNrrBt/3i44V68xBR9QOr2cUFzAT6Ae8fcS4BeMq86JwqtK6x1hNboo4P+FRRwhzaJ7cxu/OMLTACDhtua8ruc7c7prz/IQ2coyjbGOYmevuOceD9PePxbLWjcvUcPJngn3FxxXFO5fFYnqn7hVlA2MMnHSm0wVNoKYFjOeHxxNVzTQZGHOsSsp+A1UI55mtNn799ZWymWOFoDaMglKFLoPWKqeo+ddf/8I//5yYzxoXGtITalR00GxieNsNuRlKV3QG1ghzhBgqz36l7Rk3Gj9/hf/+68S3/MzQhskfNq9RLdOieHp2/NPvOg/TlZa+8p40e5mxxqNU4GVbuPUnNnew4Kkz2ityS6AHNQ8CJ4Z2lCEk6fQxaK2za4XDQR8Yq2lbAWWx0TPKIdVBFFvRvBUhD4efJ8wUMMbitDAvmtQbQwaOzhQtdYdaKlkGa26s+86pH9yFrjQv653qIrFnFCeWKSJNyFvhD7974np648/fEk57tI9gZ0Q7vPPUfCOKIreImiL3kgjTRMmVTyfPnx5/YN9Xyvea1BwdCriuryxz4LoZdtkYIjxNM/fbG5iJKT4cSWsGlYp1wuIC+2jUBJ9OPyBDkXojj8CWjmS/YietmeImci0475nsYF13rLO0vLOOxlZ2fh2WYCe21JC28XyOWKO5Js9tz7TyBWsyZSTKSKxjA68p31P/aWS27Y148qQKVk8HCloNhoboDZ1M6StaKb5c/4ppO7lAcQZdrzjjENNJ+c7j6SP3nBEtGDfTR6X0gomBbgaJgvOGvXeMV9RtBY7MlVYRrQ19DIxtx8qgrZjakQI+nnnNR5bEacNyPqH6IOWdeytUlei5Hf4G70ivV4yODC001WliuFt11N/6gaRVw9H1ATX7XO840cRgGKMwuUFzDvP8SC7tWDtIwdCxaHpVtC4oJ4gaR2h1CNI7k9VMWvjJDj7njdzkQPvWw3Vow1H9bHrQxlGT1F6R9v3wh4zKrjXBB0QMqQ3K6LgYud5ej1plSkivvLrIWd74GD7wehs82cBDXzipGZrGZIceHb87shpIU/zpD3/gYZ5Yt53/+svf2FvHDoPKmtQmlHjm8wPn6YmHW+P1+sowgrLH5BMvGC3UsjP7E0Y5qIqOxwi4YCm6s7fCZD1OOqVcmUxitErX5gg31p2eFESNU4r7bUViZAkXdGvMp85paZxOjnQ98W2fmGoh7h2RK1//939m0r/nD/+v/zd/3gyFO6PtiCyMmigyOM8Xesrk3tHOMhtBgMUMTlYR/95ugm4LmYa4ysfnjU/Lv3BW/4Kor8CG0RupFoY+aHJ9nEhF00pkGhP7t4BZBO0977eFsVvm6ZEtZUZvPEbLKAm1CNZ2tLmhzQ7mCMgpNTAaENAiDBlIrpDXYwfkK0qOL4H2HSOOUXceT9/o5c/09s7QAxVBgM5K9InWNF3uaP+dDKZmtr3x6aTw/q+M8RXxAxNuiJ0Rd8YG6GbgfATdsU5RROO9pq0DqxcGFXTAmDOjW1wIOGWYfcdyp5s7znVK+4a4dDwg9GE961qYtYXyhuEKDBBHmH4il0w4n8lbxObIT4+dWq/UfsNPZ9KoKGWYrUXqC84ZZvvOFG7U/Feib+TyG0o1/vyLYlSLEce310IqntsWaMNi1Yn7duGXd3jpd9p4Jzo5shAWTKvHQ98Wkhq834XJLkx2ofSDSPnt3kjDYk3lhykzXq7U4Xmr/gifGcspwMiNr2tjNdMR9HIaZdUhFHGF6gf3XlAvmvH6gb/9+sRN/57uHmlK4wvIWFAmck2N7fUL/+t/mHicrvxyHeQ2c69AdYi+8Ms9sLmZvWnKngjxwvX6RjqmdHywAa+Ec1DcKgwF2hqaMseDjYZWjTk6RjLk0rBimexCbzvGe96+o6B11AyOnIVTlSGVnDqVRqPQJPPnrxVrDUMUFcEuj5ymR7Rx9N5JNWHCcshNvsOhfnlJaOPZxFE+/8JD9Hy6PLPtO6IGj4+e93ui541oCs+L5pdvNxiNJWhKfUMNCP6JrTX2rGmjIm2jJ0/QUMtXUAtVKd7zC6ktPJ4fKPJKGQkzrmjTCc4fFahRmYNHUqO0SrlHlNJ0LPeicK2Sz429rtSh2PcVLAefRGCUQVFCdBGhUGol1YqxETsFatlY+47Rgc9bobZ27H11orBSVcGfHXUUnBlMEdT/j7V/3bUkW9M0oecbRzObh7WWu8dxH6uziuyE7spuEN3QCIEQSEjdEhJ3gYS4MG4AhBpRNI1QoyqqsivJncedOyJ3RHiEu6/DnNPMxnnwY6wsbmD/c7nkcrmvOc3G+L73fZ6eUEwoDNNhBHH3pDBi0NaR440uFSWV6TgTrhteNZQ39D5sq1b64JB0eLz+NGrGrbMlhTeGWCvr5Qa6UlpBzxNrudHw9DJCcMvs8LPD9dEAQjr9tSpWXyeIJWxkhESB7kenPCd6eq1oO0EZy3w8EkLkfLzjYZqxfuJDvLG3xOIdVCGkK7O3eDtTE6PdNHuq6my1s28Xegrcnj8SrEJkRhtLC6CNZ3ZnDpPBpQ1vJ7QMIZA3I0cQWsP2xtcPM+d25G8+vlB756AEf3/m46cP1A6lV5QBGDK5WtMI/5aMmQ+vyG9FyYmqGnVLKN3IJYy8goJiDde0M8uFiQ63SE6ZXXeYJq6Xx0FYVULRiZwa17//xHnWnBfN24cHttxJpaK8J/RCoPMYGn39wNF43h0mjLOczgdKCuRqSGvHMKG0p8gZq4RbG4031QXbE2d75O1y4J33Q/0+32F6pXTDT4+FNc14CjSwoon9mVZhPh7xZqCMd124sxfu3j9z3O/xzx8pt4/4r77Gms95fjR8lc8c1JFSD+S2oZtgcsVW6DUxa4tNhmtPHKY35HbE9MLblrEl/mEPA75q8rUzP7zwuf9bjvZ3qPqRUm4UEqI6vWdIF0w11KppcWXbB9kqXyDtO0V39uKZzJkYE5bONGVmdeVu2ZlVQVsZTmY7Y1TAdYvQoI/tRwM0HmM9NFAt09UGpjKJB6nEujIvwqwT0RQqQhuFEkR3jFEosWg5Ms1jVzl7izcRf/8BKVdKf4/4jDKOrmVUIntnXhZCm0AMtQbuDjK+HKohkyPXgjEHYEGYqKVhbMWrcep2klG6UmpH4chNmCZPaoLSFeM08RrReFrRZISYC90IqDNb6rysn1j3O8zlxMutEoIi1kEj21CYngfMSSe0CZT0hJaOSMG5hbI/8+33V9AnvvjigVA8L9cjob3j01bp+4nazlQ7Ya1w1ht38/A8WBHipvkQDJ+U8Jwiq6q8mRYW+4ZwyVxS5bmB8kciEaM2ZhtxCFo3lOmsJvPdGin5VUsqgU6naT2CRCpRa3jt9Ebefx84la/pzSOmYVWh1kKtjYNZ2PNQyoaesDTK9hMpr6RyQGtDr4qQMoudsKKwrRMt9LoymcSaA+D4ylVO5YYuE7bO7DBeyFh6A6MLWnZSWclV0/Ds14irM046qe90rcitw2RRTqNrRkyltoj2moPVmFwG9U40omaccuP71BIxF9bbjY6MA3EfQCmh8tPlhpGOMgaxnp8ugfKyYZSGGLCz4R/Wj9TeMAIHIn3fqfuOtIQ7HIi1oczEh1vjL97/CO3E5w+f0XXiw/pEIhDaSm6AmvCTI6Wdx1vES6f3jefrBe8VpbjR8gBM8ZRSOfu3lJcrzQyipC5Cs1dWNE/ZsHZHMRrlDddceQqZWCyXEMY31VVyvQyeRk9YfySW9MpdEFpzdAN2sSgsREMtEV0bXhytd75wMzk6qhvKbyUaZo2qna4N1k4EOkaBcup1ZXVktno8f3Km1Ia1jtkvY5WUE4v3dCpvTidcN/TqOMqJWAsxbizG4YzhFncOi8ZYxXV7ZBdD7Q1piml2XC9PGBGKnsitjnqiUhjnUcaQWmBxirAOFov1J65Vc4uJXCqmdlIKlJbYb0/oyZLzjrMTqhVaVWzlSowV7Y50Knsp3NadXuFhmamq4bThizdf0gW+/1Q4n2dC7cyHiSUlnLMs85mCGbp44+lKYxbHh8dPvDkpfr4YPl1uuGzIa+D2fAOtmK1mEk3LmW4VfrIUGmIdojVWhgDM29FA2OKKNRYETC0o0QNN3DJrKLydFfEiiPLkSbGqwNWNkX3rjVQK7uhouXOjEfZIDZVQKqVXWMfnVCZHaZ3eEtd44d56VNFc8guT1khrKDTKWpCJUiZa07h5ouYr2goHcfRwG7wStXBNM4+7Y4+ZUg371tmK0HHYGLg7OMI0U5LGH99R/1HYFhXH5YH7Xzu++Ve/IX77L3DXna38B4gYnt9/5G/+5s+4u/sjFj3iIM6+RdWZQ69MamepK5/fBGqgdCGfv+ZQFkzcma3/wx4GVFfQGp6CU++xfEPvV3qviG006VgNxgjeVnwZ9b/7yeDkG+RNw5mGUh2lPaJO7MnQdGOZNw7zPloIJdHIiNYgduhVSx/4yN7RWrOHPlKfraG9Z9tHaKyrjNIawWCo1PIjqfxEqnmcRntD6GgjOOtw1RFbx6DwfkG0x+Tra183cis3pulEqw5nZ1rX+OWIbpaJEzFknCk0GnM1TLoSa6V34RoHCtIaTc0N3YSXAj3NxGJIccNox5oGv9valdPZM02R9Jh5fhrWwCoGJQ4h0Z5fCGRKLpim0Crz+w+d3/7kKFlzmMFY4YdLwonn6Au4jcMU0H0wr5WyxHbj7D1ffqGp6p5SzcAp+8weMxqDP/iRsJdnTl64czdme6OmhFOOmiu6PZBkItWCE0HVYTtcP+yow4FJFLkUjB9K2iQKLYmSd7aYSJNmLZU9Bo62YkURe2RHUWvFqGESozT2XCnakGvhzma82gh5w+gJbzz6lR4WamXPlffvP/LrLxXGnqlbRxuD0h6tD5RqcDRaiPSyU3RA9w1TKl52fjVP2HQj5AvOOd7YmdA60+ktn1agN2Z7oorw06cb59PM48tKWytGDLMx9Hyjtcybuwe0Tog0IpEmgradSWtas7QsZBVRplH72C22XNniNoAkIiPprDqWjvRIJBMTLEXxdjrxkiI3B1UVzpPjsmeqEXoHq4WtQdcW0y2qjFvstTtqh7yv5HxkNon28nuc6zy/POKm1xt7TUPT2iq9Z+qWKT1jJ8ukM70Ucg+0Pv4OUR5BmI5w2Z/QZubgNVtcCemFb3+88OlyJVhPn2dcM3z6uFN6R1kFMjEZxWIzbyeN042qNaEWrPYYZ+nVsMwLsXQUlV4DulXeHM60NfFmPnOwE74rLnIjGai+kGom7qOT32rHqo7oMfattXGydyQJKKMHW18USlu0cThtMQ662qipcuencaigYa2hKuH2uKK1oeZhVVzLRqsFi8YswlZuYA0lFGp1JN8puaHJmMmQ68iMxLRxb4+INOiCaZmzX9hLJnVNK4Y1DqV7KRVNofdEimDnA4XMXhMlFwSYtWc2o4GT08Y0e1IYI/FZK+gDTtSAyTh0hVgCyoBxltwrt7gxTQcmp/HOQKwU61jbhgo36lYhb9j7N3Qz0eYZrzRej0B3rI19jyMk2irKGbyeMVgOVtGVUKmkx8Rtv2F043Q6jstfZcCbvFBMJANOOZxf+Cl8Aq+pFGLYxrSnbhhViTWipFGksuZIVcMMa4zBmQMxZoz3pNbZJeHNwncvN1pu3M2OxSp0F9Y10SsU0eyyktsN20a7yVGgOmLU/BQNL1vBTUdSWrnzislWup6x2tFU4e70QA0ZsW/Y1IypljVrbh+PLPe/5PP/wZ8yieXDv/kvWcMz17Rj8sLL33zg+fY7Ht4ISRT/nf/0f4H98le8kyOnvLH+5l/wRq781b/+N3h94t//z/9XHG+VsEfS9gdeExyMQ2GZVeOz45XTcoEWXit3A/phtAalCLkM17iMZOrcn0jpmd4SojWtVaybqNVTe8FMHW3vEHlHapVpcuRyRUtFS6WUNkZqIlArRilaK6iW6OnGIp1MB2WY/YnWZrwMwYxFU/WC8zKCRpJBgzcy7HF7pPbMZM9oFLU801qmtozXBilgrSXHCg2MM4NKlwB1z/U5k+uOmxZyCtxCoLbKUxwP0YMP6KKoxfJ4bbT2hkuEqh2TVUxT4+CFbu0YZbVAT2G4tKUQQmai4VUGdWOaG6fP74j7MyU/8eXXB95M99A8Nf1AKe/5p2++oJWKM41aV5zPHM73/PThE61nql5xqtFMHmraGqEaJna2mJnSz2gEusnMeox36c+k8EQvkSaeXk/UMhGrJqeOV5WSXvjwFGj9SI6C8g6V+2BE6CuFFxoRkUK3Qq8ZrQydfwwPNkqLxApNN3RpHK2hlZG/ULYhtpPDDWsMrY4RsZIF6RNNhnQmpcrHpyufvdk5nt9yTorShn/gORWydN6+uUf0yixC7grbJq7XlS9s4+f3EzlENjof1yveCe9O96CF26XydNvYbB9ApK6oAiyZ5gy5Rk53Z1oW7rrmP/7lGVs3frzeKGok6LuCu+Ud+574+PgC1eKtJ9XKh+szDo1UhxZPoBF7ptHoFDoRPR/Y95UvZ8s7hA8iBO+x3hKvG3aZ0dIokgk1jIBjKNy5MzVudK1YmyOkTpdK00MBi3Se9pXsElng4C3KFmKMVDLH2XHQQi2BSuKwzP9uYjHPJ2qzhFsgbJGnRfghPIIIR84E09m3T4huoztvGqVVTs6xTAuxgThBG4OumYe7E669oFUl0tlfIuseePduJhQB6UxS6PGGd4ZiFRRhWTz7LWBnaF24lMLHNSIOjkfPQFt2aoqUugKK1oWUMtSCmyxrCnQNpWas1YS44o8nrHIDbkZimiylJW6pcYmV0hu5ZnqrNDvoo5EdO1n2tOJnS3k97MkidCuopHEYSsmcTifqXgmxIAK5JKzROGV5s5wwtaKV8DElqnvt8bdIaINZ4LzFak1LjVIiVjQ1F0QpSk1c4xO5D9GPSKFWeF4DWjW80fzw09/jrKbsL2y5cZgUtRSM9ejljm1N0DRWRZaeSExcsuExafaws9DRi+UmmbUL+s7j3UwqeQB6/JEaYNYTxlpiy9wuiWYN++0FNynEQOlxCIlyZimRViu9DMEfItzSjdiERU0IaYTKU0Az8MWY15+lrignlNYJLVFMpBtGPdDC7EB3Tc47d8sBVzKxBa6t0BR0xuUjpcDlemGZI9o7tnxFTCWVAYc7iaPUymIU67VScUivvD3MHAkgM/9w66xKqGbGTZ7P54VDuCOlgg/P5NsP5L7gzQP69DVP9k+4+N/yx3/0z1G/+4bDL458++kjoQfibxOHL3/Bh99esO1vOf/x/4j27pd8+2z4/bf/inNsJBV4ur7wm7/8K764E56v0x/2MGC1o6sDNXYOXuNNQEuitwhGqLljRQCFVmO0mVrAaYfriV4SsQUQDSrT2JF//HI1T68GaseaBasMzp2oXKk5MFmgvVrzZHRnaXbs7Ftnrp3UhYLBVKi5ITnhiBg9ZDJKxhckl8bkFFI8MVqoFiU7uVRqT0NC0iKdijJmqJu3K/sOnQlXn+gq8xRvGKvZYqaUyN0Mve1QN46TQymIVTDG0Ol0rXjHidPhc1KNbP2R45RQ8gErj7SSON4dCNdKKIH5rjG7wXOwJiPdoN2Za7hR1Hfcv7EYDbfL35PlmWm5o8kzdoHar3i9cPAnPjy94OeZw+EN531HTZDrQo4/ou0N0595sxzoPeCnmW19x5//Fj59/AJxGb0EjN5B3ahto5WCM2e2zdIlo9IN3TUlrWwN1gCYiLN36H8ESdmA0xuprDTbKSngMNje8FIISqi9QS+gBns+t4Hh9dVgGF4JrYTOhTIVnuJO0zNNK5LamcwJUmeLjdorP61gf7hSyNR2h9ITy2y4e3vk+vhM7M+4UyHsV/Y9Y49n6suFN/MDyExsK1U0p9OZp+tOfLzycJ55er4g84Ji5Xx2rLVyKy982isZjzee58snao58oT3p5SNdNm7PK7JMbDHzHDPf6QvXW+Hx9sKsZyaj8YuQS0DXGa0MGOE0nThy4JZWkiRCbay1U7zDO49VcDfNfMoJjKD0EP5YZ9m2Ha0rynZqF2RZ2C4ZpTRbSGAMp4Pi49Mz0h1NKRIXlKpMViMSxsPNOlRtKEn4ydPlyKfLlct6Q5RmDTtyy+xPEf1a+dpaYNWFWjKtZCbvUS3DaUEMaKVpXSMN5tmT1itSLZCpaaXGV9KfqbyEwMenDe0ce2yEGFhspZQVJFPQbLWx7gWnDEcRDsagrRD2TFdqfA+VJfWCbZ2mhNse6E0Gvc8M3K5UTSwFeqVSKblhUQMoVF7oXThOC0YBSvH4eCWLJfcMZvgF5tORmDYOi6dKR7wjqYxMA4rUUKRU0V3oMoLIt5zIqFHJLYquhg1UJsvXD+9IL59Ik+Mn3XhphWmZUKVTC+QozPaEV4ZJT2gsoaSxLvOGeN2GaCglovNUZ/DEEehrga0kjm7iWiN5Hiu2WDpGFGVP4zLWBdU2Pps6D6pzc1/xD9kQusJJZjbQSmFSiud1Q2lhjSsl7szmgUl5eq1cr4O/Yp0l7ZHSMjKNxgyqUXsjUDEizNYwzx6pgzljjKWWRnad5DKX9QNVg/RBxC0lo91EYmPNEStqWF3LRteNrhVd61fp3IVWoKaBWTdaWEuh9AxaD6ETndAbapmoplH6jplHfU+pQTrUZsZgeHee+f1PBUGzoHjjDO/sRO4QtWczM0ovNFFMxvF1bUzbzuPf/Tnpt/8WfTzw9PieD+/u+Q//+D/hwf9n7PsN8R4TDJ/f/RNC+opr23mpO7ff/iv8+i0hXfhP/7f/e778L/53/H//j4lJ/8jp66+xf/JH3E2N9a9/wNvTH/YwgBofjlpGPUurjtThidddcFrGKkEs2ljELLjuqKWj9ISzHW0dtW6IWAQ9KiXK0dQd282wzH5ga/sAA6kOKEdt0LoaCNomGD2hsKQQ0N2jNdQUUHJHjorrSybtnfNy4HCKxBLoJVNToaFQ7sB6Wch5gppQIjgC2mwoM3TGSjWMHV/KjsXbDl3RawQMizuwHCYeTgslA21ltpquTohK3HlNLpbcj2QmSrRIsWzXR4wrvD1vWPU90p+wppHUTk47Jd0wTkgdchBU8YQnjbVHOo3DeeG2/wMNmKeFjzfP3/2YCHyCstN1R1k4msivDpauPR+vG2Kf+BQhXzO//NnC+Th0x7QM7RMQERG++qyzTG/5zV9E5lPneEo8vDkQq+FydeQgXG4TH6Mh2oZXcN1XnB5NjwsrqPF56NqytQ1HpaUbpUYOy5GWCzVsHGaDMZ2nW6cbQ9dQx5mcLlB6IbTMrDq9FVIMVNWI6LHHq6M1YtWRrQa0SnC4I5fOiuHKA3s/EuRIqBr9UjndEtetseYLs9PsqXKpnedPT6yx8emW+Ne/i3x6ubJLR45HEOFBKX7/3e/4h6ApYee0GO6s4ng60PYL17DTnKEoXtHUhi+s5u2dplWNuQjKVk6uc//mDd9+f4GWOS6abj2nu3t6CRyWmdN8z9O+gTXkPbGvV+L+SLOGUCKfEErOxDlBE3S40HJg2xrOjHWaqZotPHNYJvb1gryqtFuDWhTi6mDvu8JWLlRzQNsJPWkMO0JExNDVxPVyI2UFVE5mJlP59umFvXdKK7jZ0i4XXFa8vVs4LjPXkhHVOJ5mFjdWTt4aQtk4Pbwh7pm8Z7bbjjITB3+k9IyoynJwuHnheg2kWLiuEXuYWWNGhY2DtaB2quxIE1Ie8LLSC7MTSilgHZ8uK+BAN5SxxBKBSu8d59yASGmF8/MAGlVL74LVlnlayDVQpQ2/h2poJQiagzOQAiwTmcZtvyHuVVJjFOu6o6QT9kRHIcowTwsiowYYU+B4PIGUgdVtndocXRuEOC5D5kjrjee1ki1M1qGtMAXNS+6IB0pnmmZaq/RqaFFoumKdYbEeasM7RzGBpto4YIrCSuHhNA5A6y1SdKZ6Q+8NLZqwBuxhwhiP7YbrWjjNR+4aPLTIg2SoK+l6RTGcAOsemKzh+Wmn18ZhOVAMPNZIK4Gex4s94UfzqlWcHtU80cK+31BeUNpilaDijqPiW0ErRdeKRCNmsKeFy7ayUXF2JoQNK2P9G9OVpiqlJZxzSCs4J4Q0OP9KNI0RJvTWopUml0g2jtI7VcbU6O1yYDKa23XUEKdJqL2TEZQ4vDO4yUNT6G64c4rPjOGpWyajKWXDHe5wXXM/TaxB0IcFrR36BO7jM9//1/9XyvoD0lbCZYdv7zjImbs/Xvj9vvH3f/vnFAdPLwF7/xVH9xWf/4e/5seP31JXw6dL4uXPninHf8FXv/oT/r0/+U/42a/uiD97S1wF/xeGl7CQzB84M6BtGQ91DZNpOKPpmZEUlQZt3NalOYy9o4SJWgx6nulUmmRa2um1oowMuqDSKKXJfWFLdwQekN7pah9c6gLTfM++J0Ls3PaN42Q4H0fFLl4yt+szxkz45UQpBy5PmlIPODUjpbFeO+5UoAboDlGKy9Wz7nZ8iXpi8urfyXCMqljXESy93RG3z0nZgSqUPF50Sk8oObBeA+ezIe4vtB6JGyzuQE+d3qGy8BIOXF4sIgsdw8Pbtxizk8PHoR5FY+1MLYrSAyVrRCZyfUNY7wjJ8bJNdLEoHfk1gcMy4V2nVUcOdxzNGx5OhpQ+IEfLlhOzcxykc2qd4jSqRM7dgH7g9vE7PvtS6CqS65VWIh2DiObp8iNOnfnqQSiiuG03XuLENXpCPVCa4/l55n3QhNBJPWG0IHUcou5OBqSz+HFTak0o4UZuO2INuXW08ZiWMRRSrji7sNdK7ZmuMk0qoQaUEUraqSZidcOqkSNwRtNLxCohhUzuHT0pYnhGjEJPBQS8ObBuavTZK6QgPNVKEYeqESsZsZoriu/3lSidb2PgGi2RI0V1ShSMHvCSLSSKXbi0wqdbwtyuPFwOUKCUmVvNqLZTu9Bz503IfPtNHMCYVDCzYQuVWq48fRgKVyOKzQkf0zYQqTFwOFwRpZAgxHSj5J2YN2rXw5ZYC02ESGRtjaCuCAH3anOcDp7Sd5RNrGEltxVrOltNUCzGPqBFQS+EvNHNuKHmFnF9B9lQOjAUJJWyK9Y9I0ZxkUBqQ+FtvEdZzbRMNLVzEMV0cFzCjeIsbR32uFoTMe/YyeDcQgiBvAVqjBRZ8dMRbQ7UFHk4OHLM/Pjhmb0mctkQhFYyIlAo7DnRvZClAY2DOyHpwuQEYwHVeN4vNCUYozC5QE1A4zR7ShoinGWydD3CbSINeZW79Vw5PNxxi23gZ0un7BVjNHdmYlkzxjZ0h0lZ9KRY+xhZn5YDujVO08KdOtKbpcQDKim6Cjga+thJZQfrKLUDQhePEdj3TFXC1naSNHRMvCwT7wCrFLYnLBaPoWo7tLU9jEmPEpSq5FxGJmIWLvvKGle6EarxiNK43vFOkdkRKq139hzQHVqJCJnWFddUsHqiW401HbcnLJnD5AiqY1Xg7O7wraFSGxmHljlPB2oLlDKmRHm/sDBhl4VbGZmeh2XB1swtXZEiiNHk2hDROBl5qJMTVN7oDdx8Zl8zHUNXldjjwBLXgPOa1hOxF/awwyt+nVSZjCb3hjhPVhBqQyuNbiOHFkKiTgtZFKUWbIPjdMQpTbhdsL1jrSLut9Hu6QqxE6ItjeFKwVl0LTwshtuuUGZCjHCJBWdhnjQLE1Vr7HFGJs83v/tzPrsz/PAyY+zpFd29sOudf/XX/xeQv6f9PLMVT1obfNi4pk8oeeLp+QnpsNy/g7zx/q/+jI//j/83y5s7Pv/1/5K3DbbtStky7Wd3o5n2hzwMSDFopZgnj+511ProtAZ0R6+fUeqX1OJJF0NpnlIM+dJJbef+ZAlbotFeWdaK1i2xn3i8HtnL53R9pEug6ZXWnjB64zSPTmuvCucc2ipitezpREhfcQkzpR94/uS5RE+JE64b7s9H9DUTws6/90+PUBTSxu5Z+sR8fuD2cqP3iRISuc8YNSEEtAjWfMUPvz9xC1/RtOX8xlPKE9IrSh0odSKEwv1dpbWFrhZCgFpnSkh4P1PaiZ8eJxInurGIEsz1inMrjoyIYTYTk+2jh6s7SlmcPtO3r2jxc5R4jjNoWzmdKpN9xFRH33dyHS/bO1+gB8Qm9pqozTE3y/Q6Xei2E/aVmBXaLaRds58rTY9xPGJoyiFMxGqYfGW5j2xppkRPNyeoBi0GZEG042AqPW4o24gqA4WQVoI09CvX23UPodL3D+wtw3xGF4fKBaMsqVrW0seqxyhAU/qAAlWliWVD2YK4CGXFW4WijxsrCdUVi78jtkBvlm48MfzIbOzQzd4UOTRewgvNz8Q9YfwwMEqtNCfcamVLaXSMBWqPNKWI0qhK6MphRHO5rlTlhgxFyQDZ6M7jtuOUp4kfkigaSWSsQpaFUuHDDf4uK57XyLpXrBh0OWJTwOrIS7gyHY60GIktc9kDa7whGoyCsodXBHYY487CoE9ePmFM5On2QteOW4DQElvX1JbobcjARvAts5ad8/RATCvaHDCi2a4X8nqjao3oilbPKB3ZBNx0R67C8XDH4aCYdEW3KwftWeuNYKGYNrIfBqRlkBHUK0lQNHQuUBvSMnHdiVaT14DHYVSjCuxbRMmVVjuHt1/x4XrjFhLaWRZ/IMWXQWYrlVJAG03MFaFjnaETuTt6cm7oUrFd8F3hjSVLJfSCUYraCiVCbRrvD0xSUA1KK+SS2VOk9z5mU7mRb4WSoKfK0RuW3vHhZXD8/cKn55UQMt5N6BQ4e8tRGWalsbXj7IktKK44tPa4w0LanlCqYkSRSkIrRS6N+/NCTY2orphprFpjG2rlj9vKMhuWrvDW0NLo/3s/jzUlHTcrUAU96dEwkIJ2E1vasSdDbULMmdoa3s/oJig7o9lQCCW30aZyaoB0zFgtzX5GtwVXNFoizSqelWINmvKcieYFPTUmbQaqm4bTw1dT2qgh+95xMkA9MXWydF7WF+5mh1Gdp+1GrQpQdKvIPWNapsdAaxsVIVwbqThSaYRtp3aNzYrW6tD+5kyRipr0WBeIGRctZZGSMUphRLOXnVYK3s+QEkZDzIlUCgnh/NkREU3vmZ4rSje0KBKdNe40bbHKErfC5CYcB+hHUndY11FV2HrGGM+ad5R3tNxQNNTk8UTucobz1zx98Y4av0M//j2T06RzZzs98/Gnn5hsZHrzOfv3Ca6ag50IthKvG5N4cu5YKah2oXz4AbsZPv+f/kdMv/pTEi+Y+2fKMZHLZ6inyx/2MFAZGtySYd1mtDrh3EzpR55fPuf3v/+abz8eyd3ys69OvL2DWHdePl5G2Gja+PzLI81cCNGxZ0fKjstmwTwwLWd+9cUdvT5yvf3EdLTUIrS6o6WgVMVKZ17O3F7u+f73v+Zv/v5zbnWhimfHEuyElQVH42Od2daVJn/EMx9ZTEf3hNYDF1t/1OR4h9UPGNOpZdwOtFY0WfjpxyOfnt9yqzNYENOx7p7J1zFq6hVLx/xQ0foAUinN0NprBcVaNBpnTzSlaT3jNCME5xKTm7Dunjt3xdTv6P0yvsxO0/pMR1DqiNINry94/4gzz9TyTDcbLQlxA50qsI5sA5V17axtwfkTOq8kNJFC1gmceX2xdrI4Jm/oSaipolWl1opSBlrG6B1tJ/aXTGpPtNpQemBxZxG+fHBUJr7bMmtJxBJwsyHlSm8RZxpLqdxtLwgrt2lhM4bjcaKsguodbU64WjiEGzWvo0UibShEe6fQoCd6z4hkpFV662AOKFHQA4aJJg7VMqrLeHu2nVgsazaEFIfCNAlJxs3BvHrZJTae941cI80aUk642bFeL1wQdqNJaM7as10+sd2uMFns/YGUr2ggdnguCpkOOH/HXjpBFE7BrAVlDN98uPLbEOFwYC8J50BNDUfD9YpzwuIq17QRJZJToaoKqtFywWhF6x3lNIkybv/GsfXAKpnsofYyZFQ5U/OQ2qSSmf2EKNhC5XQ48tn9mR8f4eMl8vZwRiOclzJeBG1jUQptPLlbsj5z3QSx9XUqo/FyYnGazw6d328XlHJYPVHLSm6Ny7qizUyK44VXUsCiUVsgs/KxRPYER3PkvByoPRPijhKFNYZtC6QIIm6w4rXFT0eydG7ry2gXJehK8F1QFUoLeDexdMudOrIUzQGP0o4Lic4LrSoSkZorMQakaGZtaGGj18i2b2QakzHYg6ftkXdmIbXMYTK8NY2lZ0pXZDMTzYyplc/vF2bR2JNC1QC64tAs3hDCSpaJGBPOKnptJCqqFLQBRAhhw0weoxQlVbybRl5JN1SD2BvPtfG2NOL1QuoHVP//w+Zb3XEOnBsHiCoVMw20+FYTWEGUDJeMFlqBh/mEFni8XcYEwShUF0Qpei8sy0wrgVY6LYTB/g8Qs2E9TvwUFddVE+h4I8SSqDXTKRjjmOYj7z99oNhKyDeSdlg/YZVBX1ccmnketXAi2DImFl0Yz9re8Fhyi6NNYDy30MBqOooSIyVFplnTSqZYy5YD3QrESmsFrxXWOBqKkATrZ+iwiMZOCtMb2hlKiux55Erm+zPOWfY1kGtBTYp1Wzkoz+F8RufGdd3xGN4ub5GmuTt8wZIWPn1f0H7G4WkdpMM8j/ybmY8seiHahbvF4MLKv/7Xf8HnX/4px5//gunuL2jqkeOvf8Gu37E/nVkWxZqhT57jV0fi00c+7YEH+wVKW5SG/fqJt8rhY0KujTl6Pvz2Az//xc/o/SckfWR+Erje/rCHgdAsa4qsTTC//RVxE5A7nrcjPz4+8Ly/48qM9prfXjf++MuE98J61Sjj+PG6clV5QHzkQOue25bp7p49GNTLjc8PL0j5iCEjeSfFHaOFxmvu0AupGJ5/+owf/+FXbNtb+uRG6MY7Wi1M0x2GSqbSzHHUg1Sk9gPSNa0XirJsyRLThFEz8ZLIUaHdDNbz6SVDeYfMZ7QqdJVQuqCdRXlPFyHkla4toTacFqx06IXeG6Aw2qMkYLQhF4NVmYO98DBfOEwvWJPo/QVdP1DS7+kq0BFK0VAhhJ+oVaF45uB+wKnvcTqB18Q84c0viNXQTaJ0CDmzVSj6SAwTpmpMWXFV4V2nS0AbheovOL1CeEbZgEphaJM7WH2H5AMSHJNauCXDgzuRykbRBTtratmpW0TPDrFHNuUJ8cStWuxkifuNrhXSKne6ce8StIq3nWlShJyplde2hEFoo2udI9bpcdCSjm4dC0Nd3f8xM1LQdNZ1xegjoiy1bcTRLEWRyLkQc+LHPHHZDGoyXEsi7BfcNLGmjRoaYgUpGSUdekcZYbIKrw2YEUBFGUJPTLby8A7kpNlp7PICruCdIVdFMQrlDLk31hop2iGtMTlHrZ7L9kiikvYXRAsp7ogU/P0BbRTGNLpkau7kvdHIr1MwzVquGBFULRg1HA5aBGtnLmvATopdC7c9U7TGGsGIZpi+G9RCzYXZL2jxiJ6w1tLJyDThjWLCEMJGjgrsHaUWYlF8+90T1h+xHoLSPGfhYb7nsjfO7sCvppkPayYqjWXGGcctrHTdiTUzK3nNGglvzicyhlwiLmmc9tjjkd52pmRwWvPmeGaeZvamkK6ZJ8vJOWq8co0bJ5+4P70h5UYr+2A11AEEOvgFiZo77rFKcVIzerL89PKREme60mhnqRKZF8PBT3x9f8+sFdLhw8sjuVfiFggdfvbwBX/05p6TN/R6oWzv8Rr2PLHrA++vgelgmb1nEsFJYb9c6G3ssXWr7GZDiWU9dC49sG03So4oCoqIPxhEN1K6UbKjF8tBHTkexi79o0SkKYw9Mi0LNTyjusbo8VlwxuKKItSMtoZeCts+ap7izLA59sS2Fvw0Y70Z4DVVyRFut42qCl45lBa0tRhlUCLkXMgxIQQMhh4Xbjukk+fvf78xLwf6UsfkTurA9HbFrWTex5+4cMMohahGBvqkEYkcToXUBt5YjIHksW1imSwxB2ofgB5BiEVTuyZWIdahmg4lj7WFVviTw1dLyonz7LBKMZ8mbuuNyVvEem5bJ3Fk3cf3vFVhOs9IL4S40noHLXQ6ugqSoMY0aolGU3ojlgp7hKo5tZm77Dhr4bTcobug9o/ooLB64rPDV9Ard8YwGQPeEg+eFD1N3mJp3C8njv7Iy/4bzscX8uE7ZIkcz2f6dws8D6JpNhe0s9yef2DqiTd6hrKS1oROmrmOVYU7aJzqvP/Lv8Q/3PHV51/yu3/7I7//87/mn/0Hnr/86fs/7GHgh8vEnhw/Onj+8d/n5/wMaZpQjrQ2g1HMbmg/I5qn60fmPbMXT0wTQZ/5/XcruilahYxl7yeKnukifGk7+y3hlSExSIIdO9YKCkqHJp56PXJ7fEeIE8ejJ6oxmupGsAq8g7ObCSmgZoX0iFMdax0l79SaMdZR1Az2TMaS3UxBYaY71j2j5xOqe1TPKF1ofcOYgrMTui9UM7PnSs4GqDSJTB6kbdS6I11hiyPGRDAb1jnuzBMP5ntcfUJKQ9uZff/EPA+9ZO8D+YnqWB149+Z7ZvMRpz7S20cKkaygoCntDduqKM3TbSXmG9fYuQZD8MIWO5SGrh1LYdEGozqzuWHYaOVK7Td073hn0dIpymPkgfzhSAuKMgu9Ze4MNBLZtHHSbwqvFUoaOV1ZNNiumYxn2wL3bsFPltY7tkboDlKlt4Z6vdk3FFsqdFaeHh9RGl7WK4rE/dszcb+w15WiIyXv0Pch+pCKt8KeG6qMl7V1R67bhlWdgz9g1aBD7pJpRoNUkus0MsfjTNtAvEMUtFhZrFCK4dogu07Ume4reCH3SO+g3KggVdcopQ8iIY1eElbPrDHRJDBNB1ocIKTZL1Q6L9tKrRe0E1ob1TPnQNtOzCu1AmqhO0e2hVYbUhvT4hAMuwQkJt4cJopUHm9XtFleA3SKx+uYFGSl6WjO84GaArMTWi7QFN54nNFYMYRboURHUZWPt0/Mpo/kfM/02incQ2vUeOHu/oCIxntNVBMpe/R8prXGh5dPfD4v/OrNAy9h5zQputr43VNlV68VL1VouuInR/aK1hz145XleMR7O1ZurSMapGdaXqnOMB88VMUWVtbtSifRBXpV9NywShFLp+tCjgVjJmIyhC3wtD4j0vHK4aPn2/UFazWkyCSW2DW1CKU6trBTamLxw3HRXUOfZvY18f1jxEvm4ODpmqjcsd92YhUCOyFvr9VgTX5ZmZ1GM662hjaCmlkhiybTsE5DEk4PZ1LaaL1hvH2tAEbuF8dPHxPenWkxotxYTzgzkUrFTBNpF1SOzK2yr4ncJlTraBG2fcOYhqgyKnBxw4jBGNipr/+/EWnCy/aBXl/7+LajVQbplFyY5gNPz7eBf85wnC1smYN33ILm41Mlm2XAx4yi9YL0CkooJWGMoulIN4FsFKU1mDzVK2K6opeG6wZxw0Owk/HLHYufmL2m9kAMGZMts3lg7wEUmMWz54z418aNhefwicl4nDXYZpjdxGQczipKrYRdSEVjtUPShhShaMu2d1TPUOtY3VpFSoX9snGtw46rjaVtES8GAZz2HKcT03Fm1jPH6cikz4T3F/7qv/k/43Xil7/4n/FP/uSP+PTN71j3b6ifv+XdL/9jfvztP/D4+4/88//sP+L7v/8df3d94f7Xf0e7s8T9ynY54esJFTXu9sRp/gzvHVNZODwX1u8csyy8++wtP354Yls77fU5PNsToWfSXOG68Zf/n/+afv3I3/xX/5J//t//n/PuF3/K8rv/1x/2MFBqoXZPyI5cI798iJxa4pZ2nv2Jb6+eS4doBK8aU9e4Uqnas2JJyo1KYlhxHmKfeOkLT8Ug2uKsRtp7kIDTglBohOGZ1xnjFYfZEZ4eIL/jfDhRTQJXeKmWRwVeHKVEelGDJd0aTjV6TexlgIxKVZANui+ELNReCSVj9ER/TTxLEUQNwIuSSGdFyEwq4YzlUgtOKtI3Jv3Iw/TMZ8cNXjG/1p4w/TOuV001Hjdpvjx8QNXvKO2FzusHVjdaThgzkVuGFgah0K7ofMGoRm+B2v7xBQRWD5mGqhlThEjCu4XlMLO1Ti5uhB9Lfg1EGvbSMAeFqje0XNH9ilWFUtuYwkhF88pUiJ6aKpfyiOEAKSFEau+opqA5lDtBHw9zK5bhLFIocdyfz+xpJ+bXEZ9yWKd5fIn8dFn5dLlhvOJ8duQEe7+iVcfeWVKImGnHyk7ZbqAqzgqIxaoOFToKWmVeZlJ37DmD0VgrKJPGw1U6xhqa1uw5obRgHXgT8L5yyTvGaqwq6NZG3qAKW648lozkgVNGa2iVmgu5JlIdDoeaE153bO/EvGG0patEThec6TgleMk89488Xa6sbJQKtQSm2TB5S+uZkBPGWnrMLOrA8+WZVEciPb4EatNYbZnOnuXg0Qhp17jlAZktzy9PGOs4OOHx+kyqldYsqez4xlBk9w5V0VMn1xubqtwd7/nx+oIyo4u/7SvH48Rt2yhdoUVR0g2vFQ/nB+wCJSXInVICh8nB/ZlrrjzQ8Rh63KjqQkkXitNU3UYdzBme840fn194d/+GwzKRc2HfngbaNje6JGIM3NqE6E7XE7Eowr7hp5kqUCnDpJcDMBC3tddx2y2OEjTKTJRZob0mZqhOsHIcwiUikhtaw77tpGYQPZ4Xz9sL1mmq6mijmcQhXXHZN775kHlKGfTIuLSaKaShYl8j5+OAOYWQMdailKFUS+2wLPNoDLRISgklsMedTkGkU2oix4gDWoigC1u+gO74urDMJ2Ztkb3SmRB1QusXzn4at1bV0daT9x20e7WKaqw1+AalplFhVAKtolTjMDtaWVnjlaoCRnu6EkRB7ULYVxqVnBsUxRqFs/Pcto2tCXv3uMOEmgx5LxymBa9PpBYIuZFbYK8B4yvNd+qeEF94CYlSAiihS0c1z35r7DchmYDthqPT7HvElYqulXApXFXhajolVoqMdaHVnapgjTuJhu8GVRsouN0yZIPSlpAbk5uYRI33gDfcpHEpg/9wnGX8HKzhsV3QzqC7ootlmTzUzHnyNDSlCLpYmBzFeurhnk0888/e8Nl/9z+n3SrTV/9jvrtNfPvb/5ZT+g3HJ4epV775b1esfEL/k/fYf/gbylo4frmwLzOzEeKqmeYD3Xoe7jZeJHBe/gi/bez5kbx1jNdcX97z+S8tL++vICduzy8EdR2cFgmc7x1nduLf/iV/8rXm7c8zf/13/w1fzj/9YQ8DqidKcRTpTKrxbvnAXH7iPFvu+wE/n/lhveNjnDB9dMjPZmZyB2JdSK1xbJGTAt0SoQ3FZmCidIczCqU/ICpibUDJSlMbyIBw0CtxXynJoIrHtwtGXcBUknpHzSeyglYjapqQPJCS57nz7jyx5zv2HAhNyU/fVQABAABJREFU8Oae8+EdL7//SOudUnacqrSwIgieA7lOVBpdJcSMOpIxN76633loCkXFyyNvDu85TR/RvFDbTm2N2k7s16/p7kuyHJACRxNIOdHplLZT8iOHo6WGTqNgjCBm7Oy0ZIwp1FJGuwKgCUaDFo22oNqKmU9oPPvqaMWjhTHSQ9C6jxtVq2jlxp9VjVoG+1ubcbjQAkov1P41T89vkHSP9ZZQOqFamplJaaOkRE2K9x8Nl3rmmjQhFGIVivWsSdgynE6dpzXyeB1NCdKgmsU88bh36mSZz5ZiBxzDHU7D4yCFWWcOU8bMijhPXPYLVmtEWYx0Fu9ZlOIkFWs019QoCuzBUxiZANUHtKr2zl4zRQ3nvFONWm60Nl5S275x9mbAjlqjVqEa2Nvg3XfpNCr0Qs6NZBhI07ShlDA5BWnc5LUVUu/kWGh9BEWTuvLNfsVpRz55dM2cfUP6jpFGVZUcI5WKNsLl8oi1jePdkVoqISdS2gixcDh8zre3J9Jlx7g75hJwub3SOQvGeY6nz1jXSO4aqw1TuTH3QIh5rM8Wi6oZESHdPmFboOUxAhXKeIn1gNIwTx6lOkY6Ve3o5pntkXiL2ENGzPjg7KXQwyNTt9SSSDoNRbJv9FaovVFzpbVORfHp6ZnDfCCVIf2pdaaVkXWwxtG9IVKZFOhS8NYyqoAFZwTvPTUmxDZSCszLAlWzxYhB0VKmKkWzilojsYwq5bZvTFoGx7VUzs5wzRtiOqISSl5rZ8YM8ikZOzVCT2wF/HxEGYurQu9CVYaQO3HLXPfCr3/2BfNh5ofHJ563HW81OQUetw2vBAXUntBeCNt1jOZt57rFEYi1mlga02S5pMx13XGxY+2Ech2NcI2N69pIvYIMNPoaNrLs7HmM3pXuWAXyXKAP5HduG9qO8HXP47NvLZzOlrkf0W4eI3JRxL2itIUe6amgRWPEUQPYXmlFeFwvHE9CTgFyYtaVngrKdJQXjJnYtmeoBVMMs3U8vbxwayCtsfgZrxVtL5g0c7KANbReqVlQytFVZraG2guNwQVoYz6JkjSeWbWjeqO3RGkZ04TaZ7bS0aI5Wsud9yg8Z2PQVbOWRNYaKw5D5mANtMLaFGLHxUKjsNZSU+TOaO4AjGFFE1KllYb2hlvuZNHM9sSbP/0vmMuC90e6mfnjr/4PhD//rzjE3yHqn3K7fs8Xd1CSogaQemYPCrcL9ZKpHzSHu5nnH6985o5cbo5Pl4TVHW8s/SGivlh48+YE0488PMAulfvJY3Sg98AWE9uHG3f7xM/aio0T67/8v+PXiVrqH/YwQBsPPhAkFby/oeQ9RoPzivmN5hc88P7lS356/BmyO7TxHK3h0Ct7vHDSO75vqD4Ca5kNi0E4knNj2zvHqUGOeFdRauyOpHekMWolbSHHG5PXzMsjMkMp97wvliADhJNaoGOxuqL6yuX5EVFwuvuMW3R8uirq7Ynn24ZWwptZuHef8CrirMK5t6R+x54dO7C/ykq+WJ75+cNPNC7MXkjpI0oeEeLrSwa0Nahi2VUjdSGUzsEOMUquGmUs9J0crhyWN4iaXz/0GW0MvTZs7zhjqLUh6jWNIh2tFFI7SgCVKawYjvR8JAeFFHDdgrJstbJqGeN840CPl3qXhtEO229jn6yPxPw1//Y3n/P9D29x5y+ItfApdGJ35JJo3aENdHFc9gM7d2xKaErTRYhpI7eKiOOHH14QJ9Tu2btG6YmqFPbk0CoM5rieXl80YC0UceS247XG1RuZQOuZJp0qFQ2gLKWPnvDZCPMraz6LZS8NMRoEVC0sMnSslT7WP2icNpSyU3JBKYVxim3fsRNkKeSm0FVDH4fPrjVNhNYzW47MQ8cIVGrOmMORNGix9F6Q1lGqU1Kmd+FwgF1VrnnD+I7WHd8LMRQYg+TxsqQQy07ZM0oUqjhyqZQa2C4fMJPjGixGG65lZ7YTqd4wcR4PfDvRleV6e2FyM36aybfC56c7fnY/UcJOo9Clct0U111za4XZWpQStnBja2msx2xGaaH2MvbY2rDFHamKZRpkzcvLFWUtsU+EraFLwEnHWEWukSSFdImEErDesriJ0jKqF4w21B4xBzjMDosir52305H742dk7RClsWbi9vIJpTUlJ1oLlJQp16EidkzM04T1FpolrxX1quANccVaS0mRu+XMoU9MdsZWzbkbtAj6AMkKt1aJLRLWn0AqxhuwHi2DMdJCxx8ssRW2GLFGI6phXMf0ink48vjpwqVEwlpIOVNyxpmMmyDGTGJkp5Ay6JjxwqQU8+LZpVJKYSua9nJFK8MeGtKE2rbBWqGypsB32xNlXSmqUcOYvHQnpDwOeN4ppHeMhiYKozvTbGlKcdtvKNXx8zTQxEpQyvASMluspAZtzxg0kzdMVvPuy3uut0AviknNHN2MXhQvt4xqFe8NoQRebhdOk6HGihLPPB04HO8Izzfu5xO1G97frpjzHZNzeBEkJ/bLyp27w9qJ7j0VzX674YzFOzfYMShK6GwvV8zJoM0IO3fbcdaglcboUeO13vG43pB+HEHjdefg4GgcvhUmZYl9sCK8VujeUa3T8qBwkjodIfcy+DV+4e5wZM4JMEzTPY/dYt3Cm+MbbkmzB2hLJ7CytOHaCQGKvOX+j/83zPFHZqf4n/yv/4f85q/+T3xa/5pYMw+f/xOmX82U6zcgjVQPXN8rju4BxcyhR9b0DTdj4STkh4nLvpE/3fjq5zems+DuH/j0krn98Ix3kaIK5sVSfzjhZSG7hRc1I16TavvDHgakK6QJtitUrbgp0+WFJgXtClbvHLynyZXL7czt8jnVKmyJTB3OKuDZcLpgNFR2Dkpzz8JehZILqUyU5KBP2OVEaS9oGfxxOrRsSNFiFoOxATcFtDOYNsbcNRYMMm4lLQCF3ndqq2MXGzO9H6m1IMZwXO6QuvPFece3jziz4p3gXCb3Rqme2CyzVdz5zp17wZQfKOqZPSSMznQZIKPSG0peR2DaIWpCuwldPNNkUD1h9AyvN3YUtFyYvEe/1pmkJ7QdXgWtGk3VsbMWhrq5a2qxSPdUsXR1ZL0d2eOJ1Ib2FyxFLH/3GAjzGd0bJSxsT1DiAd12Hk4aq3fmw8Ls3/Dddz/jN998zVXe0NcD1Sg2bXFuItyuhLIjGUoTavU4e49XlbUmpvmE1gshXYnlyuQdygkyaZ63yN18Ry3136Wc275xmg+0WhEsuuYhaBGFKI2UQq2B0htoKDL2kfSCKEPIjc+V5qFmrJm4Fk3H0JpCS6ejoSbc66i3o8i5oA8WrQy6VEwroDR9nriGF0SE1ho6DmOikTFilVZpuaBaIV4CVjfoo9JGdxRRpJbpImirBoug7zQ0vZshQlGgTIdWSQg7CpcGFrb2Ti9xILJVI3fL7fIRax2KxuRA60orV2IzVFPQtoFAUR0pZdS9coM2Gjfb9jwwyc1i2s68aBpCCIWTPyJ41peO78Kd8qjDxEuNXNMVZw/YJqiqeJFAaK+HDd1JNTBNltbSGM87TVcNrSwlJqzWvJkeuOwXnLZM0YIUdKtYAy02vDEoPX5Odl4QDL5pfvnuC1K1/PiUyDkSwjPrbWOaDfPBoQW8tGEn7A0tHoshBQ19orVELMIshvNxZt8uLPMRo+0gVJYdwaDtkcl49OJ42Vc+Pge0ZE7HI8YOkmGogd4a9w9v+f7ygWpOZGlE2fHTwtnOtBZRcSjVdzcR4shBicDRjSbBy3oh5Ru9RfacMJMGN+yvXSCFTO8jL+HsjNaKmApYAQ2azuQF0UKvlaYS01s9Ks25DUCTDG+CahWnGI0oQKzGW4+hIUqj3bhMkRuhF5rVOK/QaqbFSIkZXw3384ROmZoq5TgT1YTVjtwczR8popjyzkkJd1bYCay5Icahe+PgFlzXlFhwTeMqXG87ZE1KIxe110zYLqTbir478ub4ObEL18uVXhslR4wkkkmDLohinidENZQWUhJOfkZUo4eC6h2lNV5pKpaqDKl2lDVQK5PJnI2n7gXEjcxPbXSgCPQmYBVaaebpQEdhzQR7YJnumL0eUCpzpPthgNR5YSoz/vT5+Lf2zmlWNN1Qm2BlIqQPGHXHNz/8lrxd+Xwp9OcfcS7i7grf/M0P5KdID5/x5df/Per+Ix8ef+CH9xeKE4IOKPWWVkFpw+Tecv9Z4bOvIrXCT8+Rl98YZvOAexM5HTJlS9hrIsXErhv9F59RnEVn94c9DLQ2RtRKhJISnYT4iNKVKhvaBJCdw3TB252byLgJ7FeaNI6u43RHtYquAa0bi2judcMoxRY7t6CxbUGlwPl0wmCpJaM0SBekeEpWGF9xLqHNeKiL0ujeWLRmYvw9oe5Apsq4oYvA82VD2wdmLE0JYmFyhqNvqKowCqzX1F7JPZBrxdkHsI2T2zjZAC3TZdzY3Ssoo5RXkFIdN3hrNDRDrxWtO9SI7hX1itTF2DGGTiP4mMZUj2W2DBIgVCpVjVG+UorOTMzvSOmOkE6s8TM+fLznuw+OK4ogndgbQRcKnceuyPNX7OuFLVRSbzwczvzs4UhQn2jWEFvnx0+f8btPn7PyJfgz3S9Uo1El8cbN7BKJJqGloTlyjYHFV/oy8X4PUC6crKVqRZIZ64SX2wtGw8kapt7YU6XZCmnF6kLIl9EYoKB7pfEqFWmdFcUqkFom1oA2mtb7v1PFeoTaFTOWkhozBtUMRStSTSxdcMqga8V0hxI9QoupMR3v2PonNJVWB2ylihm/bhUzmeHa6J1Sd1qKnLRhso6pdyBDVwjArkj9wN4Dvht6zky6c3KWrQ5an/SGtw7V+sDLdk1sCa8npBXUXjAiHLQi1MraO9pYSqtY1fDLTG0JMY1cMjhNbAFbEyXlAU3ZKsqfyN3xeLkiZkxCftpHqG92ihQyJTS6GkCU2jNvHx749f07jk7zd99e+Ngqzmg+O3U6ge/DxIe4Mk9nJm8xJRBvT3QjBOmkXAipjKxJF/TkEWMx2iN6hOMcnWlxo+plDiBlZGM6pHVHKYMuw2n/8klzuwVQkdZ3TncOKhy14eEwU9MjQSYKDu3u8fpE3BVWHzDLOKB4L2OiYzpu8uyt0ktnWjy2C13genvCZEfKQm2WNe6UtPLlZyeWZaFcL3il8F2osRLTM2ZeuD84pEd6a5ADJ6vQ3vByVaxbxx0XbK8sk/BwuMNK56nt5FrRkyGTKXUcmrzRtJZeSatj2uSXmVvdqLrTUZwOR1IolLThJsteC9WMloGfZnTPXK4rxno6cL6bqLWz5cCWEiU1VNxYphEe1oYhcROhiOXpOWK74zxNLK7SciWHC0obEE2OASMjkLcwY8RQtkx6CZyP97xVmVQtH2MeMjUn6GRoqWCqRmHpuXH0lk/bzm2/YrJH605WDXlYuLJxP2X2lx1vG/fnI5NZKHUhtI1QIo3KXkclvPeGcUeO00IKO60rDmbizsxMWnNzjV08wVpsFeZW+eJw4KQ8n8JObQqrKgerqK1hvEG19NqysGNSOR3o2g1gXF+objBq6mpIHPj6zReEIBQsLQpv+ufUsnBthu3TN9y+u/H2fMf3f/UvqV/ec7ld0PVKLP+AnhVrOKCulRIVvZxIyfLhduGkjtT5wKd9TAyNO5J2RdqfuX+4x3g1Aprul7z/+8a3f7ty2B7wbjTdplx5/ul7FqvoawR/4OnlGT+9xXb7hz0MOKPItWFqRzF6sko1St9AIl0XxGikJ9wUOZwyD3dnnp6eaFKYJjWMgbngJdN6oivLWXeul5WaI9e9cDAGVWdC9ByspTUBEVQ3OHXHDcU8R5SOiGmgPa0YdGscjeFunoj7Zdy4dCNjX/WdEekTtYM2CugoBU5VrO1o7bD2jHKQsqZK5nwEbVY6ibN5wesVoSK9Dh5277Sax3haOsaBnSwSHVQ9fk8KRjec6aQ2+rhiDK0JIo5aBIWh4uh1p/dG7I1axg2wN6HOD3x4/zU/vv8ZL7cjuR3BvGG7WUK2RBUpEhHjKDExec9iJrZnhbQzbxZPSlfu54zvFygN5e+4Plauzwd6PWGPM9U6diNkMrPKvM078EK3jeota++U2FjmSLSJnj5QiajpgB1pQT48XYd0pmqc9ZyOC7f0SJZCV52jmzC9cU07Bz8MaaVWYi0giu93KKLoAlYMtI7VBt011hhUqbQqIBqpgpMBEym1YZTmH/1cxmh0FSiC4FhT4MwBJRHddpQeFcXYFDkHmmQompbisNCVim6CBc7zTCmN0A84Z1AIqRu0tizzhLSOUwXVd7wYUArfE9LgaAxKO1bRPD9tdDSlK0rqzPaALZVprWhtiUS2lAgpYm1HkYc8JmeUnsipIiXg5xmnBGkFXQWr1RhR1g1nhuDqfa6s7yt33mGNxXTFPEEhc6Hysq6sSfGgLNp6yJpJN75+cOSy0/tbTH7L2gdX4+0y0yRjZkjW8mEFaRG6prThH8i9IVU4unv2dAXpPD/HMUEwHVGVrsfn5Gw8OgesVWyXZ7arcJwsISemxXO9BqydMNahAGNm9q2w7obaYJkMUhP+OLz3OdygN7odN3aobCmgusHOjqYh1oT3lhw7615Ba06nB6iKpzXR1kDLFdGNRx6ZjWCcQqyg1Rhf9xKZlwMeyxozvitCCYTQOM0W6YmUbhxmj7XvSPnGJd4wYsEuzM5wnj1PL4/EVz3ytm7U3rGTo8eMtEbarji9kFIgqk5rY5WVwsrpYLm/e3hdD92Y7MRijsSewFZibazrziKW8ir5cWLJtaEM3GIido33lgdr2cKVVRrRGlJTzIcTPcJRO47Wcm9mch7rRe9n3k0Td3kntMYTilQ61nlqrRxnT750lFhayRwOM1ZFvFLjxlM7rQ+2f2mZUjfW8ExvDd93jDO8bIFLTBTpiLXQDCqPC90yn1FVoavC5EgKYCfPeTZ4a3ishrBX7qaFty3zBou4meigtDasum5i3YfQ7Hy8I5TAQUZ2YlIzj9dMz4bcJy6lMM93FK2HwG4VKO+wXaHTP6DLd1j/cyxvOfVM5e9g/YZffPY9a9Ucv/yC649lIJfxbGtDvV84ui/pD5Unf6HJxq1kiiSynmhRUNVRuOLmRCrPzMcT17zwL/5vF67rW1w78nB/gLQiWfPxL39i7m9GKFg9k2Kk6YXrunPWf+DDgOiEVhXXBVoeoS92RGeUrmO3rTu2Cl5FzqcrvWpkNvRyoasEbCi1MZlCijtdHEff0SXi6Mx+BvUMVWjF0ESjxGBNRWlPTQ5tDFbqqy/ekSKDb113rErkcKXURJNCo3IJAZY2aGvbPyaRDXYatC2pnd471nsUilIiRitUDxx8BC60mjjoiFNpjPOlYayQS0LoGAH7Gj5X0glZ6LIgyiCq0ijDLuY1XXlK6+RWCalwOCmQOOhoqtJyQnQf1bcKKMPj7S3f/PRPed5+TTUH7tzCoivOPCOSaeqGNZ0aNdSFJjPn6YH1siEHy1ZuGCucrEFCxk8eiR3fPNqesOrEx6Ao1tMRWou89fAzvZP3Z4pS3KYH/u7xmTp7cn9mvT3T+hPadlJ9ZNsu1L6QOVKYaK2S9sT5eKCYwjWuWG+QkpmqkHJCdEPQY9/eKvOyUMMRMWC6ptaIkc6kDEZZWq4Ys1BzoeihPW054rShisJ2IefI2grL/YSERguV1uFFKpdv3mNNwxhNlURpCmcMy2RpqtCpaKVw3qC7oxaFL8LiHHk23KIhVj9CbxIxacPKsBAK0MUT4oYYQbrGqAmyoLTh5RKobTwQY8kgg7RIGS4DJYJuFWpBm1fWgYwHmOuKWjNlD2Ms7zStVrz2TNpjrWUqO+Ic5RXJugu06cyPTzcWZzjO8DAptlh5aoreNWvbeNYG6Tu9G1Qr6OUOWyxv3cKn724c5oWwXTkcLUoMfnaIP9Fjxi0LTjtSGVkbb6EVRWyCXtRAVRe43G60bcf4NNLutTNPglPQyXx8+onLBvgJqaCbwmvLGhLf7Y88K8tkBKVO9FKZphN7TCymsK83EE1pF0pUWDXRNVz3FypgtWZ2EyUFUB23DHph7RnTKqaDNR3vNVuNdJ1QvXJLN7y3OLugrB2rpqKY3Im8ljGO3kYb6LB4as3EWjE0tnQh75HWM1ppVFfk1lAI1hjK2rDF4a2iiiYq4bIVMpHZD1yy0wNiNKmE1RY5zFy3SladS7ihtKLulbZ2ghQuZQQDJ+eYe0Bmz+Q8Yb8wu5nYxmdmMgtaCueDZ6qNxXT8srw2OhTd3PF06VjluDuf2G8vbE1xmBbu1ELeO1MXfNfsoqFopGm6UTQzUbHc1suYDJkzJQlnd4duhb3soDpeWxodbwXVI04XiupkVkR77KTozY4Ao7M8iMFoIeSIrg3ZOroqJpmRbpB2oJbBRzDVcLKaO+t521ZcjlxqY2uOAJTeMC0MoFdr3K6BLe+IwGT0aAcVw2d3D0xmQqzD2JnpoDHryvLNvwHveXf/wPbx/8m2f8vxs1+zuC/Yru9597OFDz/8Lf0AVd7Rjgl1MNzer4g2lKC5pheWg6PNln3NnN99jifx6YeMVaAWRbMgkvjil2ckRGJb+eZvMtTP6ad7OplUCgf9juf9EykWnO1E6dwfTvSkKCmPS1qXP/BhoDUMjaPTOKBrhfFQWwEaxjIS7LpzYmMxTyymcWuRd1NA+ytG79gUR/9dC2KErVREFNZ6YgzYk0eZF5yxSDMoY6ApaCdu4Z5d3vF4K1zXjJqOGHtg8sLMj6Ch98YyHdhLp7Tx4lZowDLZilKOVCveKEQUtU7UVmh9dEstalgRKUyqAsM1r+nUKhgxODcNoYiXwRhXHaNA65GGvaSJWGdS9cS8U1XhaVK06uhtgCxQmpwTS6sYp2i1osQMyM5rGM1ooSoL2xtm8xZ5UJxPiZ/xI8f+gWouJNVG6j8diRdPLJWuTrQKz/uO8xVNwM2KliNm8uAP7Bl2ZejzmVwsRs+UrqE2bFfMpSBhpV0vJKsIfuKSOzLNw1ZpIhOJ3DJaG6zJWJO4vkRusWPtwFZfXh6ROKpCVitUKcQamTXkdKNOE944/K7RpVKLIhQGHlcmSk6InsYaRzwlVJI4Xl5BSWhFDmmMnNFM7kSrgV4sOnfu54EJLgKpdCancU7ILYD2lFpAKiMJoOhluBH2CrELe+lMRbHYmdvzjrEw6Yy0NPbfreEEYkrYuxPEgJjKWjKuWaQ7xM+kPdCA0zTjuhpVLi2EmLkpTa+RrjrOGUoOiNKjNmXVODDVhtiGnRRJMk13aAnvPSU9k+JOnc+EuIGqiFJENupUKRauOWKjcN06yDyIbDS2NhLb2ggfa+XPvn3P21kjCL0mSu7EnPm0RuYeSL0Snnam6TMmd+B2uyKF0fGmklIfqxBvWbeAiKaVhtHC8bjQcoQciKVRu6DQpJw5zgvdevasyAmwDmkVa2dKyWQqBzfx+ZsD2llira+1rIo3hnv5gpAKL/s22jTKIL1zN3vidkUqLIcD9EbTCTcXjHdjKqQysWWQSshPtF5JFBZzQHqnRsh9osY8Wg7K4HRnTRtFEkZbSo5se0bozLOn9YI2mqKgRkA0vTUcCh0SB+XwfqIqw+1yhbpznCcOeubWEmSF1yNq6s2oFt/q68RLOilvzNMB584oBO9HZsUaxR4V4hSJSmDou2sruGkii0KcYXaWk9O0HhHVOSvNcviMD58W3MFCq7SuKWj2XNDsWC2wXckyU41hDzAdDmAcVRmqDFDQdPgMawStDdoqltlS9h3lFKVnUksYA6fpMKZqViC3EdAESlFcbwWD4nxUWCXUHqnpRjLj3aOVYJeZWhx1mtj6TF5fEcrGopPG1Qj7Dqqxh8auNEE6U9Y4LB3IrQLC6XhktpaYGr1ofDvyxt9jliP2fMeiJ/bf/kvMn/2X/PKfeeqLxvidml5QW+b9p7/i9POv+YvffEetlrsv37GnAxIdP/105d3dF7zcLlya8It/9kCikm8RbOYpfOTYZtg14iu9bNRQueue4xVSuBHSDM8L/W5iVwWrNZ+dvyC9XKl7QMQyn09sP/1AQHi+JYJ2SG1k1f+whwFLQ3ThfNRYq9lD5XASKJ1Wx8vQaqh95WQvIJrTu4ZMO80VLuET2lSOXtNiIoujG7iugdbH7S0Vy/MLfHbvSanh7QFKQ7Sh1p/z7Xef8cOjQk/3GPeGIlASfO4bX70N1P6MsdD1Ay/7TMyK0hVWH8lV01pAUFjxPBzveXz8QK2eGBJORgJZa431MFk93OS1jiS/FmoStHGY1hFRhHSldT3AKWLoWaHlRMsLVgyqZUoqBDfx/lPElAPOW9AVjKbWG0atI+ikHHRQasLKOID8/1j7j2ZdszQ9D7uWf81ntjkmz0lX3rUD0N2AYEiQBClFMIIRnEgDRYhzaaJfobkG+gkKRYgSQRkySAkQQIQoNLqb3Y2uri6TlZWV7tjtPvO65TV4DzHGoH5AZuTZuc96n7We+74uiSQXx1YW6vYG7A0be2KXbrD1SDLDWhmaP+R81mitqCbgUyAlgWm3YFaokV8pCji15+Vd5LWXTNGwFI2sha542tZRBbhScHjCMhKEYtISTwVhOC0LgiONPFLKHUJIqmnQKhJSoXF7yigpGpSoHM+3pAq6cUQi0kEYR65sh4kJKRLzlFYEb40Yub6sCGkZh5HGmJUBXszKas+Jk1jFUqpWojBsTM/W7umMXV9tNgJF4WpTWKgkmdFSEEsmlUL2mSLXP58PAYykqIzVEjLEZcZXgXQtVUKUgpArNecVLhUiygiK1O9S8pIgDNTVr+BTIPmKtauzYFkqO9XxSGsccn1FawyLFMwJTsHT2wZZE41bValCrIMOgHGW8/FMFBVPhhLJuRJrIR5es5GKVlkexhFtIVeINZJTwWq33kirJE0rO71zFte0zOdA0/csIVBEJYSMsNv1lhYFsjHkuA4LxyWw37XEMJGKoPgzTq9rv7hkkg5EX3lkOq6Vg75HRsUbf0I7R66JeYZG6jWnYRylKkrKdFuDKhZtLqgpsn3UcQiBEY8wit3Wcdk58hKwRryz0yVGHxG5YoxY0+GuZcyCWBeEXKuGImc2rkNVsEIiBcQ0YazEpxOm0cQQyclTisdQ6Xct3s94ITGuZVw0fqmUrHHCkHJaX/v8SJWZmqDKRNtqzucDZZkxetXb1lpYvKffbBEkaqxQElYbSkyEkpEhcNFtkFS0B1s0trEYVZElIdV6trTKYBUonfHjwJIKTbdbg4fJM44j1rB6MVzHPESk1OQqVoaCqWw6Q6uhVZI6TWhn0bbHCsn9yTIvApkjVQyEI+S6VvoWJBsFpfE8mEg2isMykrue0zyhcsOz62c0esPpvA7KU55gXnBdgwJiqrTS8UHfcD4/4IQi5fWj71rL3jS02iK0ZMHjmpbGVKYwkdWCqZLFe2ICTEOUhSTLqi9eRkRWtLqh5oVqDdlqyIqNU1xby4whLuIdcl4wL/M7GiScTwNRabruEc44Nts914+eQpG8/PJT+qc/5FuPP8Ynz/u3mkNyqO4jjNcM8Q4nFJ//yvDVzVOKbHmg8vj5lsPbiddfRvrvb7k9n6nG8evbhD/ecmUU6TwTjhK5v6bte4bzW8opcy32qHTCHisqdIyDpZEtc4w0emVYTMcHpulAyQJTGlxuaZqWWB7wsuAtiBSpaf7NDgNaRqTI0GQ2m3fdxZxW2xeril7ISq0jrV6IPhDGNxiXSDqwc5nWOsp4RJmKbSqRI619j51V3M3r3l7pHUIa5hAw2iLzSCgNL28e8frufYK5QIoCWjDlTKGSimdjF0Q5o13CtKBqZmFHqB3noPBZI8UWIcAoCcEgS4uRmVYVVBpRziFMg7MaIwI1CsgrkMOnEaM21GJpZcaHASVYJ2sFKa6VFaKhkw9cbj8lGwi1oeRHsGzR2XE6BpbWsBgN2qBqQSyRXl+S0ojUFaNnSp6RWpIKNHakbr6myonejLRmoqYJoQPKbPDnSBYS5QKgkKZyGBYmBSEEplrw1dCZDmZJPK+sCKEU1m5ZaKhVkzXk4NkWj4l3eCYOG82tVJyXzFwEvka0DFgbEKyvIcEPaKVJteKUYqMtxmg6q4k2cDqfV7Oh0/jZE0tiChNOBJYYmSOUkslZoVTFZJAoerdno2GnFVSJyYpoGzQQasSZBrxmZy5oRIPOKxtAaMN5OqGUJOTMEpfVR549SIFrNPPoWctEklLXUGKUmiwLRSuSD+iQ6YzBx4hfzrQKupxQ0uFzYcgRayzTEDBY6m2gERakRgj+jZGSnHncdeiYMFKQVMUXwVQSUhlmsZI3jbb4eGbXbYl+WHvxQrEUD0isdIiyZhKMs7gl8bRR4CNNd0EewGtJFBldMxJFpx0XTYObBxrdUJQGt2MOilAlxIRkXZH4NDMFQwfcP9wTSmUpkdJ0nMeBstGEMCFlQymeXBxUQcWxLAahLXJn+caTx9w8DDhpaOlQejV9Fma6Zq0IO9NQfKaW8C6bIYghorQg1pXtoU0BZiRrXiL6GSl7jNGExRPCwr7rmYYZozLVZnxKzCmz7Sw1TBymt7jmEiXB6zVEOYQJXwVFVRY/YK3Aqsrx9o7GaUoSlOrI0a0K4qwoBYwzqAqygG0sbf+I8K4toLSipHdIbSlIOWKlQSiBfKd5V1KT4prw10YzLYFlTigpCHFdTy6LX5slKhFxxGghWMZ5gRrYXfQYWxlPyyr1KQ1LTTjtEKni44KU0Bi4q4FYVxeLJKOrp6ZEFpUpFUyRxCKYF8XxrPn69h5hWrZuhXSFmNFKI8QaZg7Sk5rIycJQI6lPJDGSlKRViot2Q40G9e4mas3KkSh5VZJb13HdOp63FrnZc5xGUvS0G5C1YIxhDIJQMkYpSl4Y80SowzvscaUzhvMpgFakIlhypuRhtfjtN4Q4ootkoHK2hjpkjJ+46CxDAZ81QqxVxGgcVSlqTegseHT1CKk2LNniLvakxsGp8PqP/jVD8wnds+/iuu9jyp4nWXA4K6YKdl5pjMcMz5tLkin4pJjGieunihc3gWgzdC3HaUKOlf2F4jTdkkPicXdBypLL5z9k+lpT6z3zq4HLTmJbx/HkWU55zaH0HcN0ixIL8z34IsgYlCqMIWBKRtXI7nLP2wmSX1spv9FhwFhFLKDaxO5xQam11qGkXpXhNSEzGFlRpmPmAghINWBMRQuPUh7sGYGns4bChmV5TVp6kC3n80BSmftFU+SH3J5OPOkFObZ89aVhrtdUbYBKzO9IfMyIdEa7M9qMVDGR40RnM7r05NyxeEOoEiU1UlScW18znNmiRWZvKyQLolCVJkeJaTrmWMjFUdV644s1rR3/mtZhwjicPlHlhBGRRgdEHui2BSm/pmygxGsYFKebDU25wNXKORtuisI2LU4URKprGEyvFTlRNKpqSBUjO0rOtDaAKdQwIXUgkVBGUDFQHKgAam101PSIQVjO1jIGmEsLtsdpRzo+8ER7NtvMmAYORfGZl0TbkGOiE5WdynR5QTeSpSoOteGwCKQVtKXiyGybSo4JZEC7lmXJxAxHf490F6RaOfmJ83T3bu8dkYtiGWYEivNyhtYitEPIBVXXVLkEdv2OTut1pxoGcgoIWd/RIQtWSmotxApDCLR2YQnlXd5hIZcZIQXH6YRsupUmWAtGKBqniMtIXma6TUOScPIzSSpCiKSUKRmUsIik6EyL05JCYdMr9qaFKnlzHlHGototMQW6pqHNlTgmNkYyZ09vDHEJKGNJsdLalQ1QaqbmipQaq1YkaxEGJSSd7tHVc2F7clQcvWBQCpKkM5YcIsKuSfutaXiqC6lEclmDjrfvMiyyrLUsURWESmc0Ki64Zstx8RxvZ4pWRBEoUhBKIZXCPI4kBMfT1xSnkE1DjBGnBM5tqTFSpWAKhdkHED2y0QjVE5stP76/59X5JX/nm9/m9eFM8g5hNxQ7UuuJZRrIVRFEwZgeUVpklQht8UvBdA2zP6MbgxKB1jmcyMQ4kUqi4JjmiXFeiCmw7XarIS8v+Nnjc8W2LUVV9ruOcDgRcqAoRRSSmFYevwLaVpNyWAViNbHfdDS6p7UbAgXbbJCqZU6SEGYolZQ90lSO40zf9BQKVlXSvGAMaL0OdUtZiNUjoqTVjhbNMs/Ibbe6REbPcDrTNj1t33JcPFVIVNtCXl82zrkyxUqpqw7cOcs5DeuZYSHVRIgTKRc6IVnCSAkTWki64Gi0wTpD11isylgZKERyLFhjabYb3txUDodIzAXbaBqXaGxFasN59EzLGefUOyXviLENi6yIIqjC4t8pyRu9obUdx2mizglpKlUEVM2ImOmlpTqDC3BhJBsp2JaOqSoMZxATyu45zYXjcGauCWMryq6yruTndcVnLJvO8TB5shJo5+h0RcSKn45I2RKKRFvD2+DxVdAsAe/v6LZPaYrEx4h0EmUVPi/UutbBQ2QFjkkJumWSGnaF7/wH/yFhSUR1zVX7hHh7oJkrLt1T6wEzZOL9wHP7CK17psZTdMLurhmXI8nCm/vXmM6hG0kKE0ElzJ7VdCoDqvXcpddERuzWsyyVqCSpCSy7mfPbgHEGcTpzUSeeuC0CAaWlsVtqGJniwL69ok4Rs3W02hC1Ys6/YeiQ0JpGJNwu0u4mlIkoYVFq3XcqFFoCCIKQKz4TBfWMFAvIB0QdMTYRvEeqDTE1lAI2SprJo0tAW9B6y8uj4+Ad3ZVkfkjI2bJtLXOVa5/XBJ73mVaf2MiXtPKWUg8UUQi1UBhoTSBXjfIOYzSVTCkBBSv/u0gMFRklNXf4KSBFRxEwhUI69oRasD3YxlClgqrxMVDEhOKe3j3gmhNKJsoaSKfkNbOhHfjphmqeU5Shii29UWQqtmhEVcTQ0mDRQlKyphYo1SBVB6xIUMHaz6614oyh1IksClXIFa9cFqzNKDnS6A3DNKDdE6ZQKdLhlEJogygzF93ClTvQNwe2InOhWu5fb3kTKzEuXDbgRMRJEKIitGUeWUl+ZLaNpBMVJz1RZKp0FGHo9jtubwYexiPZrDmJJc8kJ8hFkOZbWqWgLuy7DVpo2k3DeUnUknCmoW8dpQrC4plDwoeKNJbT4hGqognYKig50lhW22BZw0XEAlJSc0bkQKssMxIpJEJocgogKufD6d2XUqKUocqMRlBSJQkQGKRShDnQWUvXbFDFkyv4eWEuEasNSSYKlnGeEO84F2gLjWTXdTTzgi6eWSnGtDYiRqGoFU7TgLOWxhmSkvi2w2qN04KaKhuj2bDetEspCLdBhxmnNe22YxaVTEGNI70WTKKQhERQ2MiVp7D+jlZEXgUyqndYoSEmalWkPCCNISdDkKz0JymoIjJOI02vKOYd3CYuZKn49MXXtNbiU2QcAzWPFDVjmh2CzPHhTBBwKwvLTz/j/V3Lh/sdh7dH9k3L8/daCA1OtNyMcBsVUa9CJaM6cjuvuR+j8D6wVQI9R6SOhHmtD0aXCTWSjQYK9+c7rKooLfGhYDCIJEFWnm4fcXuXGYshkck5kUpCqIQU+V2+onJhDKoqkrIo2bIshlQkyrOGo3MmTiNVSjpbCdkzLiOotR4b00gonlo1ulZqKfgwI6SmUx2ywsY4pBCcFliEwKkOekFSgmZ/jc9HUgw4a8gRhC+cp2GVp4kKIq79eJEIVVClwChFyBltDEIrijJ07Z6961FaYbQnoiCteZ1GG85TIBdBTgvHOfL6dWS3e8q2c6Rpwk9HUpCsTk2DsgqMQSpL07VrtoVIKJCzJuNwdou2PbNPVDS9UaunxlW2OeI+f0A+6onZcdUonrczXVPJO8ur40KRCut2zEmSkyfkipdryFOVxBxGtEzrjj8nPA1LUggBWiXqMqJSQiaNVwkpe6YSiLGQY2BbK0VWiopMyhNyISfB7E/kkpFUyqYjAijYbXfUEDjVTOcu0Y8+AmMZFk24OzOMdzyWF/S64ueE2CjmryOhHIihYB91bJ4Y7suB02Gm6xpkI8jpiPUFUxXpmBBtB6qgVaEe73l4eyQqzVAz3QeORQyk9xuKa3hanqKDR0wHdhcSyT3T7popO7KqyLSu6UtcUd3VBNpmz7xknPgNcwZkETzaS9zW47oTrfYIIZE1o7VD5IIsKwgjGQ2NBudx7kyWL9D6BGWhiIp2EkpDjlvKtENmjU4LbVPRLtFuKrd3laY8xqUC6Yg0jrNQTKHS7+F7753o5M9pzBvguIJqSiAXCEumaQr4yjILcqzkd1Q5IzUyLuRpRpse5xSlWFLeE2qkFS2d0XQ+0nUNWixkExiN4i5tSNGilgm5PaLsyHVfMXJ9BitlbRSslD2QCnCacTTQbJkmha3lXbhCsiwO321p5B1SrNhhWVti6JEmQjqiyAir1o8XBWsLtawmvypgDjO2H6hVIMoZKTNKLigBBkfTCK47yUVzJi+f834X0OUBuKXUGScEH2yu8TeJ6wvLXky0+YxaZnKcUabFNVvOvpKJK0MhZZTt8MXy5l7jxZYlZt7enwhEsvDIziBqoTOGKYExDa0uLDkQ44l+d0Wp6+1SKYVWCqkU8zwzJ0+aZzauWS2QxjGlVajjhOTSNGs2QClKnjFmQ46F4CW6VpoiUD5jkyRlRQ5g7IZYRpyT69O0XBP1WvToGjHCIHJm9ImkJVk47k4znTJcayAmcvBMMRGtwbk9pyIJUnLZO/Zdh6qOUmfMfI+cEo1ThBI4pkzX9SD0eih1jqwM0hpIHpXWNslpCmiZMUlgcsI5gZEKg4BmwzSfaWxdkbgh0qqWgqRamHKlaL06BHRlGE6rpVJqJIYH1aK9wJjKWAq5LDglCRSEcO8GmkwqianMNBtHrpVzCsx5JBuF6hTneaIWRZAL2maEDFRxwpkNRir6/gppWk7FEubI06ajfe8pbrnhH32v53FnaETLNLR88TLxxQBfPBRO0eAziNHzRFxTZeJBT5xKos6BOid01zGOiaUGhujRSuCMxlqHUavvI8dVStZvHFtjya5FFcvJruvKGivWSKSUCGkRFRrt8ON5vbEjOM1lzfFgEH6hlIJ9h6jWSuApCCtZxEyeZ8DT9JaSCjWu0KYtPSlKdm5HdoFcoVaFn1efQotAYyBmOCQu644sImmYodmy1IARCasUyzyyu+hW0E5d1chNY2hNi1gqVUiMUPTNho3WGDRVK0pOgEQoyzwthFi5P0Sc7dhuDaoKup1GWkWSmag9VRdShmbbY6RExRmhC6fpFmOuMEZjpENQyNWh9J7Wrsj15Cu1KJTKZCRL1jwWkovxhnb6ktDt+cHf+A56s6E4TZYLXVUU2dA7gwsKcxSUeUFbh08jWga0FahSMFqjtWacJAaHKBJHxZSEq4FaM1U2IDUlVRRqvVwgKFIyh8AsMllVXC5c9RvQivM4Iaylvbha/RZyva3bh5G+wOX2MXcE6HbM2wvcd/4mn331hu2bV+yaK07Tkbk1+D4znWfEmHDG8WjXYmJlGkbUIaLHRHyArCxJXmBER80zIq9nNaUQjWD38Te4O9wxdC3Dqwl/d+CbV+/hvhrpd45GV5ZayI0hVkcysBeWbmkQ9yNVCIRwaK1XPsy/XbPw334Y+K//5Fc83V/zB6nndzeZ/tKjRUKqjEQgpULSUOtTsr9Edz1V36DNDdqdQS6knNd9Wiko3ROGDWm8QEVH4wq6zRi3tu6dFsQgUIuiyQLrVj99LQJTDjzefInmZ1AfiDUjlMUqzXCMtM0FuWiqcKRgUcJSq4Rq2DWObVmwuwapW2rO1AJatrS2Y+McvQS3ePp+QSjPNAte/bLwxZeRx9c7Hl9J2rwg9gKhJoReq2Wa9VXA/I9l98KaalcbmosNx3TGthUtHHux4VwXpJhQRlPIq5CoWAgdRa6Huqjrbl5bha4SqTKpRGy3rm2kzai8esmVDij2ONMi3+3GRA1ctye+dfFrFv8rjCzkOCHURGMlxiyE8YB1LY1+Q9tO5HpGuAGxwFb2aG9JUVOoZCPx2XL/UHh9B1O+ImmN0xrVRkw+oORCrInWOXRed5ZN1xD9kUQlpoVcFnIS+JgQQqNEZZknUqprT98aWiXIqZLkOigJa/GlEJ2hWoNfPE2zJrd9mdBys6KWReSy26Gs4m0GWF8FpFTrASc0sWruhhM5Z5zSyAoOA65lBgrrQDscZ/72D77H6fWRl0vG9R2bfkOUlhIMB++RojImjzbQUlAyk+tCipKu7dDJ/5s98XkaiDWyVI9Wlmk8k8uZpAzWtpQsKCXjEcxLwFcBylOLRpZKiDMxeGrw6P01UTeEmjiNJ0KWSO3wPtOIjn3XISQsoXCcEo22OK0pS2bTWowqDMvI4KHd9kgVmMKC7g05rZmSKgspL2gnGdJ6/VgbrwLVaAiRru2J6UxBM82ZK/scIQQnZbg9DmxUx9Ws+PTVyO6bkd4J3n8q+eiZ5Q/ZczduOY+aGK/wD28J00IKlZ+ftvzZYHg9z2vt8B3AS4pMYwUpeaiQfUHKSomepunW38VUydVjtCDnSFGZJQ4oXdbaawwU40hL5O68UFJYPwKuwbR7QGFag/KG5BONnNhd7/ExYGhI+ZbqPVYrtG5RVdG4nq3aUGMhy8ppjtw9zMiaMbqShCTViggBZQydVtQCD6/v6S6eEALsXIfXgrgU9t0FMU1osa7GRDXE5HHKISmExZN9xjqHqBGAJUFIgSwruSSiECzTjKqZrmpc29O6nhrOlLpQhcSHGVUlUgguH+8Zp4WlBGTQK5OlJByGvKSVPtt1TEOiJthtoK0GOa+WWUHhynT0tSd4jTvccvirX1JKQH7wGP+9J2w/+IBZaYawUJQj50KKgZglWkuEUkgR0Wo10Soq19sdvbWMScJJsdEdxghII7ZAr9t3LI0N56xJUeL6CxrTQ/LEmhhzQUmJq/DEOoyQnEJi0+2IsXJaPIVKsZqLvuOpu+TVTz/F7m5opMRPguR7lmd/A3H1iMgO70945RmeaBYDAs3+WYPbVV58dQ+/tjyeM7siaHiGR3DOidDs+cbH3+HLL34Fg8B2Fr0R3A1HDl8fadSegqCIitEngj9w2bcoIiVVqtYMHryqOOMY7l/RFkOrNhQhoVaEMjRbh5G/Yc7A2/GK17cX/PjzkR/8ycz/+n91yaP9HbvrFl0MsXTEdM3bu2eUw7fRzZbmalhVjE5RdM8wDThZMO+eNJ25YFg2qCpwjQMNRS2UCC5IUlwwdn0ejyVjpaYEyCGhyg2ivqYyo6WjVoGskb5tiPTMqWPxW3Ju6JxFNZYkYes8ewlGJ6qaEEhUFWuVSjpEgSQjm05gm5Y3p5b/4h+/5F/9SWSeZtrmnm890vzhjwz/4D+5RF/1oE5QQch3P9D07ocmJVpuiUpS9Zn9dYNTM3NNsBTCcCJoSbVXRGn45ReJT342UJLmuz+84tvffowzA0WdyHJAVUEBlDZIXSgxo4R4R3xTkDU5OiQd2ji0EnQ68ah/QNXP0OKBJab1kFQe7Qw5LZxfvaU1T2jlEeNeUdRIvRKUMfH8XBnTwtk3JKVhqowpkfSWoCsjgIyI5HGurs/4KIIUCCHQwrCcR6S11LD2rpWVqJIQWqKkREtNqy25VOZlohjB3jmaGFBKknLkVAoYidSSscL5MFIoKCWJ40AOESkTvlqWnMgRstxweJhIoqBlJMc7RCNpuwZqIoweSkGZhuQnpGuQRdLrlqotCwWZoQ+VfenxOWKcI2fB+WHGFsG3XUurKgXNQ3FMasscLVIvdG5FlPatoghFeaebTaWuFTEtoVScdSsTXSZkrujGrTvKlcJNoyzLsv53ai3ZkAjVMNfMXfKUoghZUHWl1IztG1RZb4uNqsR0ZomJqMBIxTQ/YFWgFkXbaoYw4/3D+hRbA4dZrBAY16CcomklqYYVoiXflSa0Ipe1oxJzRdqGcJronKGTgrP3nIaEVhuOKCZzxf/5p0d+8krx7UcbfvBsw3euJ/rmlg86j2t34M/U1qOkIAvLD8c9X54io9mTy5loBvTWch5OGNPRq45OCtK8kKl02w2bpmVePGFZEBTa3hEHT4gLtQRMY7BSEovifvDIkkBLSJ6qGrLMWJ1QUhH9xHh3j9UaLTN5OrLbXhCEpaY97t3rnxAVjaZ1LdOhEJbETMBXqFrjLi4YpglpNbYIeil5vOmxsnA4nZmMZykTrl8lV3OMRB+JMWCcwlnDPB0w1hKqB1VobItErSTTFPF+4TAs5CxpjUNISdUS6ko9td6jY4YcELWgU8K0htRohqGgGoUUilQSkAlxwEiLLAVBWdkHuqMGvb7mBbBJ8NgrtktGyg1l6bEYbD7RvDzRxMqFmKmLZjovpM4S9GNy0yOcpRwDQiiafk9NE1OYiHga55hKxGiDK4GttPRF4pCMUaOFRWmFLAUtW6SUyFrorWGpgqwssTGYpqVtt6RlRsiKLxFnG9R54rnqMHX11ww6o5zGZENtNpR2j3IXxOaC/e9/g/nFLRv5KcsXf4HwC+rZW467wMm+Ro13ZFYvBcpigkdWC8NEVyTpNnIZd/StZAwTerNBuC3FPYbt++jLE1mfSTuN76HJe8bXR5zW5NdHZK1Y29BkQ0YhN9BvLUJnzKFHhIZ0XLDbnnD0FLMhhco83+NrYWoUSv2G1wTGOqZJM5en/OQLyX/539zxD//uY+QLyTBd8stPDEf/hJc3lq5OfPdS8jd/v+GHf3eDNC3ESK8ajC4IJCnb9ZYvNzhZqWKgKrFytwsYVdjoSE0JrQ1FaJS2sECYE1Ymcs4rqldUSpmR76QaKUqsvMSHDifB2Aiu4HVg28008ojVgqQiIRQUjiVJ/JAYY6VtLM+ur1iK5H//f/hjPvkrhXAbsrI8RMdfvpB88cWRP/vFmf/t/+4bPP/mLRARch0EagRhWCuJYqZl/aBXbxCi5ae/8vzj/+YV96+O9NvEB995yutD5bNPBFN6SjWa7l+N/P7vbvje+4ZvfXTF+++PNJcVbWYQkppGrHG02VKlIsYJKR1C74g4Fu9JNbPfVXp5QFPxWKoSa++eSBKaPApEcFhXkUwUHqhmIJQzpqu4ZuAHl5nubsMcWu5HB/OO1yFgGos/nZA1YGWhMYq+2VPlags8zoFhTlTTYBO0quEcC1VmnFUcphFjtusAlgI+R7KuxLpCpramIZ9PPGocU8jMVVNioWqQSlJkIuSZJCrYgrSRMRVULgzzic12j95FLpsOlRYWI0BlYhiwXbea7qSAutrBCgIrHEpalDaI2tJUx64W3ttLlmXHTVVkNJt2g5WSD/qerVwlVX/6dkTudmjb4xqPTAPKCrq64FlNewpFjRFjVmDS7Ge2zqKVIedCjYWBhWk4UUpg6yz75pI5VU5xojOw+BNZKu5j4W5YWzrSrKhiq2CaBpSS9M0GKzWdqwQyAk/wnq4XkDRSaKxUWOGhFqwpWKtXuIsVDPGMVQ1LXINn1kJKCVgR0T5GetcgRUaSaZ2krYmmQtaOoUDXNYSQiQpee8Ny+5i/fKPY/cLztx5l/tO/0/G8mXC6UtOCsQZUQUrPldP81jeuOP/qlg7Hr6d7UgVLiypgdMEoSRER2/QrzEtXwjIjZSZq8DExLCOhrLbOOmWqkbT9DqnO5DpQNbS2owpLqoJYPT54mhDXijEJn/zavPCSYR7x0ZOMXg/aUth1mjlUbsaZ3XaLTHXd8WqNcBpVGqgVqTOdVaToiVUQVYPeG8aUibVSpzNWKzZbSQgR3VSsUhjbE0qhzBklYVhGFJKw+LW10moyEtttUM7ihxkjJE5ZpsMdyhpySOx7y9VFz7h4mk5zvhkhRqzUVFEIS1j/OSPIfqHbNJzOK/9jnBassDzaP2bXK7KXdNs97SlT3lT2myt8lTQiIesb3CPL4ZO/4Lp8gXFbRmn5/GdvkO9dcPnRHtVImjRyPg8EXzmMCVsVj4zifl7PiA+N5WknmJjx7ZbbKeNFQpNXd4iW63q1QCcNZanEulClZjivrZF10BO0uz1zTLQ4rvVjtrZlKwSq02S9kObEG1+ZhaCXG4Tc4htNeuKYjwv62Yw//jEPy/+TIxsejvfYZmEqLdK1PLXPaKKjDplZLhQpqSaTskPuGsqcELuOgubV6Bk/+ZpHXUv/NCNMw2HynI+JZrB8/A2NFQIxS+5+Jbn81iWmCRRVaLYKH44sdcZni41Q0+qdkL0jLK8RNq/0SgW6+t/sMHBpKtIIkjQEf8W//tV7fPT9K27GyF//wnB//5jDsvKa3ZJ4dX7Fy08+44d/r0G3kqoNJnokgYokJ4W2O2QXUWoEdUY4KDkhk6TbBWgUxEK2q053WvTavY6grESEVde7+u4NqhiU7dCm5Ty10G4xySPw4DKoleokRKRKy+2D4P6+5/Xbwh/96UsebhZycTRN4X/5H/6IH37vY9581rEzimITOSu8EPhSmOUlP/9l4f/1X2n+s//Nx2j3KyjvnrNb3mleKypIRLnkJz/T/PEf33G+G/j8heaXt9ekLGDW/CJseHh1RKZLotlSS8OhKt7+uPDP/+zIvrzit58f+J//L3p+9w/3KCopZuo7QFIplcY1pLTDlyfcPawhssdbxbOLARmPRCWoRSNIUCXKdJAty+JoLi3CPODtHUXcY8WMUTNaQ1wEUULfOKTcMS/PkeUKKRW6ZqwRNKbj2kEjItYYCpbjMLHkhoxg12SuyoGNmrnFcswrwEkAiMw4LRRjkVau4hYpqASktrROYFrDJFpeeoBEKgtaZ6gBZRIxrypgqqAYxbQsBO1WWmUn6axHThMhJ2a/7u9rljjdUJVi07SIvBDTakpURiBVRaBXB0WNbEVmd3XBr5NmiIGmsWyt5bLfIMYzqSR2bY+2DhdGUopULbBGEMpa5dLGrqIfKjpHYkhYUSixrJVIKcjJk7Sl222oyXPRV4wYCDFTSyGkjHaGBAzjgpaQ50DbNBRZ8GkgF48EptnTb3c4Y5DDSgmUoiCEJHiBcA6FpG83bC9aYrpnnAYoGmEyrjfMs0doQ0mZRjXk6BEi44RCOocWAmcss5/o24atNMRcGeaINIbFr3XXVGb6ItY2QtvytVc8vDYc/vuZv/uh5Lc/1DxyDZs6YmpCSInOiX/wLcvvf+Mjpnnkj34V+eWDZ1SGF+cDMq70Q+FadAXrDLYY4tThWknXaA7LxOVlR5VQfUTJQBGBcUmEVLBygxMNnVZ476lZknLGGEOn3Jqr1BKiRuhKTIEsCkUWIpFYBCJBRHIcJ6JehUq5RKyRK9diWGiVoiCJfsHXhJ8mSjVUuyGU9E50VkEltK0YVxFNJdaJWCIpSapWJCqlQK2CGBJt268vkVWRckZnkFGylRueXe0JJXNU0DYWHyaUSSQyKYwYu+W6cfgpI0Jh0zfknNeB8DysMCNliSGD1ChpsbbFNv1ab9U7gtuxe+x48dNPeWY/ph7uGd78kiDPpPyYR80FarthHhRhiWytQaqW4Sx4GAQu7BBeUha4CBmdC42wPGr2bOsdv/+Rx1rH29zx//v8gTlUSjkR6toIiTFhlGYOMzoHnLGMWeBjhAQhJYzRNGalZUYM2a8cAuV2XFRLPd3w85//GU2z47J7xMvpU7onD4im4/XD57z/rR/yar5DLHdcdY+ZgufFOPIQLRshWRaL6Xv8mzNPpKRTBTF5RDH4UaKbx4yHEXux55gCZ5mIteP2cMu+UTwSDSV3lNtMm6+4Uhrz4ohMhpwU6WZifn7AfEdz+/KIGraEk4DYIKRkOZ8gRoqWxHck1pIMyLTSf/NvuFpo08hWNkx0HIvi1XHHX/zVK/x0T5rfR1XL1lnmk0cGxRQ0ubnAdAlpFbXmVfIiLaK01PKc44NjIbHpZkQ9EkVEKosuFtFo0KsYJAgNs6ayRVLpWoXKC1I1aKlXSdC8IoWFrpS8oFRC1QNWB3JOxDKidcvB7/jZLyy//PXM//DXZ05pgwRE3CHVHmUbhtrwf/rnt/ytP/+abjkScwdtgzYNoQQmIzkvhcIV/+S/fODb3+j5d/7THdIc1yFAQpWSyhO+/vy7/H//Wc///f8debtcIFVLmE5srCaPCm0a6s4hcs90F/HzCMVQhSKLTDU95/Axf/XFhvx/+YTv/tZzGpMwrSOUd68RcW1vLPkZX795j8Vf0HWJvDlixC1KgZBbRBVYKdZnfSURakNSEvN0JKTXmN0ZZETJhRoXimjX/294rMtk2VBMRUiHioXOFTZNBAqPLjdUv3B79DS7R2QGWm3YisoH+siPHgm6IvjFy46fHhWnOhJEJIiB9mKLQq0vPWW9CWmViWXgcqtojKBZIM8ZYSrKrVWyUEa0FCgJUiiW7FF2lVBlp7kbBva6wyiHyoVOG4RuqAJSyWSREEqSamLnOloTcVTmmoisnolujpgYVjOmBoclCAslsZUNNWVKgRwFl7sdqXoaORJUwhuJwHNtFh43imGcWGLlMASc6JhjwilFDms4qm0ssmnINeC6luUUiDmQpxkhLFKsKxMj7SoLk2JV8oqMIFP8jHUKpRJWCYwuFJW5v5+JCfbG0Paa+4PHk/ExkVIkkzifEz56MoK+65E6My8jokjIBSctnWwQXYNfjugS0VJhlEKUvN5eRabUheNpIAqLNZZaKiFFeiV4b7vFjwFURBrHiYa/XHpe/eLIv3y18GET+IMPJO/vBfu9QeN4f5dR6oAqmd95+iE3J8nrh4Xb9Ixfvnrgi+mEFFvQlUYXrLnkr9tHeJeYREC1e6SInOcHdMpsNChrOJw9OjdctD29XKmQW71moLxYZVRaFkr262pLKHINKG3QuVJFWSuSObDrL/BZ4XOg37QIVZnqwka3aGkpIVFJzPNAwrNRO1TRnOeJmApFSoyoaBJCeRbvkcZSCCgrmJZE8HX9OFSIuSCqpm9aetex3W2Ylsj13qGrRdTK1lm21vDyfmCOCUwlF48RGaRkv9nQqpYgJkSBi3aPyjAsMznzzrbaYXSHaRwqC1rXUqXmOI5QGx7vd7Sbp6iu4ckPLJ//2U/41u6MESe07ZjfBjpxxSKfseSJtARMV3j7+nOe6e/TpECjWkqzYesucU4Q/MjD8BmvPvsl733U8GTfc78YTlNLwlLymVZWavVos/7dNbIgc0SpykVriCNY46jScFo8ptsR0IgkSaKShWJoLKJrqaVy/PGf0Xz5c65//w/pvvV9hjvNcvtLpi9OXD+7Zrl5Qzq+RTWCz28nrnYbJj/R9z3+HIiqJZtK6eCmLDzeGC5zx3IG1UpOacFG0MkwGkFMko1tQQm2+5ZW7RiGIy0BWy1NOZPTSAmOaUo41/Fwc+aRe8Ku2xNS4D4fMe0FW6M5jAsZiCVynm+ww0yqhqlJWFtozW94GLgwjrERjMpymj2i7GhEwsgzqBHtD5xuA202VBowikPWvPxMQPOU4SyYTpHdfo/u3+P/8Z9P/Phnd+z6I//Rv/eU9x4/5fMXR371ZcBZxdWuYwkLDzcHFr/w4bMrrp+2mHCmZa0gOdMgWW/jyjqk0IAllmf8xc97XryK5NHTXW55+ebEFCJfvLznk5eSUC3KvkdxDZaM1Z6aArYzqK2jtPDV+YZsT/i4fkzaEijTGaEsVe8J1fD1+AH/9X91w9/8+x9ixbDu3vpLTscn/NN/vOGf/HdX/PKmp+YG1Vqk27Dft5SvBpxxjPOCv5u5ajc8+qDl1dtxZb0rh0qFvJwRyjHXCz599ZT//P/4GVe7kd/5w0seP7cY5xhGzTw94uif8Pqrnk0n2bYZ0Q645g4hCrFEpNCkJBG6oeiOeQRRPVbeU8WJvoE5VmSVVNUjqkIBCL8Ge4Rjyfqdc8HTtS2N8CQyoVTO48i5KMKS6ZoLLmrmuTvxrd0Nzy5uEVny6l6ij1tqlRRxwOiKzxMxSbReK6CrejoRyWRZ0Qa0UJATRXkQBWVXrn+uCSUEUkisNcSyCqQyglwzdRr4Zr8evlJUoKz1pARaF2pdUFIgtMFVwxMhsLXhWGCshqvWcJkDfRbsUuBZbLlizQ883u2w0a+/gzqTsBwaSG2lFImVmraObJbK9UXHr5cXlMaQ6zVeOYxSaJkIObIm4dYb57BMzKWAksgC2QcqqwRICI3KlY3ryHOik3atJvlIb7p3YqMRRUSJhBQF7SybdoMyjnEO3JzPGNeilaIKQ82CnAqu6fB5YR5ntAQtVphKFZVGa5y2zEtAiYbOaVrrMBJyLixLpmssSygEsWBMpGsMhIoviV27Zd84DlPkHCeKEqSo8UJwUwWnSfHpsuXTqbIVmasNPHcPfO8DeNJ5Nnol/H18pfn25YrG/ne/+03OY+A8B25uvyKWB14HxS/SlodksRuH1hIINJuGMt3Q6wJh4lJYrBYw5RWsZCzFrDv4kDyyQtN3hGUi5ImMQGjL5D05Rlpjialg3rExDucTVWmW+UxShVw8S5Q0piPVQAgeYSrOWnKJiDizt4Zk4VxhSQEroYhMLh6/JIzVhGUhzRHX7hDKrEqzjSH6gK5iZRosC1oKNlZDKXgfyVVwOh5WZXqEEBeuLzpqPCNlwzgrfFqxztpZ7u8mWBLWaFICURRCGmLU7NsN22bLMCeOSyHlyqbvEKpBiA33SJ786ENcSsSf/3/QHraXv8M+V8rdZ7iSaGphqR6Y2Zg9XS6cbl5yWAqN/RjTO5SWmHvF5//0FwTxhu/8R/+Iuo2gFQ2Wj0rkalYQOmIOHJcR2bYsKWD2F2y1wClNaiRvFkAYnNsRljU3IXKmGkXRjrM2zCIjm475499ne/VthnbH9ul3ebR9yoH3Cbsj92ognR+YvvoZzaOeMd1iTIvtBH1T8anhVA263SGtJsuW8TSTholxKbhLiz/fkZOBO0XoHTiNbAOBheM50Feo0xGhTmipaEvCnjPLCFRN4cQ4JTbmGtm+ReqG4VCZHt7i7My2ZIxztMuCe/EV+lQJusPqLVpI9L/dluDffhggBlSKEBa2pqcEx8Xlc8Q+0J8zx8my0esB50NlMoZhdvzpX1QGvsGbh4ZD2BJKw5vXE29/uQe148VY+eq/XZsEr19a5npBkZbKTK4Z/A6dDKJ8zfe/OVAPr/nRN4783b9zQdfdkIrh5ZsdcXnOwzGj5CP+r/8k8C/+PJJVYNPtKCIwjS399WOM6ZjSLTEXLjcGJRfSeeAUZhpj2G03IAs2Klz/BH/lmV6DnaBrLEVA02iGlCnCkUXPq+E5t58uIHoGf8HLwxN++sl7/Om/uuRc3iebjJIZ/a4C1VJIyiOUoXRbTrPndhZ0TWa332PtBS+/fItLgq1bK2lBFMZyxY+/8jSu4xeDZftIoIPli18lzpPmOJ+4uz3waH/D977T8wd/qHj8PUOpYkXjlgVhOrJsmVMly0xRGWVW7zqyWXW00q0wFetAa3zqEDzidHNFCBuka9Ai4svCHBZiiNxXw1IkRSu0gg+awvd2Rz5oP6FzL9A6oNUTGtPixLrLLnXCikRShlufQBXIEWRFaoVfRoI1RCJCCFonyQYqC0oJDIroI67r/s2zoNGGzmrIApQh+UhYFjopEQhS9EhZaAVshSbXitWGnDLWtlyphqsC12pH1pYqFranA01Q9HcjV7s9w8ZSiqaRjibGtfonFEUbolUsUhOnhBTr4X3Vb7hqDmw+npijRsiWryaJEYYpnYl1IJdE3+2Z4kwSA7nMKCnBaAw7ZBXYXLi8uiT6ysM0YoTByAtkrdQSkSnwqGsowlDLjDMKqkAqw8Mw8vr+HqEVWQpKmtn3W6zeAOtapprCaVJI3SBLoYqMaSzjeFqhM6cjWVe0hjHld6rtgqirZlvKNR1fygT5RJkmGuWoFdKguJ8Cpq6tnyokQmYuu5aNuSDVzP3suVGOr6aC8pXHVvOv7yb+xvOGbz/SPMaws5pGrNyAjY5susRll/j2B1fIrPHhEilbfnIDv7p7YJhmWhJVzMyq5VAC13LPx9sdD7NntSVVfI7M4fzOeihXI2scKXUg1YIXlhotRRSUWC8iKnliXOiMY3GWJeS10isl1TiU1ITk0UCjJahMtTBGj6HQAiW9ExppSQgJUTONMLRe0AmF3mw51MwUJaMPWKcRORLnAWc1y1xocNQ00u2uCULhQ0FUgTQGRSYeZ1KY8S6wdTBNC1/fVFy7wVrDMk20qmF//ZiQM2OcEEoyDp5FVrSSoFp0FWzbDUkYrLGkODGPZ0IyKCXot5IPfvd3KSfJzVljbr7g976lefFrjx0POCsJ0wMxN/zkj/5vhMOJ6yd/E/fx+6Q48usv/5LDL17x+P0nmMffo+6esJg3yCrQU+RRn3mkNcUbQmh5bDdk1/Hl2xO5htWGqFd5194ahHZsnzTk6YhMGdu1hKZB25791WO8cGRhaX7092hcy3w88FAmkipcffghQXxMSvDi5/8duQuc8i1iq5hFxqvVt/Lsg+ektwcebg5sr95jGde6ttpuiUOhTolNryEv+NSiOkdJEdlIGtczjqvVUo0jm04jMPTdhl61LNuBt796i8YTEoTzyKOPCnevjyuSeK7UWJFK44eZvmSehEpjLC9CxCfF/vopNqrf8DBQdhAcdR4xuiMXQwwjRg4Mb2f8a0Hne2QWJFFppeBwyHx1c8V9nhmXjrG2eHXBmBqcXSBHjGlISI61R2w1+JnO7EBV/PlM3Rqc3qCCZ2bLsAT+4hdn/uWfbfjgo4/5659k/vTHF9zNVxS7JyTDzUFRe4uOnq27Yk4Tk1aMs4IlsHENssKldtRlYvYLgUyrd8w3A1kZzkEgusKuseuNsiqmnJCmx9QW6wNWRarOZNHz+vAeU2348afP+cX9U07LnqMBLbbYcACvmA5QTCSows5uMEoyFwFqQ1aa83ykSwu6Gdk+UixvBghq7cxqkKbjmJ+zNO9x85A43UlUecL49gjScH+XGUfBl2PlFzeJP/7JG/6n//7Co0eRZ493XO02VFXxYkuqAqUnUjMjrMI2G6oudP0OLcLa7xctIT3iOFzz8tByeNgTxpZRPNDaTPCeS23QxSLDjJaarBSPdOL7uzMf7z6lc6+oeUTKFemqcDzq3iOXe/pmwbqZB7/2qatQ5FqoOROLYuM6ihTMKSN0pWlX058SBg043bIkQfUggsQZQw5AkUjp0M6BiyxBsJENrRJr7Q5FIyRPugsOD2eCcMTW0dKypeNJI/nIaHSUxGCpqYVSeKJa5FQ5tIrc7XmgciEjXUlY3RBMw04vPGtncAONhb0eudJvMHzGlF4xiI5b85Sv6jWVgmkyIS1YW8GONB0wF6Qw1CApCKYYSUtepT/3Z5Y5QGMotaBtgWiZs8Hmho6GjVY4u0FZOAvNm1NEaUVrLEUIfPGYHHjUW6JfSFUTa2GeMilVGtfhx2ENJc6JUEDVim0UJS2IktFasxakDdNSaEyHT5rFF7ROLMMrNl2DkVvOU1kbHd0G6Ry5BGqNSAHLdF4riqJg5olNyexL4aHCi2J4rfZ8/kng+cvKk/Yt37xqeX6x5aPLljLfYGRGWYegIgRYOfH73zT83vff4+UbwcvDC5paeHss/PVDw23ak2c4tYaDKFzYjrKMFFHQukWYitSJeTqgNbiNfMefyKSaMVpjlcGPM1shiDnSKQhKkSps2o5UPMEvNEZi0WycI+fK7mLLWCO3IXJ/vueiaSk1k0oG4lof1h0yJgwBWRPn08KUBT4qSs1AoaZ5ZbogGKcDQne0SiBLJCtN3+1JQ17Ry+MdsS5Yk8mlED2MS6TYjllIlHZsrhQiRWY5M6eCcoaaM9tugyyGzm5olKXGhAoFHwda07I1DY0KlGIwpdC/OZD8wkd/++9xpZ/x8K//hDm+QXdXbO1CHBL6TnP/2ZHXv164aD/C9D/kx3/0F/T3Lzk+fAF7TeAxRjuUKuToqdlDDRgkp2GksVfkuuqNi1JMWSOso4jMkgXa7dm0K8nzuo9c7cHWgGy2fHEAxol09zXRXBC7LVluGV3m0u5R4z3Tyzt+9cs/ot8YxKP3mMcX1E7ggydHyRhmtG0gFr64e81cFUoIbD4TjzO6v0Jut/jmgHMtcRI8eeL49ddHVNzR2C0hWbLKlKApVTEezsTThEiB3XvXfPzd3+b1i0+Y5B1DmslaMOU7nrWO3lfOy9oWCqUSqkddGBoE8UXELplOO0S3IbvVRPsbHQaUySR5j1GK5BtKnInzTDIj07D+kEIsaKmxtiXFhF8Ep6Vh0hl/TkzLEY9HnAKX3SVpiqhGrH/Yk8SIhn5J7ITAx4VWCIxz6M4Q0xpOM/aCVI58cdfyyd2Gt2/3vM0bvGpIJOzeIs8DfQsSMMUzp0TXbVlSwQGyBWsd4/EtbTzjlgVXBVqd2D+54O2rE8sciJhVirQdWE6rdlZUqH5eXegbTRYJHzJfHD/gT34i+fzmilFvmMORq+0TTm8HpAQhV4d9rorJNpA9XSkUBdbaFc/aOGo8UpbEdr8lDTDcPeDsDqEdKRceDhB9JFFxTy5o91eY2nD71Q3eJxq74zBVklaEwfBHP7nhG98y/PQXC7vNhrk+4Itn3zX83m9fs+s7LIJkPcgz1rTIXBC1ZQlPuXl1xeF4QSotyvdoU9lZxd5Vts2Gz14mTDbYzUQ2moPxvLePvLe/Z2OO5FSQcoNUkLIF0yA6h/YW4TqMqYh5wuqGKNZdrNGOmgtaN9QKXmgSilZLrN7hZEDKzFwDOS8gBX2/peRCLgJlGsAghEa2HeMx8p7bcqUanttHbK3BFcEWx+A23MvK2ax+BitbNkg6FoxODEGT909ZTiekTjypBXsaWHTPC1F50vY80Q2olugadrs7vv/oU5x6gZVHYr6jlCNSztTqGdMWtX3GzecbHu5OVFGoZZWwOC2ZUyTlTM6Z7BOLh0zFdA1aOsZYSUJgYmFjDXkZUHq3aoxNy2W74+PLi5UBogWfnQNWjOhW4MvCmD1SgFOKuqw2v1wUZz9RlXwnxjEItVZ4bdfirEPU1b+wsZLKhFECJSuzz5ymSm0lUgqWXNENKOW5bMGgucuO0hqiMutLUpHoxtNoR/QzExLdWlCR4eEN77cdjWh4EzO16/Ha8csJXnjDT29HduaOf/Dxlj/4rqKXmTqdsKrirEGrjqf7HbN4w3vfuuVvFU+MgZob3v9J5E++iByU4fOx4NsdD6XSdy3RV2SjqTkicsLUTJgHrFqDnUJIpNLIIhCi4pSj06CNYxaaZclY5QgpIaRg4zZoNI1RdNZSk8EkoCwcz2dqIxjkWm1tN+u6M+dIUgnJSv48zyOL1EhnKcHTbh1RzPStJoSZnCtSVxIzND2m0XAuHO4Xsi/kEMh55JGRqKLZJoWIa+4lKYlPHhtX2FcikoJniYmt7ZEClCogI5u+YcMFO9kgpWAvMuploD5ILrPgKGbaET48Z8zpDflXf0L55h/w/Pu/g1yeMp7vcYeCCgt3b0YmDVn/kLO64NXtDdfDn6MePkV2V+z/xn+Aad7n2Y/2uO6OWOxKBpWASDR9R0pQmo4FwRwLXhYKCpFh4xwpena6YtINcplZOFJtRqYTl/ES//U946uFj775NzlMEXmlsXUhffU5t19+yvvf/Q72m5LTcOD2xac8QjHKDWqciRQWW/A+k5vKw9s7lNzy3uY91OvCZrmkhEJ1kZISteypeaKUwKaxHG8X7HVPFJ6sBRHBA5JQBUZA1YUvv77l/PZfYIaXXEiYa0QUhdKKfHbUsUOR8U3mQTywaMn2kWaaH8BO9GGHUIK8t5xsZjv/hoeB3BwQJbPZ7vAvj1Rmorzm/qYjlg5zsWU4efz8hlbvyUqDb7j7zMNuJJ0TRIXNlVZ3KGeJRXMmMGdDIxvC6YEmRhqg3zsGAk3tWG5uEToxaEjGMs2Km3NPrS1vjp7FHyGdAUmYIvtoUU7RPX+yJvyFxRfLpdU01XCqB+bzQruRNDmhhAEf0XgeNZLZgoqFrTEkqaiPC0FGRBBYY8kUXN+itGA+B2zf8eNPEz97uaOyhuE2QiFEJMUJVQtWWjZty8M0U3RDMXrtadeJKgW1rlUpmwK2TsQwcvF0RxUOUVrmuB4SKcNyVOAKj7sNIU3M4YF2I2jdDiG31NuZUCrguPOP6NKOMp64DRbXPsGZyN1w5Gd/MbDVJ6zx7C4sZit5+vQ5jYK3bys/+Qu4fX1g10coHmev6WyD1YJNb3Gy8NU//SXX10+5erbhw4+u+f5V4vrpPY0KEA3S7ldDWwoIc02wPYteUY0pFrRUbJot41KIUqHlO5RjKmzdlq21DLXQNwpdI52VaBZCXci14JqekAOIgjZr9dBYTQp11fc1DbrfseOSb+uOy+0KOdGiInziopc8co4vneB+WJ3zV73hIi/IqhivLnjlJX3tKfdv2LnCvkY+1AJx2fFx49guJ8YM+qJwsTlz2fwMqz5F6jNVJ2ZfqAL6VtBFwZgtxUuksGi13hxrGtGxUkIhZ0moUBDkCqYzhAzzNGGroVFw2Ro+6K9IvuEYIZkWVzc42bNtBLLOoATVR1rlqFqQFk+JiVICzmlqDsQQwLTUGpjmVT1c4kx9l1uxVdKIBkFDCQNdb9AojBXEUjiFjG56irIMy4y0Gq0FuoFdK94hyS25WOYMpcS1+14Humat4I7zkUVohC4s+cAmzuy3z5mPd8zpHuM2KDqUtpyQvFkCw6dHXs6WXg082lg6LXh6vaM3hkZ5ut4jVEDWijYKQ+If/UHD732/5+7c8l/8xZE/PyYmsYbNigLvZ5qiMNJh5B6DQhCQUvLBZs8UQJNpZKGTipTOJCm4GxJBKqRULIun1YrgB7auo1bBkCZyGmnzag+8MIYsoMhELJmH5UjVFac69l3LfDuSaen6PeN0T8wLuvsf5VeaktKqgK8FZxtkUQhtuT2PnA6WMCoapdcPSNW4Aq1U9EXi04BzPcRKLauLpBeVYy6rKh6Jj55t25EyaNlwf5qITQdqS6v3PCuC8uJnDF/c8IO/+4zXDy+x0ws2w2uezl9z98/+GvM/gRdXV/zev/M/Q5pL7sp/y2L+kkE/0DbPuHjvh+Rp4PrwM3bjF8wxEJJkvPM8+/s95tJxCIVd/xhBQtsJYSVZeGrUkA1+WKu5TgjGFNi1midtQqR7Lm1C2zPWZKSCgmLMhU7DJ3/0CdPNiX7zDLfPzDc/hfnHGPU1bWi5/atAFJqLbosJE9MpY1XDeDIkGWk3hlO5ZYkJ3V5QQmUuYQX99Bu8nRiGW3LK1Lrw5LHFqLUIZ0xleynRFO5KIlGYd4a2OOrxRK0FWyzGBObplkbvqFGgOskyZz7/4g6Lo33maNPACEQVCbszGoX6eSbcHyjumvFwwotAl3e/2WEg+kQj9kRfsM0JufHcHQ3zceWSD+ZAeu8ee70wqolUL/GvKsfDhPQDvenY6dXVPahI82TP4S4xzYFGdpR6R3t5wOWVQZ/LHikNXatpqGjb8PqcGGIiLYmvPjvQcGLXbig107otL94O6ArbWCDPlPaE7fdMLx6o9oJqI4M/MYWJvtmifUCZnpQekMEjG8f9m1vStN7WTecQViFHQeQEfUT3it1uj7UdcorEKdH2W+4eTsRFYUyFFNFJgvT0/SrcUTmTxcy5nJGiIC2YmkBkNt1qsNNpxsrI3jrmJKDVdE87jm9neuVp2kCpGhkavJlJy8TFxSNqIzncnsE0K+Rm22IXv75aJMnNaFH6feZh4lFuef965tnzPe10wiaHrp77mxPjq8IXn1X2jz9ini3VOCKv8WpV1arzgFEDpQbu32TOv1Lo8yVv7gJvfnbPi/7E1VXiD/59ycXf2pFcQJR1l1ekY/Qb5tLjmh6jKq3MlHrGh4ptGh58ptb11aZJcNlCq7dkNGaZ2TWe1imWINcAYpVUWRFqDRdSQKAoJYGzGKlWHW+rMElyoRQNqzWu1kDuBdsMqm+47xKybNlYA3ZtxDStpG8Fy8uW27Ngv9+j68AmQLs8sH2Y2T83FJERRHoT0WZAiFukHhC2kGoFA1oAqkIWZNFj6WlqJcQznesRMiOFZYyeTjdYaRjyQokFkSW6GHbdhq12mFq4NoZHsUGXDc5YBtFiVY/SDcpa/HCDSInLpufoPUstzNmwaTbYrHHVI3JenRgpYmUFa1BKY1k/bJ1ucLLDipZYInMsjKeJq90FNXpKDcQlEVVCdZmYV6QxCbSWpFSoOVFqRRqY/RFnKq3TbLYGqReG8UwwERD47JEucq/zyuW4UKtcrC6ovHAOlVNYXy5eLJLj6xU1rTjgtMbZiWfbDd/sFH//d2DfL5ic6GxHLZIqZp7tC3s78p/89gXfvev49Hbgr96+RRnFdU7szOq8r+oC4jVVBeY48Ng95hQzra3odMc8n5lqZhSGQ4QiMy4v7IxEMuM6IM/MYaDdtiRmjnFAO4MoHr/MZOkpqiKaVdyki+RS7jE1o5VhmSesNvSbhuO0MEePbRzWdhhhETVgTc88J05DRsoeZywXjw3jOIJtkdnhXIfKK5EQmchF0EloI7yvLHUZGUNlVpLOdatZ09f17JMWazsSjilZgug49C2Pvv89LstH/OznL/hRfMHu7me09sQ+HkiTYvjsr9k+/RFxOSEeG9RV4Op0YvP4GfNSEeNbLkxgP73EDIXjwVFj4dVff4p4lunbj9he77h78CyL4xxhKZkUockVVxPXu5b2LnOdGu6XgW99q+G9a01JCsmALJFSEku2jKXF5y30Oz76j/8+4/1C/vgbDLs9j/kWpz9+wJoOJsUQJO3V+yxbSSMUIR7ZIFj0wmmMVKeg6ViGgZlM17eYrcZuJcfXt4i8arc3qWDPD+i0x9aWtuu4rweax56t3MCrTBWXxGlmt1VIYUk1ku/u1n9HoxHRIyT0O8c4C7bvb0h+JJxP4BONLIQxoWVDTSsBddeD6RXHRRMOir5vf7PDQHcZMXLk8eYxX98NLLaSdhPKBOpyRjce0RQCCa09xBFhAy5couIzYvK0RrHbXDNUz+3xQHCGkhTFLDzaVna9Iw0LaTxD1pjyiCQil7sLlmVCOk9jN5xGQyoV11f2e0seZtQo2KWWmBb8dCSPCsW6y9xfXFP3e5LPoCdiOJDSDSYmLvsrFpHJ4YBaMtUnOmOJauR4f0sKlZp2uKXw7PlH6L5jSmvFKPqB+XDGtw2LVTRBU8NMqYJKYY4HaFouLz6gqQ513cKrF5y+OuCkxUpLyQsNGec2uKyRk2AZJee50D/bwUaw6x/YXzt0gfuHmcN8whrFo2DplsL9qwE9augssUiMlrTd2ut1RaCmitSguGA4SV7dRj5qrvnGj37A4f4zpILTixtSCPjz6m2Xfc94/5pOzZj+PdruMToOlHrDftfyqz8/E4873nvyMdOSqTVisicePX/6T+45jTPf+hsO5yw+G25uCtO5ZVcvqKahqJGuSHK5wGsFRSLa9TLvjKXLYrWsVYvRjmfXDaSJJUWygpotj6TiQjsmf0ZbQZgzRQmgklOhGgiqw7lruiny2N2jZcuRSx5f7NGjQOaGa7/A55/ykD6mvPcBqd8yXW3ZyDue1sT48sCe58y0LOYJwt6hXvycEirtb/0uwlfs3Zn45y949vsBs8soZ0llodR1ELBqBd1JYfHLlp2+4nm/oRiFEBuUuuJuOLG3W86lMKVCh6HpG7QypKxphaMRq1zLRs3edvT9Flkv6NQFWnco4TmPkU7vcSrwXjlzvh94KIoaViYHIdDahr3RzNHjlKLfdBx8QiiHkw2VhcftFceHTBGKWANFN0wpMt4sNL1E2Y6qNeREWgIigxYCaw2iZBapKLGiu0rgRFEzWltE9fhlIUo4TSOlKpRZ6Xupek4USrkjqY4YPSV7rLYsY0VicVgEDp8yYqMZg8C5luM0clMDX9zCkcyPPkzsnaSThV5rentBUytWZb7z2PPhteHf+/iST+8dxSg2ZcDVhc9eB378tnKTBFLvqckRaNFNolcaUSeqWoN1IUV0J+m1xYWZfaeZhploDT5VqoFzGkAkdGM5h4FQA1N9hzFXFS0lqoCqkU5AKIVcZwoLRkh65RjwSCGZzwtON2hh2LaG07mwRE2pGqsc/W5LGs5EAqEqjGsZfMCSMGKFtJnOsouGOC5cVI20DYecSa4F1naJrgYTNU13RddfQdkxnSoXrsUsjlwdzbXi6ZMLwk+/JsfA485wEXYUrXhzjFzLhm3+kof4hiv3NY2ZePlw4ni6wV3uEeLIM90iywbKQD4OyLRBD4WUPPef3DMcb1HPHuOdJheFkVsUBj1MNGLEvHnDiz/6Maa1/Na/+/eIduF+eA0mrzmSXFAlsNv0HJdLHuRj5Lc/QHxP4ZuemjuOXNP+4X9G+sm/5Bs/uOJzf+LlJ3dEFM8e7/nOP/weX04HXn6WSEcYw5E0B1TWNCpjiyIvE9pu6ecWu70kzA/s3Jb9UwhfV1SthLSw3exppMAW2Ic1P1NyYrtk8hwRtbIUQzrBZXvNeX5NqIVu+4iYHZ+/fIDbEe0tplxSp8gT09IkSz18iR0cIiaKr8iuIp0iUn+zw4DpZrCJbA2iKaiTgIeZy4uGEiLBNZyOoGOHSnlVTdYjuysJYUPigs1OcxfOKL3ubJ0QmF2DlJ69dMgT5BvJdfuY25PHPO6JvQDbEx8KUx1IZkF+c8d8Wbm7E8TRIBMgPIpIe9HyMI5I48iD4utfveC9H/yI0BeKruR5pskKoyKtzuj5RJcVbvsYv5zQvUJqSXf1HrMfOJyObK6u+bLesPiJVu4IS6A6S1UNSinaviMJybVh5SzIFutaatei7TU5NAwpIsLC82+9x4UKjIeZ7fYJy6sTnALNvsWUyvbiElEzPt/z6FnP42cNZqikr4/E3BBnhSuO77z/HlvrePPrF1zFnsuu53aecI3DNC1aWIRO5PMDpMwpHnH7C1Tb8bDAX/3iHoB2b3i4vYGQUZ1gc6W5e/MWf3OmlZFxSsSbhbw54+TIrjGchsL9zRajr9FpQbQj9egp1lBqwvgLvvjzBd0ZphB4cdvw+qZyOg0k91OefvCYrV4IYsC5R1zvH9OIxHHJaNdRSkKpGdH2aGGxS6CpLbEUrNZr1Q1BZxSRSCsrOSWqq1ij6bRFUBmFxIctzdxx2QTUrrJxG4J4tqavk8ZgqS/+e/Q/+3PG9gc8+kf/EPP8KWyeYEoDw5kXn/0Zj75jmO5vePyt3+PmdiT8608obstj+bfZ+DN/+s//BbdfHvn4/e/SfLgliwPGqHUHX1dfhZQwRkM8NVzrZ4R4x93dLbOMmI2GuiHmzPF4JlPZuZ4NhirNmsQ3BofEiRU5a9yGUgy7vsfQ4OeJfZPZyoULd0KXe2gD5ZHiRXzE3dJirSTIiV0rsfUB3VSq05zjjMwT8xzQxnDRbNjoDQ9pQluDUhIfB5RoMF3HWAcsmpQKre1AVIqMtNoiaVjywmQdRRqmcAaZkDJyfLjhcrchloLb9CutMi2UoHDK4RxU4ZnlmdN44GJ/AUtgzDMP3tM1F5zmA9pITGhQNJRcmbKnFoHPkocq+dPPj7w4B5yQXLQd+52kE7DVlp1uuKCltw0blfl+L9lvBY1cMOrAbz1T/N685394ceSwwCdvFEebUDlwuevwN465Gm4PE3m/o9QFXaH4CeUaus5yFIK7hxO2sVAkpawNFpQkBZBKYduGEBcohUYJmrzAcoPKCdPs0NWAVKv8KCjIgcYZ/Hlks9thSmWejyAdWkJjNJREebf6kNR3HpJKkQrtHMY4CpJpmLBGE0VG5oJVln13TcaigmCjHNvtnoojxw2VDRsjMFqRZGSY7pC//HOCcDAPuCeP6OWXXKoTD6ND1ULzdMuHv/0U9xVMv94wz1fY8hir9pznA3U5UneXyMYwz7/CSIsOhvuvBna/Y+iud3zwnfe5Oy20oydOR8wikJuMOwa+/qMf8/DVz2iC5xv/8e9Tu45leAk54uOItQojFdtGE3XFyZZhkOi+X2ulpRBPL1n8LccvH3gmB8T+Q+zthuhmFma+GhKf/OWvuXy8RbU7+n6POQzMX3zFvn/EgOdwPyM7zW6/5XC/IMUlWSaOuXDZXlI3C7kouC+UdGZ4NXC9b6m2IkKmk5muFJZJ4o8RpXpKlQSdKU2ApLBaEsdErZJ6uGQ2hixapJ9pZYt4HRB3ipouiSUgUoPpM1rMpMn9ZocB/+sd/fsXzJOgPhgsFcfEJjTo9hlvP3vN9WaPJyPThNs2hFGhlgbZSLzwfP0wM5YRv9E0Fz1lDkgRUPJdFUZJdN9Rm0RnN8z9hsPdDaOcCNkwz5rgPVkEPh0SG9Ujj57LJiEZqNKQywm5qUiVwCa6R3uCDUhZqelAHt+wVVt6tUPkM1FI7H5LzgWlLKJZePy8Z14yy7QghcGfM+998BGH5USWI7JZqKIhCY2yPcI4dralzgrp3rLdGO4H+PLFjFMT232Dah2tc5QcaDeF7f7/z9p/LN22nemZ2DPc9Mv+bvvjcQAkEsxkZpKlZJGSQqViKKoUalRTDUWoo3tQW5fDjtioUBRJkSoWyUwigYQ9wDF7n21/v+z0czg1fl5ANXALM9aa8x3feL/nOUXpM5LDhsQNzBaebt/j7UR37FC6oPOWKAuOlzuyzlL3lmEwBCkpV4q6vgY883WCUinifqIn0hORi4rx/p5UaURveLyqSGaCaRhohoAf57y/7mDbk7SaqHI0gu1ec9x6jHyYgNwdJj49F1SlIJWnjEi+e7lh9IZ0nlCmM4buwXOQFJJpb2mHmgMB4z7i/mhpRs22PTA5yGVK975mczzSdCMxdVQLy7AL9PuGbC559NmSz36wZDlTSB/YDg3lbE2UFf20IxOWoBxDULggqXyJbCMyJA9dD58QlOPmasvq7gOPVjWPfvwIc/4CSWTevOfd11vOP/9n5OOO7g+/Yby54t4n3NmEj92f8tF/84zGwdhOdB92vHW/ZRoXZJ+fUHSGY71hEjk3v68pM5C/eUtoBL/5jxv+jz86QZ2+RwhPEA8j8gdMvEJwTnssGEKk1IpJFYRkoMczSEEzdBids8pzLtIZp6bgOAUcAydyzSxElmlBZgTOP3R4ZrMExQE937JIW/LkgGRDngcic6L8mLdvCspsjgYS3aKFRZiEMNVo71nIyHqR0cYF7fhg8ptGgdQZIUoEKcYOnJ3mHPojBME4BnJZoW1EGYUxCa7t6WPP6C1CGpq6pVcBJSOpFuRVRMoWKUG6lDC4h91+O4Bs0alCFJKur5lG94AWzzRX9xvaVNL6nqLMsAJ27T3GZ2iVIQeFTgqi8XR2IM8c27ojN5q9C6Q2QJgxxgTrPLHfs6bniYInVcZHj1N+9DhHmzsydeDT+Y4XTyQ3+4aPVobrPZRGI9XXDPrAq+0jXoYUREYlJFX0KDPDiEjA07YjNkhykzF1jkQnDxAjrRmaCZU+SLOQGiE9ZZFQTJ480cQqYfSCxBRIk3CYLCEqgpAE7/6LijowBY8xkiADwQ4QOgq9QomCSRhSA4Nt0RoGb8FpxmNH5IEzEqWncwoTMwpSAjk9miAC8/kS5T1ptPTDiHaaRTQUQpDPT1hXAcZvSLXlXQOyH/mvfzInra9Jm3tUf0IsZ6TPP6H75p7F8yf84Wd77FShkpxcaM7WT1l99JT9zT3t9w1S1Ogi5e7DPU+s5fST50ip6DffEm8nht01i+oRH759zd3ffU1+fcUP/+wHLB9/SfzsjJ3rSVSCMRkiPNBujbEkiUAI6HaR0AawjlBKzCwhOVsRrz7gDv+O1XKGpqDZHpiffsyz51/gzcCh2XH59c94Oq9AKOz8Gc1Tyf52g1A5ubmgblveDR4tDKt8RmlO2d1fUosEqUacjagyIUkdenIMhzvGLiPPTxF1w9gHXP1A4KUQqMrQ2Y7F2QmDbZGhYTZFsgit1CSnF/hMIxYd6yzl7W+/ZoVmHDxpmiCE5rxKcXeA7f64YUDkgjQFLSPNJOnGmkwHYua5aUd6tUL5FKkiOp3jtkeyrKI+HJmGLdNM0v6XUW4yrzh7vGR/f4+9b5hdzNi+vMPVkkj5IBBK9+zGPff304NRS2UPo9d+g0kckzMcy4SsVDSyY15l9PseUxQkOsHtBrowsX5yyhAlBEtsRvJeEwdP/vQC8hNkNTH1E/2hIy81jy9OORy2+OBxUdDbh5PvYvaY0FiQirQoGaNn2x1RquRoAz4Gqpg8uKtTS3MLXaMJeYCuRaqMsREkSY4+lrgJxjn0g8LYiSfLNZyuwba8/cORwY5UlWK0A1ZqqiphlgZ2m4lRWmbLFNmluO6hiyCSwCxN8Z3neDggpSZVOZlyeD0wNbfkq5ysgCRNyORIe9xQPZoj0gw1S6iVp+sVrEqcO6D6BFWt0cU5+7blfh/47ddbpnFJIhNyqbjb78jFHC8HopdolTGqI2MDX/3nO7qsxCQZw5By9iQnTHtiUWPjiFc5PqQMXYIdFUmWEsaB/TeBX/7+Hc9+EDk5n7C2Jl9/ClHQ3F4iQsq4t4xDi9tahss9ze7BDJjKQCGBoHBHQe4PHGeXvP/9Df+3/9f/HaH2fP///bdsv1eYv/i/MFx/hZiOlE8KpmvD5bZl+u0bFi9+w6gSrr/dsFi1jO6aR0//Ke/r33H/7c9QoWJ7MGy2HpNvUEXP08Uc9fhjrl6957PHOUK2D3hq9RAEhuEFu+2P2dZLBuEo85SJGRs3UtsWqzUmNagkkopILjQzZbBOcRJOeRxOMd2eVUjJVYFMJRcLwcViR6JvyM0BSYeLLZGeJFH4eErYzMn1mhkjRRxJjeCkFJTGYcJImexYVCMDS/7+2lC7wCQUbV9j4wMQSbrART4wF4FypbFbTZAJeIc0kXGcUInBWk/vB4SW9HJgsBbrPbYbiEJxVqYkOGKItO0Rh0SgiW4C/V9YEE4ihEYwkiQPGHESzzD2aGMYRCAAExOVUYzjxLyY4+gfTsPjER8CsghM3mOTlG3ncW5gJGXnHwBCJ1pzoxJe2Yxffn2PNTn/8EWOkRu0HlGh5qwa+CefCRApWIe1O2rn+eFhxuO3P+LqmHJVbxlsgzMVzdjgXIFtIUtzfGeR9gHmM/Y1k/b44IkEujaQpwYjJcZLyiSjc4raawgaLXKaesQai1PDgy4+S9FIZquC3fU9ISqil+RJjgwRLSa6yeIx9OOEkJ5MpygRkTLBZIrJG6RQJEIiZfGAdweGpqdPMoJXKOYsM4Npthx/8/f8+OP/mjzM+PDhltVHhnFXM/YtC50zFKfc9FtCmqHFCUViH+Rkh4bx7XsWicKsXjBmd9SDRFQJvhVstefn795Qb1t8dcHF4ikyFxyHllNdsfn2Dcfesv3uO3xn6W7vuKt/S19vma1mzHjEZhJkhaPY9yQnPcmpJitmHA8PGGyTSLTMGfyS/V7jtUFMllI5iiRDm8cMJwXJn91Q3/+Wb/7mb2muDDyuCc969DSiuyOfPk8J+z2NM6h0QSZOWT4uuL/ekUjHkGrMcknY3nO4e8Xq5BHzQuFjx1CMKKGYjorxdqRalXTtxNQYumRPhcfohOWLC2x7TygT+knRbg8UvUdHRxIdydyRpgkxF8yen3O3GflwfU+v9hjx0E8SqaRXD96ReOgoqjm2tn/cMJBRsh4/Rs5nfLh7xTTkNDoSxxl7AeZZgggWERw2C8hTg08n5HhEJYrxMCJ1STpFZNNTf7+lzNZgFwx/uIXLgNeGY9IQfMCqwJgGymSJ0hIfPSFpeHK+otSKu6bDlpJSSM7jmoSJNNcMSqHzhPWjU65ud8jeITpL00yI+kHbO1tWDOWeadII+2By08sl1rc0C0nbZCSTwtVHnj3+kuLTMz6EiWEm8OOEqz0mT7GDJEkNYX/EzHLSxLKoKm67G3yXkklNno4oFN0x0FoHRc+niwoxNHRyT2McWZHThYzQ9PjuYb1xd31Eari5afAHGHtPDA4hPN3Y8M1v3vMoXVNoia0K6ron1wUmHSnERFodOc0F1BPd0TG1ktSWpNkSuTL48J6sM5zM59zEnrevDkwuxatIXjnKaBAhcnSK//jzkSlIhrZD6VNkUkCuuRv2pKPjrKrY1z0hpBRTz1I6tFUUdUp6nIjFAdtKTk8Fb95ekpk5eqWRxnNSZAhnud1sGacFUs+YuoAJgq9/teXLPzlSzm748Ic/sBhGOH7L7U5ht3NmizWyPSD7DtspVJqyXFQcbzcoG8lSyOcBESRhiGzeO1Qq2bw8MgyKP7z9DxTf/lvMW0Hx5ITiWGCd57A/8odvfs+brkQPM56cPkOnL2nH77j/+0ueiIianbPrIj/7w2/58+WOs2yJqM748P2O2fOUVC3x6h4ZBSIxdONzvvrlD3h19ZhJLTFGU5UpPjhmU8CSYIWiWBhGZ8mmlkQIFJEqKxmHCnOEJ6qk9ApXj5gqpch3ZNk7SrVDqYGAQEWPkPHhWqKVuKh5cZHx0WnLutxQ5XfMyh4tdhi5xegdQmpebzO+O1yQDJ60CpRVwyp2JAzkypFoy0TPIVYPSFw1QwnPYjHj2E0Pq8VFgZGRKHg4oeBYq5RhnAjjxCIvSVXgbr8hK1IQKe1gUfMMKQOD77FjfHATJIKpbxnjhMQRwsA4DhgzY7IOkSpC6JkVJVH26GL2AF8pBEUaKUoFKn1AXQdFkAm2exCfKZkwCs3d5OmwSLvgX/z9AaM1P3k6IeUeZIfSHWVmIUCYOlZzzyOT8+J0ZB5gd+Mpz854WUveDQlv9pbvDw1nWcExNITo0WlOsANRS45Ng0wVSkkkAqUFUnjcJHCzlMudZewdJsuIY4MwEs9EWkaqfI71IKyk7x3WCbSQmKxERIGRmjIp6EyDJSIF5Foho4UYqbsWQorRKVWSkflIkWdMTNi+w6kUG1Kky8CvcDJB7wULt6B+/5Jy0bG/abh98x3P5jV5POI7y+mTH2AvXhD+LIdmjSzOSb7vefvVL1mWkeb9NYt5oJoNXI2GcSlYfvkpu3FD825HUZTM/+wpXkzMhUUqwy9+c8/b7S35dofbdpQXgtMM5mczXrw44dGPfsz97+8J45756Ug0t1RSAS3kAU/Brs/xY4nwc26PFTEtIUmQKnB48wtudzUf/cP/lvLip0yz/xPj9ITFiwM2vmNXvyT9j1tUMPjhnvlyxjRT5M+fkC5nqNOSY7tHysjm+IZstUKVnmAfCsZKB6plRZKCP3jG/YAqEtQ0cv22BxZI7RHa0uueVGrssSH0jql3RA9+sBipySeBrEvy7JQ+DiR5wuHma8RNwguZY5sWO050UySdBOSSQIfvHKEqyKo/8jWBOj7wAu6v35KqnM5EVClpg2IqFGNiyWKgQuP8hM5P6JMaY3KMTLnIE0wsmGYTx5uGZK+oosEdO+oPD8Ui7SUXH805bhtsmNFEh04GslWBVgVDkJw9XzHWG1RZsXWSEHsIhmZ3QEmFcZ4yS/j4009Zro68/3DDLKacL9ZsZUdSPGh7m2GPbwWr+RkqFEyTpG9G9ALyIjIcWkgd6sTRdfeIqIh1x76uH+h3w4ILrTkxFpM4lmeBXDnMWvPyVwPz5BFlNUcnPbv9EWNyxuhomont6Hk61+ybHfsuMPaa7qsrZoXGiJIwTRy7lDdfvyFVJckA+75HRFDzHHEcoMm52fXoKDjWe0KRsO0PRNeT5innWeTZU0Wzjbidw3YJu+sdj14s8fUek7fEyfD+l3u2PtJZgzYFQkRCbwlT5Hjo6dIEFzUugpOaqRse3AFC0o+R1mpsOyB9jhJLjq3j7CxBNx3LcsHUb7FRIKTgpHK8tJY3Xw08fjpnHCy3HPni8WM2WuOPgrQ06HQgnxn8tEQmkrTfsf3wB1bnKXmcODt5xJtNym5f84M/09xfKYTW+KzGPbYE4+m+jxTrU1588UP+cPU9u6nj11/9S54VEZPc0FcJE7/Ev79l6WeQnpFmBWVX0DcJx0PPPM+5vLzBqI5yljONKfb9RMg9afnQy9gfBsyzU+yjjOzxGf/8v/9TmH9A+NeoqJCJYfLP+PbXn/Lm9Z/RJc/wRoDwKKmo8oInaC7EKTcu0oqI9S2FyDjN4aM1TMMKNjlzZcjnEwuTITvHvqkxmSCpjpxnA84ruinH2QkNWG/o+wQtJRfZnkX6ksL8FpPeIfSREGuEmB6EWnpNVWim4cHL68U9WXrAhBbjW4x6wDgHk+CmiFYaqQS5TKnSiuOxZppAJposTXFTR0IK2jLLK4weaX2NIkELSZ5ZOjuis4Q8T6n7HmUEZWIQwTFYT9+NlKYiMxldNzBLFUlRPryPhEAQkd6ihUMVCd2ww3nNPEkeJneNQyg4DiPNcHg4jGQn6G4idQ4RH1bOemXJ8jm3w5yfval5unAszRGh2odJg3PE4NAy4JBk+gFC9ekqA52RxB2PswNv9Kc0v9Ts5wtQPWdmou72jGqHTBXbOFHoFCcmgnQPp8Xp4ZnKqNh0jlE6iiphDC0higcFc32gFCXeRbyV2MHjrYOYI5THjhNCygc0ubfUscc6i5EPtEQfPSrJEZNgPq/wTqKjxET10LcYW0RicCKSIliUK6Q26HJOfpazGivab7/n7fbXJMWc4AcOVxtOKLDTHuG2HKVk+OivMSpyOP4rHtdv+O31Jd//hw3PjMK3HZVOcXWCHCeOzcj89IxpFykqw9X1FeXkKYzkQMv95sg4ZWgbmS9mfPEnP8KmHXQT558uUZ/lmCcTp8LzZG0hN0RvsbbFxwERUnZ1BuMZ6+IJzlrKRcYoE0SUPJ8/5+rue9788m+oHh+ZVQnKf8tHjwwfrZ/S7xc0r7a47TW7YUfbp6TdkvNzz7G54f4oUcbTtO/QNFTqjObQkB4ijNkD0jkBO0zoXeSUHKUkPF9zIzfcvTtwuj4n6gk7CmR84NDkszPCcEQMFiaNGwbEEGgPE/rQk59Ekv5ArnJa8gcKZddDaQi2IRWOXAqK1RO6mCPmJ7id/+OGAfv5hp1s6C1ol5NPHcuLBU0UiH4gUTnGB9z9PUlI0Msc25RE/5T9YWCsJ7BHnAj4ZEaZzNG6YQzvCElCvppR6oTEOxbzU5Jqxqu715hEc1Kc4F3CfdtxrudUj+Z89/odh37AeU+tIjN5wlBvkCpldxP4xf07ssrQOUHiLIh78rWmUwMxeKSVmEzRt3t8vae3Hhctql9DniCWmmJewaqj73pMMCxVy6ISpEmBETArPWa3x1iNaQbyVUYdHVfX8GhREgX4GB6sglqzXK1oDnvc1DLVd6yXFXKEJFsTFRyHB/WsyTIWs5JPPxf022uc1DQu0neWxJWU8zleew67I2oyVAvF2eNzpiip7zuKQhGZuHx/ZJZKZitB21yzOi8wxfckJuN8ecYfXr2maSJOlyR4iiJlaEaE9QxxQjmBnBygKbICkWSMU08cG9I0ebh+URClfkAmH1uq1LD+yDDMElrtiL2luXNkC02pRy4+vmB3mBiC5/SjRzQ43tqOxQ/WlLcpP/3Bl/zqF99ia0HdGg7v4MXpGU6/g3xNPV8ySlj8yUQ1V9Tnlg92IlsrnPHc6TsW+WP+8h/8NScZmFXKv9vtEdFw4VvC+wY3aXSVU0jJ9W2Kkmd4vSCkI7kCnxna3jI/OXIyN4h+yf2loL7zzEWGUwNNf8PKX5A8f8Sf/l//B7q3b7l4VGFWjrrZI6dzpN7gRsWxvsDZHyOyj4gyxZiHcW5ZZEgToJAskoL+w4alyDAoyvSEsohcZC3x8J77es66+inRSw5jz6P9ET4cmH+Wc/4CUuNwR4N2J6hgkaojxjnEJ2QqQYdrtLhHmwMy6bDxgA0PXgUfBZO19KOn70dCiPTtloyW6Gu0ETSDJUZBiBJHAtJjMkVpKuwUUWT4YUQHQ56UeBlJ4kNT3rYdAU9hDN2hxxaGAbg9NCTWonXywNo3BZGIZ6LuB6KDcfLYwwEXRjCC6C06TXAhMvU9y2zOvMpoxwYpBM5ZsmWJQNAME845RJ6hDFjXQ9hg0iVZmqC1oG07ZukakoCP0PJw/dB2E1mSkBmBkCNSQ7APp9YoZrhhSSZn6EwT2yvO1oL/9HbD+/oJtckwiWSUE7PyhLnO8fmOw65Fu4RZVT38RoRCK01mMpyMuAmKJCXXmtDvCUhwD/bDcbBs6p4YM6LUFApmSYIRgWVZMglBMzru7lqkmkhS0MFT5TlYQZGVJLJgHCWjG5jsgNIzGicQs4o0K1ihcEPK+eKEPM1JdY61Nbfvf8NSRlJhUdt7hCxYXjynvbpmlZXsL1+yj5Kf/+GKLz5JKVYLktqgXm1RqwuaIEgOaxhyVHdAtT1i7lipM3bJkeP2hua+oRURkwVMCovTRyTJjCybODlbMhaG1Q++QGtBL0dIDLOnOXk8EuUHgog0dY+QkX5ybDrPvk2YyZLUJqyNYsJz1IEJRfXoOZ/YiM0MVglG+7eszc9Rl3ckdxc8zf4S//yM2lxzsy9pG8382LLeCn6z7dgNFck6Q4mcLKTIWrJITjEqp9eSDslw55iLikVfMI170syjYuTpTx7Rui3DsWWez3BtDxcPmyfD4DCZJIxHKp0xLwV139IOEzrN+fTHM9rXrxlcjlw9YrreUaQ5eTzB+S00bzDjhAzn6JOCsM7x93d/3DAgvETrnERCU7dE6zArHmAph47MQO4UKiYIPCpp6cf2wVNvJ2bZHJIRPys4OAGzJdbcE9WMEAqOStKNA4+LlOViQeMaSjnjbPUMay37DzVZsuLupefaHdFTweePcjq2dP0OnToSaanyNZ2IWKewQZOnBX5scbEnLeYIFMNhZLAPqlYveoq8Istn9FaxuTtykIrnT06w7RFpDXoQCKE4X2QQA4fdyLY+0LewipFCRZRcoLOM799/g7OSap3z4WrHFEac1RQYnBdMMmXvR0TnOCtSFmlgSGpKLchCT7Sa7VDSuByzkSyzj5nmPR//YEV7e8fd7ogbRnrbkmXQtCMrVaCTFkGClZHojsjkhP5qDtmeTz9f8OzzHJE4RCbpxkA9TXgtUPOHXkA6PFDIlFHE0WLyDBknwvWGNBVYN1CZkuACYRoYj7doKRFR4DiQlktKpTk79aw+sQw/esR1PzF8d8mzj54iz+HgbtFJhkwss9kJfTTsjoZsNmdoN0g78eH9N0zTHU1rmFQKOrLOE9J1xU5WvLqck5Yp60f3zD8J3NaGup7hU8ExDPRTQUrBWDXc1B3j7S0LFyhWEjneMx1qcrHE2Uj9/ohLFmwqS51K4scpc7EgDkvaCW6+74jvI4tBo0KJsjtQNW7fMdeawlqaw4Ff/uxrHn2+4uLxjG1o6JmhwifoEAnBg86QlWQIEZ+WJMaRmvSBBzA50qSgLAzxZstcK+Sx4fT8S8r8nO7tbyiuXtG/mVP8gz+jDCccppfY/Xve/f4l8mzFn/7DBUnmcXbB2M6RNhBEwuBWTEPGTEpcAirPSYqCIHqENWhpESIglcLHirY3lKnmTAuErsizQMThXGR/cCRJhpApgeSBF6BT1sWMLAjONQwpzKuKPEs4bC1ajmS5JipFPTmGLmDKOUcfqPsaoST4SJob6q4jjCNmprAxIJRGGo3TCl2kSJ0SY3iQAz24yPFKMAVH3e0RQmDHEe8cfUhprKO3HUIKxqNDyASpMg79jvVC4H2HtBJpEibxcErOTMFsZikXmmMtOPrImYokcSA1ksRkKGXwQ4aWFxA1IrYIGYkhIfEZJx8t+f72e9brEmsjIQnkacKbuxahSlLpCZNllq2I8cFpUYgKHz1WBrybkMmDFCrGSIhgJ4kixSQaGwQ6STB5yjC1ZFLgGJmINNOA1Q+bCYmSpEKhfaAsCqTKuL1r6caIVGB0jtOOo++RUTEESWJKotdEGUFHjuOReneNtztmqzniw8hCBdptz+gsVeqIY0GeCPy+5mf/9u94/uILzn/8Me+rBPXtv0FNFdkXH/P3f/M9ojuwWkfq9EBvv+Zv/+ZrzBTJJo+WD//34vlTPv/zf8iUTjQZ4Ef85sDsSYUhYVQWoxT1YMnpmFc7pLjHTyNFoRlHgRYKiScvHybPsxKEhbu+JdUFPRGrC2o30W2v0VZQuh3L9Zy0e0/28oZV95Ywv8WdwHjMefJ4RdpfMd1vSGefoGXF0Cdozsh8R4yOY39HUWbshgExzclCASZ/cA9Q4sYMw8j0bsdqfkqPom0bVicrot5jVMF06Mmdw44O7wayZYa3MzZWkKYl89UT5nXHuyvPyYsLjqJk9/sNrW+Rx4YzMSKCI/proqiQYUGa/6/7zP+vDwMC+tYjVMVMGcZpQvsMQ01qAwmgRocIAzJJuX6z5/awYXSOZ8sLYtxix56pW8NsyfZ6Ry9rUmdw1gA5KMt9E4jB8X6zxaYZbt9yuloStOW+PvLsozOoU2ZLBaZH7EeW9QqtAmM6IWwPacm7ZkuWKpJoicPAZBO60dGZESMS0Eu8EKwer6FzTMNInsMXF2ua7RERW2wPo025fl/TdRMmg1WV0jVHnMg5DIL8TOEZ8Zs77kZFcfqIVfmS+eqKuxtJ10pkIpmC51h3WCmxSuPtAu0ThDuS6oFn8xMuqjlD1/HrTc2lTejvKzIg0SlkhnNZUrg7trf9w59XSeIkubkaGPwdy/UMGoeSnuN4R54847BN+Da2VGc5zg54pbm8i7y73CEHzXyRk+jI1DmUcPTHhjLJ2d+OlFXCenaK6x114wjGY2QHhaXIS3wAGSTnzx9R6AuEFvzDv1hx/eHnvH59h0hSclsQJkc/9sRHKVEr8tzgjxMsIiFoOno++mzBfKoZ/dc8/SjQTXPmZ39C0X2FD3v0ao04PmGMGbuNQ5+eosOAkJGnH8+4fH1HeXHKeFcjxYyby54izhh2O4xx6GrALxeoMcWPGeQj+13EPy1onWNYGN71MPqM/W33UEKaLZho6e4dL56fU8iImQbGRjAEQRAGd+jIlOAHP/qS6BviMPD+zQ2LL0/xaLTqSMSR5HCJOCzJZl+gUk3TXOJzSapnJEQKfcZnjyW//v/8S/z1gZunI//kv/+nbK/eE+6+x09P+dXv/55//OP/A++/ek17+3fE/cQffnFE8CcYs0fFhO4dVOWCqGC8FQRl8WmLQdLmkSzGh8KayhHSopR72Fjp13TdHKNTMjMhTYJRKVMoQGUIs8IUBVpJxtZTJZpVtWCdaUzXomJDdSopig4RWl5ceErtWeaBGARv7xzbsmAnMt7eHOl9fPiATyP4hESDjAEjJalMOUaHiwmtVTjrUUHjgmSyE0ZP6FRQpMXDTv9oOV2siHbE64n9dM84PgivIo7eT8zSOUo9OBZGv6HUBTLJ8H5idJAkBiMsOh5xomZ+otA6pe8brJ9xt4nMFysKnaLsM5RdEKYDnnuC3NN2KUkyY2kiZZtyuzsyiYHgIod24nY7MNCRJQpjA0VZ4BE4JxiDQ8ZAVea03hEnSUKJEg/Po40pST4jTjUmy2g6RxhHqjwlE5Fx7GgnR6pypIDMGEotYZgIQTGOhnoY6KJHpA89J+8tPQIVFAtVYkwO3rA8fYTKS6yCbD7j4q9+inn6BHszkqhPYf8S3dwT7nps2mLSJcVyTacTbm86bsNjZusn3P3uJQSQIeF4KzDyMcNwhT/uUaKjSnMelxVeKcxMEXTC6ALMI9thIOiM61d3RDfh7ho+WTzjQhS8ev83lBcV99dXMN9y8pcdbR8Q5kGBPg0CpSvKvMIMmhA8mbZ0vuN0taBWIxMGLxMef/ynHPYWe3fN3GQU0yOKokdlHyhuXpOMkU6VXORPuZQ1h5XBmRXVqiDtJtzkkaVkZs4JvmMUR1oRmVAoeyQEsMkKGzyFWnIMAdE35B5ECJxcPOfd9Rv2bs+T+QVvv7vjyUWJvbzGkzKGmsNxZP7oMfZyy3Z3TX3IOb4+cugHxuPPqcw5a3VGnzTYpMPtDGWq0KHnuLnB9QPrbP3HDQPOBoSKHA8NM1mgMbi9oD92lKJgnpTs3C3WdoR9R7IuWeqSyQpMVrDf9OhQIYVmYRSJDhSiIiewtyNtODD6ASVnbO5a2jalWp/SuMDx8p7DrkHqGdc3O0SMHI8d82jJSVksV+yuX+NJCcqQP6o4L3PWzxZkwtHe7tjujhyHlrlZAhNCBWxePhDAFppkmSL6hvpyS7O74fHHT3nyoxM2ly3pmSaXOSaJ5LLAy4AcHIMcefzJc775+/dsW0HaKRLb8/En55ydOV5/NzB2GjGCcj1htkamBU4kxOSCzjnSKTI7WbI8XyPtjqcfr9lnEzev9sQxZfKegONu3LM8SajSnMp3yDEi5Mi6lIhZTioEpXMURYKfWsLgkSuLlDPuxjk//7pFJo/pR0t3HFHiBKEC/eAw1mFEJLGC0BTUIiVg2G96UtlTaMksTejEjrOLE374xV8Q1Jy3337g7uY9/qBxZ4aTz875+Ic/YL+biPE14cM9SiS0osMgyHTBq3cbnq2fIfzAse2Y1Ut6KTnEI4OueXyuiP6AyjX3w9e8SCzdYcH9u4go1lw8SXn/7g1OzPnmrabe7plrgS3h7EJQpIaiG7C1p5kcOtX0OnI8ar79UHG83aFipHx+xuR6BmkR1YrLY2A7LTmbnTNbR0Jzg+l65KKgbwJ3K89sPWO47nCdoJMpjRuRyZHd4Tf87N+8YlaWbO8cr9695s/+/H9PayVTCxyPbL77FZcvO+bnf4GrLbdXb7i7Uex7QZganj56xkfLZ6hoOO6hmw787X/6BT+sbjl/VpJOjt10w6+++tfM9l+jkg3lImfoJMd7wWItmJqa65eCT3/ygrtvf8d47Xnx0/8tu3ZirgyUmkIlKG0IKoM4InXC6Fd03Sk9FSQCsCADSIlSFV2rIckZnWC0jmH06CQjNRVYSxIlZTZ7wAuLiUQFLtY5BR2FaVFSM4Q5mXnO3Q1EFcikItUZZeJZLQs2fUHTTcySFOccGZYhRsbBIrKEdmgw6uFDptQDwGWwAZKH6ZR1PcJ4tIhYHF5DN7QoJdCpxMcR5ANDZHI1iQjI6JCDIkkE0oMlYXAPjIex35HlGTbkbI4X3O+ecP8q4FzgxWLJTy8iS7UDe09UgU0oeBdGLptbvHNonyKyJSqzTMOOclYRR0uqJWZKkJ3B24m0zAgyYocO13u00DgLMihmaYbxGhsncJCHDO01SZ7g4oMHpu/3iChJRUIIAe81nVXkRY73hnaM4AJjiDgZkCrionh48duH3oJJPJXrWKQrusHRjOPDlYsv0IXmiRkx+28oDz1DmrB/XhDfBk6rZ/huz9DvUEmCPgq+/vn3xFRyuB+w8jnvrxvMfosdU+RUMW1umaykTyc+/skz3rw/sHx2SnrxiN3dHYfQ8XrznvqDoFI5jz9+yvLjBKlSrn77jm/+/R+4XR9ZrQt+8N9VpJmhH6AoMoJzWFdxt1lymErcmFKmKVF60AEfI7hImSqqLMFu9wR75OKLz0mL/4ah3nB8c8bjs1+xa//A6eDRL3fM8x5XPGVj5gw6Z9wF1pUgQzOOE4f7juBGyosKaRyh8sxMiq/vkMqjZxXN4chSl6hBU6Vreuvphp5sfs5w9wbbdig0Xi6ISSA0t1TaYOKErlJMKbj9cIPwX6CHFNnXTO2BQxWojoqzJHAVRgaj+fKjcw6HPfkE+9DSHsIfNwzofoWNOQHDh3ZL5kvKSWBChR8jTkSqULC/71k/OyHkFsI95ckTuqOl6aBQpyjjKFbA2D00kidHtk5w0wihJ1OR9v2ASAsWZwn3l/eYxQLVtHR9i5jA95beG0SW08eGu82W23uLdIHTx5JpOKK15OpNg5QpJzLlxeM1nfTETlI3NYJAlhYkSUljO6KNuM2AGEZKKZiZkZM5vP3llvFQUD1asR8HDlIRouHps3MO3T39OLBtNYU8pe22nKx6Vssl2jh6f49JVwQ3YX2HCBmZmjPEjGacmHnNSbXkk49fcHV35CBybo8j/ShYZhUhCQg7kYuRTAf29UAhFXqRkEjN+TrjpJB4N+KCxO5uqT/scWEinp08FOp8QnAKJkXdemSENEiMskQiSbRoWVOZiJoEkxegHTGLxCyQFxl5qnj+0afM0gqTzwmm4Ls3v+FkXfJk/YRhqLn1N+jGs90+4tl6xVBKXv3dkfFYE13B1PccR8d0l9Oe56T5Ent9RaVyAoLLDyNearanhs8+KtHe4ceJ4z7BDGta71guE+apJK17tt3AZhQMdsbuOFHOV9T3lnQyXP7+wIv5U3RqsUbTi8jdbYabzhgnw1TfUopAtnrO3f2RfHnCm+3I0RbUGJ4YS5hqvPWkiwT9+Yp+loDOyS96jt++YxYT0qTg4kng+PKW7389MBl48eIRf/r5Obe/+z11/TWIW0Z3wA0jd/dXvPrNf6Df14jpmgNH1PwUFRTHseb9u1cgU/JyDiJn8+0Hzv93kSpb8OL0hPY7S07LSQzQllhb0sWE735xycVTzWx+wpv3L+lFxF5+R98p+ssPvP7tnuc/XPCXP6owiyWTh64DYTKsM9TdGe30mGhykmIiifJhTUkZuran68GGniSv8CGlXJQkpkK4FNuOzGY5szwh6sjQH8lTz7wSiLEjskfqgnJ+wdGuqOuWTFyQxYF5JclVT6UDg9/io6ISMzb9DhE91tUPlk1vURk415OYHOcFQoMPIyF6TosZ3k5oLVHJjEPTE9FonSMVyOiRAbR8oLEpKRnciDEPbJM80SyUwIWatJiYwsT6ZEaIgbx4xq9fPuVvvlrSqTlROl5bIFN8LHpOshKVVAzdGucCSznyw/nE1b7jnhQ/SrJsQUwl435CEslmhqapQUnGocNkEiUceZmhZEbT2ocrJGlpR4f1gakeiTGiosIJS4iRafJIBN4HsiwnTRL6QXN3cIjFjMG19E3D6XxNYQz7tibPM7SPxDiRiAe51UwvyUZB4hxZpTlMKWm1JFePScYDZ7tL5G/+JaltccUL8rNPiHNFYIPzHTKf46JDiMD3v3vP/DPD2EOXlqRFpDleEUipCjh5Hrj3CW2ckz/6hMdm4PM/+zP+528+cOxa8tmSGCImbJmi45vf/pZPH5/SvH+LPN6RWot2JU8+/ZTTZw3BtaQiQ9npAflNxaFesQsGoRNigEMYaKWjcxOT9czLGRM9stTY3vLu6is+v5gjDFwfcoryv0J+9gL14TvK6QNi2qOv9/DBs/x8hkMyTA221SwWZxxyaPqGYz8wmoHOtSzLU0qzRDYD2eMZPJqw11tS50kpCBK6ZED7nEcnK6Td0bmO68vAMqvIz9fkg6UMkaG95qQQ3NRw9+qKH84K9FSww9NtbpGHFisdkDKlho8+f0K3zxm/vsO2hkKZP24YQDhGYWldRTAXCB1xeiJUCSq1hNhih5EsSQitxe0hbRboNjKblZRZwIsEpGB18ZQ5M6RrUT7g+4nzVKISR+4HXo6vWZSnlHLJMd2iY8dikWHTA+WTFXIrqTc1XZioZgs8UCcTxVlOWC8gGSkzx3DZUD254DCM3N+0hKFFTgN96zkqy7MvHiGayHQcmIxCiTnrz2akdoeKE9b3WKmpu5S7ty1jlrA4O2NxvmIqE+LgaOqRYDLqtmdeJfzlX/2Eosz4fvMSq4+oMqUUOYMZMOuMEDyinejbhqPTfPnDT2l3B4amIxoIImVT14goUJ2AvsOGA73S9MEzM+DdAicFPo/odCS1R5xO2Hd3LIrIoVdMhSZfSsZ6IBlgOU2kRiOMROTTwy57lnFS5fzJj18wyz2HXctv/3DJcn2KzjPOnz9jeXbGZbfHpZDqOXov8Ntf8Cj//yG0wnUpUVUs4jOa74/8vH7HPLvmdzeOT0/+hNEo3n1/SRh6ZvGcRTCMhxqnR/J5oDwVjDfvmE8GbxbsrxouReDTzwom33M/GLyLJEVPk7zGJ5Lk1CFGyaNc0yU19SS4vh45vBeciIKVXTKZOXftDfW1Y/CeRBU458EIzPyU/fFB1jNwRtcWRLFApIa9Hylyxbw6pZQjHtgeDzAtqbVmvbggpoFu31PFwLPFklevjrTqOaP2DHUkr99g0pasuEPHO1TvGfIf0u1LQn2kPLHEIuH9NxN5nMgXOfPnD0AYfao5zda4MOH8SJbn9N2G8wvN/j6leTMxHQpm6jm2HwhVSXbyMe9//wv6/Wu++LHEcsc3+4qtL7h59x1pl/DV7+75y//zF6TpCuNS8OfsW8VxSGinBUFXqESwLAX9pscGQ9OA7XNyl6NVwPceEQ060QQ3keics9mK87MUZwf64BDFiqLy5GaLJiK8xySKaaM4bBSZO2OdJQThSUwgC3vyseWzGWyExdYeIyVKDawWhj5EgrNU8wLbQdfWZEmGCB4lQHtPoSNNCOxrzzCOCJEgdEBqgTCKrhsQqWG0Ee8DiQA9BUyER/OSL87OeLxIaNpr/vqnF1TVhKdAyiWv3z7h11+dcx/OaXWEONI6x/3LSHaQvFhmSAcyd6zXMz5fnrBIHW+urvn3X9/zoTa0DoKUZGbJ4FoOTIRcoRFkWoLyZLpg7CdSLUhEyjAqVKqQSSTXGbapydMUKRWJgX4acD4+WBOnjuAHJOoB94ykGzyDD5i8RCqFcJ4EReYz5OhZrOYYCpotOLVkVhTMXI5iRipWDO4E60t885by5X/mOSO1Ltl0I83v36IcHOYjSXVKXQdMuiSJO/rNwOX7W8J+Ii5zqk8ziqYnNg3DqxuSLMENLcV6zic/+Me8/OY9X//9b+nvr4hNz2L1Eft2g2jeocOSZH7KsduTlmcsF2e8uv3PiLHCpzNk0iPtRBYUwktINd2QsD1KDkaSJQmJ67nc3IHO6OKDptpNI4MsUYsV589eUHSWq3/3C8psxypsES4lnH0B8zMG9a/Z3YzUVca9qHBtS7AHlF5zWp3RjDA4iygWxCyhswMx09jRciGXJKVlHHvW55K2tfjtAyVyHGry9QyrRsK0R0w1s0RybI4cZSSkglI9dGb6waHSBSI6dtdHxJOPiFOCGzuyoee0cCzPCn71LXg9h+Vj/F3DnAlpFRR/ZGthsDfkxTnWlYRY4rwlFQkmVUQCIQYm2+C7jqw8x2qNKCr6JjDQkyWRUdT0reU3f/cN1dMvcH1NGAcEE6kp8E7R1UfkUfJJbujvjmRdoNsN9EGhkpyCFU7VzJLAelXRDwPb9sjAyOnZY0hPqesd7f6aODp29wduWo8NgjB6SjExdhP6tKCpj/T7PdYq3PkjDnZk9+7I2qQ8rk746uqO6wOYVD6keCLb9sjZsyf4CGcfPSfWNTK+QiaB+Tyh2TZ8uNnzbrvDWQFekqQlqhRUTzImP6A+dIh+JFeC6/cvwUzMZ2u8STi4CSFS8jEy7DZkicS2I+nYkScJy6qibsEbzz/+Rz/lH3z5iMp7/tXf/IJXvz+yGDqiLFAyJ18/YvX5jDcvb3huM+arObN1xnqxZPlxzhQ6pJxYrhUyTHySpuRffoRJMpra0zUpXUiwxZqhszS2I1z+GrP/DYQRi2JyATsJkmRALh2zQjI0R9p3muyLgqPYsv7xE6Zhi9YTyQeL95HqWUl2ltL5hk8uTvj+lw1dOKJ6QTgeKTPB+XLGpW0h00TVMaUP5Ee8wvQTwgak7jlJlwxtoG0FN3ZPWqZc3luOWx5KPIMF1eHTh9JZcxxIdEasa5bFGUcRCXFikUfs4RYfFviwpu9v8Wqka0EOUJaw2Xu0f0SRHFk8dexjypg8ouv30EeuXmV0u4L+J49wPVTzBafGcOfWTGcGzhyujIRUoWXGnArRDozNhrq1PP20JEtarJXY71sOxxyjTjAyY9qNqLBCnJ2SlwXl2zeMj2dsg6PfOnwtKU5P+frnt9wcnzFgWOYGQ8IXP/2McjEj6oRXX/2CoYZR/wNsWGKdwKSGWdcRDkdeNDO6DgYfUOIBSdvrkTgpWjEgjSHNDYnKKBNBe+hR2pPlCcpoVsVIgiVTkCUZzguSsCCOJfN0xugzRhxCHDmtBE+THqm2XE+G+2aGMHO61GH1xLHe4ZzluN8jfAQio5uAiEkV89nswVY5CYbBEbUCFFoLUB4fHVEppDSA5ctHTyhHy+l8ycXJI4Q0DMcDu2bPD55WTMPEVxuNdY/YHTLq+oIxvcCPD5bAqRd0IXAVerxQvA8Zj1eP8NPEfD/xw7Igywyff3LBxZNzfvV9zXXfctjvuO8i70bHqAVpVhL8gwdFCkeeKkSIBBcYjHwwpEaDdx1oT1WVpFmBSnLawx6jDEYLhAikJmdRZsSgaY4dwYLz9oE9kKYEoR4U365HTglROqIoSXcN1b4gmJxRZpRmSVKtsb0GndNGybMsZY3Fjg50yTooPlkvaGcp/XDk+q0jSeakRIzJ6MYJuhnCJAw6YbuvOd9Lslgxdi37jaVLFIuZ5Jvvf8+3L99Rf/0dKrG4ZcZ9PTLcTcSDIV0vSC++IMw05gj2cKRcP8FNG777+7/hix/PmT8XpHJOdJZmjMRpRWHWWJmQKsMizVFTJAhFIjSFKhhUTivWOBSTs/jr9yxOLPXl95yFO5LdHV3/GfdiQZGuMC/WfBi33FYl0c/RoaVtB+x4izMl6TxjSixt8rAB48eIUZYkGQn3jsRJwnYg04YPtx2roiKVCce3O3xeMoyK8bbnZHlGSDJaD7aVnJ09paoy2v0lcj5nPNQcrw78Ot2wdwlGrJmLK/7JXz3BtBW//8Ml/WJg/k//KWNREl/dkeULqqcf/XHDQO499fXhoXi3kA+ju3EgjQVKpBz278mMpGVCDw2+KNjolvnsMYMPJNohuxq0orOaREwsljPcDvJVges1u3YkWZewcGw3WxIBy6wkTpJN0zMWOd+/3FAkjio2WD9g5hWKDN17Nh82NKIl16DERPAW0RwxIkVgmWRPuQiUsSV9nvPo8QnDdWDft7zbX9GLgp4MpxNu3x1wNkDUzBeS2DUYA2ezlNx39B7MLGGaBj75JGdRGJTbsd19xc0QKXPJPB1pmh6vA0r0PH38nCRP6Z5rNvsj7evXJMWM+XLF8Taw3bQcGRCVoVCSxEm++MFHfHj5gaxpeLI4oSg1H/wI64pBrPmwH3i2XvD1yx4r1oyZwkqPSAxfff2OJkD0MzIh+VB/4JGo+Olf/4Sj3XJ/3UAwyFkBEapMI5OKwRSoPKOyKaKdKPoNaneLHr5DHr4jSMvVN4r1agXqgdR4u3lPaRS9KGm3kdU8I2QtX79pyWcVZycrRDgy2A2EOVpqUIFu25PpgvTUk2YKuTEkGOxhx1Q3iEPgrk+YzxaIziNJyE1JVEeUMHh/5NmzhDevOqYhxZicfrIk1RxlJrSJSPWwhqZ1QbpYMDSOVJVM9Z7oL8lOz5kHxyJJCQj664aJFKPA2gkj5hgy7LHB9ZFeGeRpyvmP1tz8fgeTIpURpTUfthN7UXD4SmMPT4kInswNaSVxjxOmKNB1h5hynpw9oxgk+14Sbuf0hw216yie5ci85OLjwKHfUZYV9/ucwRYINTI+NlwNgSNrhnbg/X3NfPYUbxPaW9huIFEJifYkhcLvR969/g03u7/g7Vbzv/yLHf7Q0FcrTNXx+KzgYg7ZmPLtz27o7wY00LuIKAElKR5nnK5O8bOO1FdoFBOKjIliZpBpJBcRnQkS2YEP6ACJePA/GJEgY0IhJcsyEmPNsrzn+folzxbfY5Kar37/iFnxBXZsuW3hbnPNsXNEDTpCkmjSPMO5gDTgoqVzDhEVt3uPDZosLxm6nlSnuGlEekUiU6KPZEaTCIcdRgZVcyccXkuGbmSVpcxPP+J2e8XoPiUxj2ndxCjmuCog7UQVI52amLqWID1H27O97TkOhlmqOYjA4e17rk4L/vzTGSdZ5C++LLBCY+KMY9Pz1bs9r697rpqJzdRyqlZYO0LnMRbmyxm3bqJxjlEJdHSkRJQIqBCZJhhGS6oFNgwoLSlMSiISDq1FiYS08Lg4oXXKOFl0YjC6pAwZpSypgqI8KMKvfkvVOaTK4OlPIHtBr5bERJHNT6idwPUpbloyiwtsO5GFAZPBcv1DHmXnbLZX+ETQHo6s5YzMSbg50tuBIyVn6yekB0EcesKx5FhDk+csnGN7/zsWpeOo9jRWEPSMbb/neGj5ybOfYuZzBqEZrw989/o7kt0dyR5WT06ZTMrb94LHjwqiGSEp2L5xHL5NsUlBoguWWvNsqcjkA2tBJgUbZnzfpAz7gj5OmE5zMntOcjLDfHoGv/qfUNfXyLt7LrdbTlZrYioRbYEcDYMUWKUQTiJtZH0qyVcjN5t3ZGrOUkomtULLFlH2qD5DporNmxuQFccpRYYOJSxqucLFBNQclTSIoJglhqMd6O3EXZeQGskoc4rFGX3/De3lwHYoERefI0+fs3Il6ephejrVE+msYL/p0FFx8uRL4mzGrvkjbxMkWU7qM9xoaccWPwXSQpLJQEw9hUuJnaaocspScX/c8vSzR2yPE3GSyGgRTUORzolKEMXE1LZ4N7EqFkijcAyMViBFhlIOESaKIuF2t8eJCVkVeCJKaUSEtuuIUVGWjzi0YLRjNfMs04zubs8Q9qyKlKg8m82OReb4Yu05LWfstiNaD0QLZ0WKbDpkd880euLekU05yoI3gVwLnjx5wuXtB6q6xbrAy809l1eRJzLyqJqh2z3ueE+RCP7xR1/w6AfP+c3JJb/62yuK0jPYnqufv6I8u2BIDVZGeg9DuqI7DJTLGYUZGPY11VnF3asPVDaj3nnqOmIRxDxwOL4jkQs6U/H7D0d+9+qIOXzN7dtr1qc5xCWDaChLjbrvyPQ521YwzSJPzz7mOI78v//lf2BRFbx6e431DX/xT37IX//1D8iLiTSPXG4ir97uaK7vMPoDbL4jGTvSYmA9RoRRbJKCNAhOLjSXx8DmlcfJiXo3EjtFvrBM2w2InKu7HjtMrNREns45Npr6pmOZz3j3dcdyDavKUK5L3t7eIm3JYE85NHsObzvOnpxylJ5x2lMmE15ETi8K7rZbooXWDySLjMorZOuJzlPv7khSSZJAcIJoC9rJYdyATBKcG4nGM441J3pBnj8gcY99ZJoCIemZLyX2w8CiLJFixzDUiHhC7yPpLGN2tmZ/DITryNRpsrkkHCeiO+W+XXLcXLPMBuy64O6geL8/oPMFieyY5YpsrlnmC97/quXujWV+5vjhJ0/YeMmxbUlzi1sLRinYHaD48RxXH+hUQ5ADV71Fd4JuTBAz6IJArZbo84TgXqOWCvE4RQ4Popz/+d+85agT6m5FexvJCk07NNz/bsf7PGXepVz+riYRBr0UDB5UN3JWnTB8P/Dh9RuOvuH05IL5sxR5qvnky1OKKjC4hmIEekeyLsnyOYlvMSKFMEPKDIxgWTnOZ+94vHjJ49VrZulLhLkjyTK293Nefy+4PkLrDYleUiYjZBHLiNQaN03oICh0QtT6AT7kE3IJ5azCWtBpifBQiZLSRISwzNKUi4XCTzVePCiH664mn6VIaUEn3Ow7hFvz3fuIj548L5iYOA4HDl1DoSR2tCQIBkbEOCFEJDWW1u5ogmRvJF+/rvnZ1S0nOvL8pODx0nCSKBZ5wV99nvPF6Ug9GPbdiJsC27rl2Druhx1yGHFNh0lmiDHBaE0yPUxWUy35cOwps4owtSyKCmcHijQnkQYpA1pLciQjgWlqUKJEi4q2kWgUpc6ZX0ue9pbpNrI9vuduzJipZ1wsAlZkxHRONwnGaWTXO/bdxIm4YzbP2bXQ34zIcmL1xSkXT59y0+yJ/cj+fs8sHcgOA8aP9JNDvXhOUxvE6PCTZPSCkz95zPLjOWPUlGlB/tFn/ObqPc3Ocd/t0YkifbTg2LTsbzfUb98iuw+YyaKl5vTJC1ZP/4rvd+95/uaetKz5+nXD5S8Gni2eUMRrZi8escjPKbKEGBom2yNjQgwd/vI7zs01w/hbVrOMk/mfsst/RF1+iTjp6ec9yvfokHHINP3U0KoedI9UgWyuGW9G0pgymyTutsN0gtKUDLFB6ZKu6YhVSktkvGto3wVOzudotigp8FaReIMTkqSaE4TF68iw61hWOZt9y6Hp8YeWz7/8CZvtlnwKpErgaofVDf7sirt45Not+clffUby7244NiNWHvn4J09p+pxdCMTN8McNA41wID1iaHES6sGRuoFyrhG9xR87ZCJI52uE6BHdgL5yJBLMMBDDjlRqBDnOa+xhw9h7vNa8e79FukDbHTjsHImVlEqTF9AFcN4jVSApQBtJ6qcHfGOnONQ78pNIoQdy49F1y2I8Zb215POKuZVsuoFxr9DlitjvcHXPKi2o8wPzkxmzdmLJgfOzjNr2KFuR6xVXH74jRMFZccLdzUuEg6YXOB94dLrAeM/ZQlEBkYKsWcA4MCD42b//lpvhntQZZPBcPDnFeuj2N6i1gbGkbyO7okYmgtpuiDGQzDUiSbDSMGrLh9dfk1AipSQzHrmWNHZgUB9IfIpODJftDUvjMKbGyxm7fYO9TZiZAjm1yMkxXz6lKjWbmy2/u7mB0DIr5/hixv/0r77n+2/e88/++jnNNPL7r28w9pYsXEO6496PlOvHKCVZzVLs9o7VRWD1tCAZ9yTvMh7FFVX05E3LYhF4d7imva/IAsheAhKdpw9XLvsJfQRxOXBaLdCLDKbDgwJ2VvDhvePq24apT8n7lI+rFWZmSWcKYQdktExqoljk5F3CeAXVGEiHjhgUo0oZhhG8Jo2CqkiwU0IfNPVmJJUFo21RKsHVke7NlpimHKMhJDP8WpEXkSSfyNMcppFMDpRKcC8co4BYpBxdzjRPaFsPZ2umtEb5FFzP5vaAyeD8BxXOHLHXhmmEqBWz81NCnNhMA4tVjswirul4tp4xE5Zf/+GOcr7En8AkU0xScm+3zB8v+RA0pcsJdk/6+Zop6bi+bXlSGcysYHuYOI4NuZwhJ8Vgc25RbI811XtL+twwf5yQmYhL79l1a378V/8tx+0Nm/e/xb9wzNYz0rJl7DXarFExwTYHUpFjt4bdYcC1A7d/6Ni8OvK/+Wf/gJ/9x++Id/ckPvDFP1zxj/+5IMkSTDAM04JsdsG5mbEqjvxw/TXr8m8R8RobRpwMaD1QzQcCE8acIPxElgaiWtKJnihTpBAkKmWRR0y0jF2PzHPaJnJanXPsPF3jSLOSyfUYKfjRF5/THu55erHixRxuDzvuty2jPeBdz8lqTWIUsySgdeTljeb7XmGt40WWc1pkfPH8lIkrUqdx0y1eBezQEaLDS8vkr6m0wHtDF+Hgelpd8urQ8avdlkfFipKU03nO83XBqSqZJYJFrjiZT0gDnY90Tco0TXx9Gdm0kXFwjEFSTytEFAjnGMOEkQnGCLIYMKlB55F+DKR5ga8dmU5QwlJmOTKWiDFhdXJB+WiOHiNJOfHmb3/Fx2fP+eT5p3x4e8v923c8fvopTIF+NCQnf0Gu5whxJJ5+RrP9lsXqHGJNOkUEE27SCA0xqXj6J0+5/5uv0Gmg3m8Ro0AmFZffvGHmBqpSEeuEvnb8yV98Rv7RimB+wO/+3c+xb2rW1SOCjFxP97TK8u9/8z9SJYZ00oy1RXiNDhWPLhY4Hzi8u2JaT/zHvzti7mps6WjfWi7Db6lv3hP/0eecPf1LZPYY7zNMpvBZRnt3jY+/QR2/5oX+ivHecfXuS+Tn/0/87B8xLD9js1qA/IB9kTLKBhcF85hy+L5juldkFEQf8Fjq2xYnJf1Q0g4Doz+ymHm2Nw3dccZgHbFzzPMF3kO20EitiKOiDw3j6CB1TAtHI0fGqWNZJpzYlDKfPRQL45Z+e8dJmlCZgsY4RvsO8faKQ5rR//AZmDlJUtAfaxYXOa6S7D5Zc+gcIm7/uGHA9gJ7GDkr5ujU0YmBZOp5nC4J9YBez7mr7xinI64bKerA1H3gyScrpDhS2wGbzHExMNzvKBcL2sbiqwRdzGi7DUFJLlYrRDeSmox8nWMKw3KVM+wg9B7X16gQ8TpCHnl2uiavDAk5uRNUyRztBbab448HxNRQJhIxWYLsSEvLTClUlnMz7dFjZLq/YTFOLM4/5e2byLw45y//yZ/xdz/rqG9G9DHnWa456pFmlBzriTwGFvOU4+aeIDRZInh2doZzG7759jW2nbP4bMVCOWShUJmj7XsWVYWqDNfdntPTEwa7YVKevDhFZRn+0NJ+uMFMkkoIzk9ShM2o25ZikXD3viZPVpyhGa4uud33jH1L0U7oImNX90gxx4gSk2pinDBiIMsHuuZb0sGSMzAaTZrl7A8DTiS8fdPxr+ufE9yWUVu+fNHRH+9RNsMkMz60ll2Yk85Szp4ETpOJdndk/zIg7pbMeoMYBKJKWM0Ce6E4+JH141Pa0CLlyPEwkMQC1XpyQIeJqsr5/g93qN6RfJTgN471/An7IGh2NWUlsSXcNbeszmcINMf7Gp2VuL5j1ln8+4J1W2L0QCMajtWc3bVHYzBFitYepwNBaop1hjwOVKsF982ORBkqOcdaz1Tm7K2BIiE9n3H7+juEFWgZifWIUCkYT8ST5nNud46vfjehwzlPP884ikvSL1b0716xECVnLy4Y5J6b95Ys1zz6iUIuFLkQiFozxjPuqozjcsNZ6XlyInHHFtPAcNOT6xnHzQ79PKE4nfP9h0AfzxmSA6ePK6ztMScrun4D3mBDi65Syo8Me2+om4qrW0lnM+ouYbWv8N09FVDEGfG+ZJafMyWCYSYZZhmr9XOKXJMnFbev4Oq1wIQGoSNZ5hGZRKaRKA2MkvffNfzrm19wt71HTJHQpby/vmP5dMUXPzaI0BGCZKEEF1UgN3fM8rfI+J4gLUiBBER0nKwG8uSIHSoynyCkJE9n+ME/tO6Lkug8y9Igh5EyzzlYSKNhLhQ+SkavmImCbLlm6Bv6JoBL0TKhbluGYcK7DhMjAoGOnnkhebIoyKKkOwja3mGngVevd/iTNXkSGMeRspyxujjFuQ7tIh+aLUJE7LRDyYRD4xlVymQtoxsxKsVpwSEE7qLlQ6f4T5s9Pzx7wnOZUAnNT1eGk9Jxnjj0mcdIy1/9iQE7MHU9V3cjv7vKuWoKPtRQipTYPZAHy8SAFnTHI+MEQSpkN1LqlPV6jbSGnpSwOCWfrUmqGYd6IMxnhP/qS+5uJs7EjLl5w2Hw/Po//BvyUpKeFEjnyauP0cOOrw937JOSP//4n2POO+S3/5m23XP1zSX3KKaQ0iOZn5xSzQQ+Bo43G+63W6QNIAXV8oSYCbwZ2d00uDzj7t2v6d7f0jV3zLMTTrM59bzi1eEtQgmC81gzJ+YVqSzIdUqRgtjcc9vcE24Cg224cDkytYR2wA3XpEEyy5fo5SPkfI63HRFPh8Kc/4QXyxPiH/6e9sOGeZ6T5Fd88+t/gV2+Q5vA6smf4x99yrb/jjC9AnrSyZAtz9i9DegYyZdL2n7PFDuUWOBigqlycg6Uxcj+KkXEOVrvcXmkLTxeNyRrjS0FeoxkA5jwgLwXucQ6QaISxDSwXqWUhSAVju31V8TjnNnqKWkqaacd1VKTNIH545zi9ITLyw/MM0Gy7xgONemsIqt3jH5GY+MfNwzULcQeRKw5L2bYUcBREhJNuw/Y5hZjJEY6CBG5Shi7kbyfWCYJHAcOgJcji1yTqEA6K7gPLX0zkvXhwdUsFNEIpPScJjn0W6buCFNC3PUIaRiDZP3kjJGW2A5M+xqpC6QusStoxp44gmDGqCR1feTRacFcD5xm95h9QlsLztcZH81TclUyDQW/fP2Gs+VTrA38/nevON5KUpczjgNFOuNQj3RNjckr9BSo3x9YL1a0/fiwivcoYawnmnuPTSwXy5JP/vQpv/n9W+5vOpJJo8rA/eaWgUBmMsw48mj5hO2xYdtuyVrJiap4ka8wQGxr7u/uSBaeyXkkS9rrCSsswThMGYmJgLJks28ZyUiLEpXm+O5h5Hu6OmGWl2zfvEd0AkZNkmX0zQBaP/DZO4Wn41weCGeGcVPQNmtEnuKFZmagKnIOveLNd7AulvTXAb0T5PLBRR+rjPKHhnhRUzQFV1iYafSpQNqR+UnBcNNwOku4KALdfc3Ul7jtnCRKDm8SnMsJs4woNFkCRg4cDzUqDdB3qBDJ84DxBx49Tqjqhuu3mnEXyaTnJNdMSYKaC2wM9OFI6hOiFFgxkpmUYAcCBmE0fhqZup48TWmExMqRVEVEHxFBM5kEJz1KlUSZgDCUSnA+T2k3l9BE6rjh6M/5ZisZkWSHlBfnCygjb1/uyfqc5arAzCN3wy2r6imtd7RuYhonsmXGk8oy7TZcv18z3SRoFtz/bmC+yjGPCw6XB757K4knkYWas7s7cpZcMI17nn1WsVxkvL/bAZI8n/HKei4vC+ZnipPVivoQub9p8EHg5k9JCk3X7pEmMkwjoxekJy+w0fGuPTBca3Z3Kdlywcl5SZIM5DlIe0MuJ4gGIQu+/7s3jO8PaNcyUqFUiWXO3/yP9zTf7ZibEdfdc3Yx8mT+hPzFluSkQRmIUiADCC8IITIvjzxZDIjvO4zylDzAWURSEKKm6ScqaUiQKCWISuJihktyElWgosWIADZQWcj1jCRJONiRbz5sOEkkF9Wa9Tm0/ZHWaqrlM5RUvHrdctx84LJZMGYJk69xruft5RGlKmrvONoD1o8wtSAcggi+hzRhFLCpD8RiRlSaLEsJdiRYxygkY6/onMUJzTe7nms8p6Pl8ycFy2yErEAZi1Ijgj1GHCjlnnkp+PSTM7oxsm9OaO1Tbm8s7TDg3Ug3dtzEE952HuEky3lK4VPu3mwR5YKdnNGqNVFUdHVGUAWNCnD2EfWZYNIC/WTObBNwyYhwDX63Y/fb/4Uh/C2puOf04oYdPb+5eU1azikWJcdm4PbNPdPZBZ2QvN1cc9p1nJyvEN2RR598SvPNhuGoUaslQ5OyqXumxGPFge7yiuZmZBwhmadM4wbvd5g+sm5rZB+RIgOjOZm9QCcKwp67Zk+Bpgsjug1o1lwNW5ayJC0T8sJwutJUZzPefbjjcZowoNnakTpYisU5U7/k9Py/oxB/yuHyNYNruHr5DXp5iygC98VnLB//BeL0BSmf0m+/oznskXpFMhfki4JYtRQXCWPXQzvR3w4kPvLJj+egWq63ntofqE4NPipYG5rokXqC3FM1E+f3M8phQTfskfXAmSzACW4OO/rWQCpRUrIIKVtpyZ5f4LpIMkg+/vJj6j+8J7EH3vz93/FdB4dDS0bk1b//JadPInN/zrevrxGT+uOGAakVQ/QI26I2ioWeMSQ1MQR8IbjbdqSUzFeGZneP8xoZU2Ju8NIjAoR9ixcdOilY5QX5GuzVxHbXIhqPiQI771BJSh8e7rxkvwVlHzywvaWalwxmpA0tImqChWI+Y1EV1PWe3fUVrhsp2wDCUcRz9OoRC3XN8+aexRTo0IwhsJIFp+MT4mh49+4St00AUHPBZntDmk6obKKtPdu7LcmyYpGNJGmN8xNJsUBGTVEKklyz70bef7/BColMUxg87dRjYoLvNU6nrB4tYPtgUrTtEdVZvD+Qqo5PHi+Rc8W0G/jw9iW68xTJAh8dZ89KdtuIlGtGtSXGQIgPH5NpcAhZMboUHwTLx2fUdYftBItHp3S7A91X16TFnNb2yMRgbU+WJEwxElWCTxOSNOFJXmL0it/9dsJkFwxFy/xJTrvZc/pCEYBi+YL7t9fMwwn5OlDOcup2B5WHzzT3rqG9zwidANNy+twgJoFRER8NelJ4t2c+UyRHS0GGjCNTN5InCV13ic6XmDzDMWI3I+v5EikHVBaYdpJ4LzDPBwoZaNyEFZGk0CRqgOCQWhNGi+8sLlWIAKnQqD4+6D1loG96TvM5BIuXlugHKqNwdcvhckREiSWgQ2CKnmmKDFIQQmDcXePGhlVyyrMXK4KyaAr6wTOZHD1LieMR2RlmJ4Zsrtgc9pydppyuFJuXA6M4I6ky1heW2BmCTWhaz0jG5AOXu4k2KtqfHem7BVqcUg8dSZ4TpcDkKWUxcJwC10fF97eGoiygOSDjCpUozp6vyReS41Yw3Y8opfCpRjyesVqkDFMgdh7TWmQuAE1zsBRhwcWzBdF5ohtoQsf7qz1ubDFBMZMwHo8kdxMz5kRfEdKRcuGQOiAaiO8Set+igqW5+p5X2ze4VcM//X8oTj7TRBzRgs4VWjqUGniy7vnkJOUoDKflCpOOvNlu2B62iGgI0pPqhAxLFBL0CYPOwBsMIwspEd6RScE0dKRyQZCGQyepDy1aGewYUfmSJMv49Y1ktx/JGsFKF5hCkyiHnzpEvyOalKAsUir2o8PbQK4VhIl5mTE5SQg1h14zRMs47AlRMrY1aW4wRhNih9Yp0xQwSc44tdww0g0dm8MA3QeO5SMuLgKzcsTHAaPAhUAUjszcUGQHzhcON3nUxQztEnwDOi64alZc3s043jwE585pXLTc7g98eromZhkhLbjclxRzRVpJJrskKsGQOLLZAv04pVeSQkuyq5FwMvJhs2H3dkMQGVQld23NWVgwv/gpMfsOtauRxwNmFcllCbKmud+iW8/Zn75ArTV+N3EwDhJHci4wznF6FmkOH1g8O2H7wdD2imQK5IXiqclIjhNmfFiJbBqBHHvSCoLeIMKAbTVGKiYG9u2RtKqw64LP/vwz1kWkfPcr+s2Wt+/vmcQ9Tz7/EYWsUHHF1GegnjMuzsHu+dD8jE27Y1+8pdh/x9JDt33L2w8T/dMlq0ef8vj0f+AuvOeYXlKv7shWkcCG4uwZeZNQNz1ZaVgWGeu1x5qBRz8q2IUHF0SRJ9z1l8RUkmYGKwVFESCNjPsjth0wPmVoWuQ0sSgqTFphhCXonv7+LYtHTzi4W5ZqzboUhN0lx2NDetNho2dSMI4tTkYuf/cdzbcDKrlBnH3EyYsv/7hhoJsmspOSaqVZLueUxZxu63DRw35AZZohtCRWMFvNuH29xSQRihVhTCjmpzBraIeJJjrmKmM5mzMdI8XUENOe/tiTR4GdGoZsTSwFwxAQMiNblky9Jy9SEuXodEc2X+JqQZjPsFZgppYfPSkYrx3GWPoYeLd5y3xZ8NGzgrkrqVSBXGdsbq45PY+Yk3fskx2bm0vEZyVq+R6TWS7mJURLdArdFvSNA32k64589tEL0rLk8qoh3BumpuH29RVPn804/cHAZ4lkbCZ273suu/8/a/+1q8uWnmlizxgjRvjfT7v8tpk7k8wkWUWW6WKxnKoFeTUgoC9AhxKgQ12HTqVzNRoQJEFqCUI1qppkMYss2p1u+2X2Wmva34eP4XQweQl5FlcQ8cb3mkeSOM+8NBz7irurjv5gaVtLkmXoKGdoDOLv6WhiCgKPcQ4bJMFaghA0QXB0t5yc7YmfbUgKze33O6rrFNGdMGLBREitqNY7Gj+g0hyjJUPoGQfNMBiCMESRppgoslLhdw/qXKQxfZVwqDPGm4piuaDexUhxzruvDIQFGQFZ36FON3z6BxU4C5Fl7Hrms5xO5IRYsX2rqY8Zk3KCCIrj3mPGmE0lqO5KkiRjkAJRBBLn4b4lS2ZMEosyLd4LVKlwKdApZJMT2gx1kBgPzRAT0j0//anCtRuUmpJlJTo+4pyhP+zR5pQ8nqCHFmMFuZgQ5Qkte0RmqOueRGXMlhOEGem7Cs1IbCWh9YQxQGQRacxoLcvFDLP2RFKTpgrTjXRdh+vuEN1Au9P0m45sOYU8omljRJ9Qbwy/95PHJFPN1z/b8Gm+JFMV7c5x1DlOOLJzze5YU+oJJrkFJ7k8n6Ebz359z1G1xAVkLuVQOzaHniKfct+BdSVh6BBasW0kX3xxy7SBYjKQPklYuz3tqxZtYxIpkVHPsRuZJpLVJKaUCVbUGDPSXG3pa8vQakwRobI9ynSYBnZtg0hX9F6w7wy1DCxMjHCaIXP4PmZxDj/+4ZyzkzN8OLBSL7HGsXt5RV/D3sSEjeaL/3TkX3+WokSH1xJvHVpLvFMEb7nQE0xvaaqeYT/iXMzZ9IIwePq+p9l1fPhkybSccLWFXghKqZgsc7SS3DQVXRgwZUx36NjtK7rI48LINxvDMg48VSUiLfnuZsciLZkuJcsy525fMxjDvMgYnGEMDSo4do1k03VIoA2eJIHga7QOJMowjs0DYEYq4jglmJ4sl3jp6boeERIiNSEOGQ7osThlqZoRGXrq7RVar5jkEc70yMRhnAURwJkHVLFq0EpgxwNerFBSYnq4PVpevnZo5uhixpOPT1G2Yv3vfsbm7bckmz3aK+JJwnKaohAY4xBNT8gNpYzY3b8hhITTFz/FzyNaWfF4copb7+h1jLsOLGfPuKkTrmRM183xp55pVKKCpt7vUbFgtzPMopTd5oCexUhXI33NNFGM3pHHEW3b4nxKcBmXn/6Q7143hL0hSjOm+ZTd3RVmtEwfJdje4gdDljkef/wY0Qs2rzfUruXy+UeMrefxxWOWn5aUH19yeP+W9f/vBjcmjIuY7bu3LKbvMORoXtBu7imLFb2a8e72htd3vyJMLjj97J8Sv8mIxhtIIu7uN/T9HtErpudz0vgcc5pzfexgPPC4nCF0Qqss+cUKM91jT1uMn9DVCYuLlPtXB4ZG0Y4tInIEoRkHjwwCXay42ayRt0eKLEOMHcorIpVjzEDrrinLnERlNKQcrvasHktiB1m3o361wZcz5CSgxhStHdxWaLHk4CQnT35I+9bjj0d2/cvfrBj40R9+QKInCGWwtsGMNZPVinwSkd9aFpcFURZhh47EtKw+WGKtRCWe7d0tcZ6CDBSLGNcJvvr6cyKVURYFq8WC0Es6+1AhaoNglyoOwiCiHBsCQ3B4LRlFS5EK0sWSNjbUtUUfDqRWcbaISOoGdYTZasQNFhUZzuYxCycQesEmkaSfRHyWeOYf/Ry/dCT6CUXUsMojzpea668d2zeeJgQiNaEZwMeGTEvkqDi8BRkNrK8PcKgZxgHhA88/GPn4RxJ5HPjj//AWoxXGd6Rlw+q0YCamrG8r2mOMk4Ihauh2BhllpDqjnKWk2YicBJo7C3aOtwYRG6ou4fHvKaaPr2jNjmYn8dMUVSuGg6HMc9q4QQaBqTouz5Ysz0/ZNNfMJwf62pEQY4wmkjHKC1TboXdHVnKCVwldZTGzCUlWU/WW3qckISFSCZWUvD1CWQWert7x4aeway2NGYjjhynVsCtZvxEInoC2xNJhzMjYp1SHKetri2DGelQMziOHFjKNEh1D77BOkqkC4zrcOKKVJUtjRJNx9BYxOM4Xc4JuMFXMzVtHtVNsd4FS94jC03aCoQ5I5clWDx8XkUpsbXG1oziZ4KKaftMQBVA6YKcl1d1IGpWMh5Y0niJkizdHEBabZByqiCA1UWzx7sDuEOHiOSHsyI41YlyQHyLipaKqRtaHmq464E3P01Lxi2++51hlDOuYqmpxtaRYaYbuSFsFpudT4jQlerlg/2BioPsObQcm5zNCmWL+5pqJW2GdYDAOF6WMSrI3mkeXOfNDRbuz0ChaMWJPPVUPh40lGwWPdEphOuJJzvu7mqEULKea/X6Huz/FrUuCDmSzgtp7JjKjFDlRbnF2JFLQppI6OEpjSMYObwJt7nEx/Kt/+2/4N//sKUbGrO0bpolg3N9Q/XZMc2h4apIHcM3sGrojOldYHASwvaIbLnHukjEIYp0xSkFUgmk3SAducOh0hvSCTx5rlnnC86wjuBrvboljQx8UaZrwdj1y3TjqkOB8QJAw9C19SLnvOoz1xElLLmNkJDm5nPJiWWBTw5kqiCvHbQc2i1jXWyoLLgRkJBFSYdWAUJ6AJLiAFgGROvADzvYMg0WkjrF3KK8I1uCDIy8jrOlIdcY88bRNxTTT9DLhz36xZt/HPF0WZOnmAS2NxKPxQiGUx5uRQVzyH/52wqHVFDLlF68Gbt52LMsJJ+cFwy9eMuy+J7+uqZsDd+1bfvjoKcavYSXxVnL78iUf50+o3m2p7MDt678gtI66tiw//gPkMuG7l39JEe/ZuwiVnDLaPbIX3F+tefr4BXFyQfHogrPFCV998+ccv7ljmpyRfVDCZGCRtKh5B4eKJMnxYyD1U46HhHGYUNuI1u+wQ4mOJqx+8BQjAqfTU5qbEdvvyVOJ2R3IV7C4WHJ4X3Ny8oizFznZR4+ITyaYZiQ7mXJrLcN9j6hniIs5BzVAnfLmveE43LO//YJnF9+ivOOuKlm3E0yksFKiVcm9r1nMZjR4mmpAxSnr27e4+sDp4oKgYCkGQhJR9w5ZNfSjIJ5CeZlCVvH17YZ9X7N6/ow+hU1/DUXAjSOma8n0DHEYuL5rKIQif66YPk0IR0/zNjzshxQSWznGuyOHpoFkStsMDO/uefJpRtw0eBsosynRh2dI1TEfDLkZuX5X4coT5HTOKCsmqwlKFb9ZMTApExCATLBKoJOYj374KabdYnxNueooJgXWJsySCVJ0RNIQfIsdHEPX0vkJB5sw62IetymFSJBBUx1rqmOPTAZc3zJNCoTuGf0RlWiefrgkLiYMY0O7viUvNVINHOo7poUl7h1xtCRtI8K+AC14/JOPOX7/K+p3NY/yc04vpwQq4qTh8ndb4rDGcEWUnlDtFbdfLSjTnMUnCrG3tLct8mRGMZ9xHK8RccJkNqd+J7h93zKZT1nNzmnNhhAC8cmE0wvFRO/Zmw3pyQmpgoPdk7opoS8Io6K67Qk64tnTnotLS70xfPXFQDIU+CZgJhavBiYXJ4xDTGxbtIa4cyQOQJCIE8w64vD1EW0NsRqZJJKziwWpDdzfNOjWcPzyFfHS8s9/8hlm3/DmqzvuNh7fGhbPzhntA/+7r1uGRj8MwTwrOa4bdp3CCcXY75BpilIz6qMnhJL1OiLxOevv1+zahM/+0YdkpzHfdZK2FozSUz6fUm+/I09mdN8G/LAiC5a4LIgVDG1NnswYnWVxWrK/qRCypBs6Wm8Zu5bJao40AzLO6axChJhhqInlga5RfP6XAsYLpvOnYCuO/Z79LmBDSd832EEyn63ofMPgOsYtOBEzXy1J/IGuO3B8q+EioXUJw84T5VOSyYrD9SuUc/je42OBKieIQaDjnvFg6W4FJgSePF9yfql4/dUO7RWH+xFZLHD+gI4bnj85Y2zXuLElyECUWar7PXGXoQ87RKlp7gJ3fYQVhnIsGUdLImL6fs9iNgcVsds1PJ8nNMOe2ihu25Zs8RQ5yem3Kd9+dU+8PTAZY0alaZzE28C22eGASZLifcAMgcjE7G72fPLjJYvQcnejSEOKzAyJUrgoohoHNpsNKI92lkR5JnkgCYKQdDw5nRBNJOu+ZvSQzQquv/05f1p/w8WnP+TkRcJs5pCPA8vW0B43KD8ghgmm7wj+IawXRxBr8KLg1e0Zm/UZPkgi2aGkYWdaurGlDw91wDxOSUOMtAPDsCXWG6K4QgiHcz2KGakvUVYzdi1R5shEQnWsQXhaW5MlknVTEQePFR7TwGBzSg1/8GLKYrZicxtIU8vNcc3W5Ax3NwhpwFvQgHRMpw/Y5jRSMAwMtiY4w6g0nRc0W0OsY2JiZPDM8gxp9qQkD8jmIuFdN3D0AiMdO5fy6leWp/OUT04FHz5d4NoIQYk3llwmzIpLtvVj9uZjzNmcfdvBRyPSv+f7N9+SiAvE27d0919SSElRxogoYf3l59SXJTJ+imkCetTc1e9IzxacrT5j01r6V99y9fmfs79/TXH6gserGZtdR105svMnuNww3N+gVUIte2wbOBwqsuyCiw9/yt567r/Y8vTkBUXpSHRLHTqcEdjBk8QrqtuY6l33sE3RCw7HmlZM+OT3f8SinPL5559z1zSczj/ih48+4rtf/pzj9o7V8zPy2QuSJCAGx4Y7JnpJkU8YFylmHEgOW/xmDTLDDp7Z8pRIPqfPP6Q8f0Q/+zVD939hdK+QqaM/DBiZMCsNklvahaM2FnEQzMtTWtGi84S6OZC5QH29RrpAlGVMVpfcf9Fw6D23d9+i5wcefRRzvetZ155+FmF0R6121PsBvKWcT2nVCDowOEtYKpgkNCvB4+cr4rOCzV/tyMeeZGxZ5tD0jqN7sDVda+nWB4q6Iw2BTg6c/eN/xOl8yfqP/wQZbVF9jdssSIcU096R3CTEj3/DYiAmBaUYBHgVYSPBF69eEZxBDBHTWUGalghZolRNysA0GUmiDdg9xivWXYSvC6JCsHKSMvYYNeBSjQ8PrPIYiRYG2TkUOdU4YQwgowPnM09iLSGsKYA0tnjZ8dd/1fP1nzjS6VNuuz2rjy74nX/8D/iL3tAnv6RJrjh5dAuz18wnAzI3ODESuQg1GtR+TxQFdB5hvGd27rnfddRDCtJT5jE727Op1shM0XY18Syj29fo2DEpFZO5pheC0V1Qd45xN8WGBfujZycFIq7IplPyacEHPyo4O1HQ3rJPI66+NURdIBYF774xEBdkJiFNJEIcKPKANmB+ZXgSP+Or17eoNyt+V18SnQWMj2Gr6b+v6JuBgOIY9pxPT3j0yQXKJmRnUH6bsfYbUAJ8x7reUldb0qgk1S3lQtF1kqpf0DlLkiQYM1LEmpSexlsO3pK7FEwgtJJhmNG3Eza395gqY/QR8SSirizff6c4XyoKEiQDqBwbLD2GMlEQBK4LjIeKNAhi5YnnmuPVSDI5ZTqds1vfkZcF+12PUBm4Fr1pydSUde/BFg8J6lTi4zmjV1gFeTmnb48Mxj0wLaIcFTRGHCmfTZF1wv5dR9dVKL8gXc3p11tEnNAFi4gKxn58WNd0AUNPkZdMdMH+LqMPMUFrmrrl/t0av7VMyJicp/R2R6ELvEyhbnj9xQgkFDi6XSDpSnQuGQ/3FNEz9p2lP0gOOiILgXjf0wwNQxUQpwtudht8NMGoQDxxJEfHxIMZd8RyQu9GmiHGhpREC4KPECrQtNeUGmTtmJuI0K8JiAfvfRNov7nj0dOYYZtw3e2Z55qJStDeUsaG5W+fU6jhgWzYHoldylm+gO8N273H7joWjy4IMuEwGP7q1+/427+0OPGKs8ea/+3/ccrHZweM2KPc7qHWKzxOaUarYQRvJd6DUDGmmVEdRrLM0ruaoe2wxwOno+eJTHiWzzECTssSMwoi1gi9x8oBpSJwBvxAUCWtixBxhpJ/H1AcBV4JklwinMMLRd/3RKmiay2qLUiN5GIyst83dLtTXr6vOBpD7ROms1OiaERHAyEaiNMJxrR0dqBMYpJUoH3OaAesf6AC5mVJFAnyOMN2DtyIloFIZnjxANXadR17z0Mg2FsMgrqdcvO6431jkaMmTqZEQXG6fIqvTnn3qiYvBbOiZHY649lsjv/9n3L95Tu6r3ZsN68YnMIoj0kdjQ+0t/cYV5EsEt5/+z0fFp9Qo9hzpLcwffwDTrNT7n75Z9z/8hX3s2teXDzi1MObtqJ78hgXUu7DNcmjBZujR48DRT+wOaxpunuyMSNKZrgxoW07ei1oao2OZox1z3S64O5+x6cff8b55YTb+7cs9BkurKj7joWeECdTdt+8Jg0pv7h6S331LcbegS2o1kdUPmNSFKT3O17/1S/53T/6PRpT8/rz16TrLe3f/oKZVXRHSTqdkc8Svr97w6OTKdkyRsp/zptXJ1y/vGLIcjoxYO8rkgRalzJJSoLYE7o9i7Tk2I0Im1C/rchCznSuyDON31bMihglRnosy1WGCnuy0BNLyLMWL3vqTmCiBzqpUoaqXSMt5EXO3jpUNqcdDeVixfXre4ZGEHcZ3jS4sUHGM05Xz7HuNb6pUf0BG/VEccxEQ/Xqe9KLnsP6Dt3viJH0d3uqm2uGmWO6kozmNzw6ZIVD6wLjJTYMeNfiXEdbjzgvGYQlMJBEgYSRcpozjvckUYXWLXkUU7cOczuS8VDnCSbgksC0bCnTHXG0ww0bprmiFAXDsee7V4GtLZmeOz5bjETHNb7f48YOJUYaUVEPc/TsY+q44Sf/+BnT04Kv/+5nZOp7PvqXB05/1KDOb4niiiSNUSGgZIYcpjQuRs6O/Pife4K6IgnnRO0L8jvLdFKQnywwSrKIB4ppTqMb+itLZAcWqwnT9BIb9xx373n5ssCHT1lfl5ijZxw6chVjJ56nP7jg/PKCwVmOY82vvq5p9gvMMDJRj/EI2us1J6dzoiJlt9mhLzRzN+dyPqFYZMzVyKpPef2q4+NHl6RC8t3dLZu7lkxm+LYhKhVpDOnqjGJ2yo6eHzw75fvb12zu7pmolMWTM8aFoa47bFwQ5wuUsMzLntz33OwhyjOiMOCCI1hDmkQkIlBLyX6EP//ZHdWhxM5jCDExJZNYs1WC9a1jVArTTah2MUEYQvAomeJVIChDOk+w9oB2EflEMZoK6QZkqhGpYjQdLhiS1QLvC+hSAi2RGHBDgxECIWIQOdZKgnYc+j1dCAgNeSJwlaf3R6TyGF8TZ0u87In1CjVdMcyOHA5HlGkRoSAYGLuBvj8gzEiUJPSux6WCWldYMUDI6CNNZx7ChRMvWIhL/Lgm9iPi0NMcO3yU44TE2JJaNvTWMDETrLHoZIIwlswPqH4k6SBRKT5SeDlguSGZxXRjxDdXew6tRpASpzmD68kxzGO4rWo6ZVFYmmHAhIBIFVooCgTq6JGlolt3JIMhjjzOC1zUUEQ50eCwO8PiZMKbY4pMUuzxiBgN2WXBbFkQhcBh32K6CNsLdl+8x48RdRVI2gzbA31N8B3oiDFkDCJm+83I//v/9p7/3Q88iV5jhEFECbFuEbkh1hFSFQTnQAlEtMD2KWGAmdDgJW4cOAuWmamZescmCNTkFKc8zlZkqwipSoKICVYilSb4iDhNMEojIkEIhkimFHHMZJaho4448pjekChNohQ6F7xIE2aqIXY9iVUcDxnf3WyR00BvKhSOXCukAiM1deeoKkM5WbKrexIRSHSO9RFd1aNkihkcdhzRKFSI6ccRpTXSH0l1znFnGboGkVh0HhNnOfXY0bUdwg2YeYF0DmF6iBf82Tc1x4MkCwVTUWNffoOJFMlixcXqhOWnT3ly/gm3+Zy3479nc/09Ks24/Cjj3avv0VKyyidU0ZLq6oo4jSFbcjxcIdmS1z2iiSjMDL8b2W2+ws0ceZlz8+pbTIgf7KlkiossaenRIQKRU+hTKnOHW1TUqkbQEVwgySeY0dInI4erO7wwFI9PWG9GHn3yO0gTc9wPHN6/I806EimIpik31WsSC+ky4od/8I+YXeaIc8HTDy/phpHn+3/I+7/5Nf/f/9ef84OPn/H48il/+83XRMbgUkkvBmZhJNnccLZKqW9+ztvNey4vHsO4Iykr7GqOsJJ6SBk7QRJABUGpNUJ2+HEgm0t22yPRKMiyCeePnjDs1/S1YaYyZpmi7yTyytNeg99KymzJ4r6kso76fU9+7kmFwDRH5rEjnUuQHYPxBFODmFBXjuPtHpsU9H5A1wLnEkZb8eyyZ5WMdKql7itsHpGm4LcV8m8+53CekEnDb/2Tf8l3zyr+7s/+gpu/fMnsDz+j/KPf4vjr8TcsBpxh7Hr2wiETwTSJkdaTyIjN7ogZFNJJpLQkcUqejejYoXVPllhSLZnsGlbjhGd6yuLYUYsEM5WcLAamaQs0+DSg/YD0niw+Q/iUIE5xYUsUd6gsgbDAW0cUBoTV2LDi9/7H/4Lpo0CU3nF/3DCYe0L6l5w8GogLB6Jg/W7K6vycrMy4fRdzuDvlcCcIHFjOHHphuXr9lH7/iF4OjF3LsYI2TtEhxQ4xV9dvOckvOS9PUWNFY2pEHmP7OX/3i5G//uUb5pHiRE+pDt9iZoqTp6dcPjmn2h35/nbLuvYMXYTUF0hTkfYNURqhFLihZnc8YpcZyTTC+glmvqJKJeOw48s//yVnj0pOP015+/I9N/sOIxYY2/HRD1YkWU4zCNys4Ko/QnPgB9Mz3v7NDXU7UqYZrhmR8wKpVoyhpxkVqaq5WFakYuTr1xOEzUm1JI0Vh+M9Ks7Q5KgwIqVh32l6W+B7Tb0RxIOgrwy7KsHaGS0tIpG0nSXoCKkFgYGAIJIRbnSIAWIEIQmI3EJ7JPXnTBYX1Cqg44jBwzAGiARpDGNjETZGJhFZqeiaGm8n+GNO0x7wAeZlAuZIkkKwYI0gi2KkbOlty/71AbF2BBRCpASdIsYaUe8hEsTnK7yyJJHABUUrJY1rSVTNmCxIznLa+wEED6uZh4ZUREzTlNtNg681g3BksUamUJwW1Nf3FErR3e6I5iW2bgniAU+tpMKbnrFSBOWZKsHsxYLo8Yxv/nJLMuboIqIfLQqwYYTgyacRLT0i1JCOWD1gtSR4R+gEykQMxuBTxVikdNKi8pzWggTunIWdwxQjcemp2wNlrKl3nu2949Wf/IJ5GqGAJJnSHY7Y+4Y8m3B6mhNvDywyBWFAaIsSgroPTNOCwyD41c8OfP4XIz/9/RQfDmSpQ2uLDxInLEkyJdQDQZQgFuSrmPz9wGOfYqMZx+qO/V/eIzYbGue58pC+eMLih6cP2OHLCBklBF/QNjVlnBKJiNAJRFCMg4E0oieQaMUkyfjk+WOeT+bYrqXrG272t+AC/eh5+XbPydwSoTC+JOiB0YAZLVHkqI8bBCMmBPZ2JGiB70YmWQleoVKFaw2DBesGFmlJFCW4bsCMgUxlRJFCjJ5+qAgahDLoSBBEQEQJY9/jR8/z4mFGV2swbuB24/jmnSNoiJVjfVxjtCTNzrDrPTddQ3z9lvNDynk2wWtFvLrg+b/8Cf/sX/0Tvvv6W/67//v/le72NZELFGVEVd8htpY4GZAqYUhy5LOCeN9zGaXU24oxWMZdQxl7Wn8k2Bw3GRA+wcU5LhLsxtfQb8nPjpSTgSG7QaKY6SdkMmZ9daBqe1wkmT+SXH//F/RbiVe/w9547P0aAnzXXpM/ucTZETu2JJngh//0H/PZP/t9elHT7G8QduTdly+R71Puvr1mf6yYyTXCvmc4ViQ//i2i2SknJYwusFMDh1c3DKrHzyfcvfmeR+LX6LAnLT5j6GP8sWbqYrzfk1jDmT7F+hbjLbMoJzERvj/i2wP0T1iuPuHdbkP16oapGsiLiNGXRAdB2kM/5By+DAgx44mIMe/vmU0EVQ+T8yX9fUWZLhh6iTWCOgx8c/0aaSJgB1mCbztSEYjsSNK8YsmRYZqw2Vn8POPJTx7x5tdbRHbg6cmSpDljsfqYabNHuF9x6Gsupgt0Ixm737AYECisD8gUJnmC8jVRbFikipPiFE3CJNYY32FdjdaGJG5RskLKEU/HfJrgohmruqA0llEYknLJrIxIo5Fh6JFKggeCJAqaOJswtAm6cbhuh3AWqQuizBISyIvnuF8UnD2LEZMdUh2R77/GmWuSlSZNJ9RvAzdvVtzdpLx7/ITVjx7zd396T2RhPvXM5it6F/GL/7zj1+9ymq6haQ02LnCZJJaCVTSymhiObxyx75Fyw4vnCy4+vuTbrze0tac8X6JsTeYbfGiJ8xwTJRyagj/5D18jjCVfxVxOpthyRCYK4zTeN0gRo72kqWp0VlATEXcp3161fPP+mtMXM85jy/T5jMXjiJ3bYROHlJLWjyxOEj7957/Ld1+8IZiew+aWemz45Lc/5eqq5vq7isvHz0hVwnY8sL8RRDpDzgeSSKC7isIqdHaOTBQ4zXDosGLA2pEgIoT2KG/J4wQVTXCRxAVNZwWkJbs6sB08DotQGSFWOMA5jbIRufh7uzUZqesd7hATzIiSllzk5IlH+sAw9kTLGREWe2gIIUWogJKWelRk5YJ0kRGkY6xGgm/woX+Y+bQ9pnaU2QwhLU3TE+UznIfWDUiRMe46XLMmn5Qkg6EdG9zQE0yLMhLpRlafTcjoqQ6GYy1QtkRIxd56YjXgyhjfDSwuprDb4YMnFoE8EqjBQNTh8Mgo5ri1OFEwPy25b9Z044iPLaIX9IcaLxKC8yALjIJ2KKn2mmO/RTiJSz3OVXgTYZOOOFG4vqf3nvi0JI+nHNqbBwtAa1zfMxx6inhFJAuOWc+YgfGecWwoslMUgh5NbWe4PiWNBPtjg9ULrJeEuMCLmOvbI5GKiJKWou2ZS00YOpquJSEm2lpcIzF5xk6MOO8xuzV9LyhdyrtfeX7845IkrEnTAaXAG4/UDoFFcE4Y5tCXXMQRP/u717z/7i2xthzud6h7ix49ey3wKuPq/YG19Ihdy//qJy+w4oizjqxM8apDEtCDJopigpcMZkDFGaEWON8x9h450UziiFl5yrG+52h33Iwdle3ZbwRhWPDluz0iUYymJ8kypBwhiumrFi9BRRDn+oEyKC3WGgYrOOz2xDpFCcskLQjCczzWZDowLXMEI045vBAYa0gjQZ4mSCnRQpCpFBEJJnFBV3V0zpFNLrm6S4jiKSrLkCIlkhGqUESxYxg8Q2cIfuR4GxA1jLZmV408ic747372mizJ+cGH/5JnH8359s0rhpuKqDLs71rUWcT87Aw9OcEtWkTaUFRr7KTg9PmP2QVF1zZE2/fkxwShCo6yoG0E8YUkpSLJLdO9IO0DrdlQWcXyxQl9v2fc13AdQDmKacS88Bg/sP/yL9EnC6bakRRLtn3AJ5DlOf3QcjoruHv/c+yfHSCdMG7v+GL9n9h8d0c6CEYp6PB8+e4dSTNSlCX6xWOii6c8/+mP2PoDomgZX72jvvo1H/7wYw7v/xzWa0qnSUxOvWnIv7tjFkrUokRqKPQS93jB9nBkb3uK2RlNvabUKbsvbig/PePi4x+xjSO67XvyLiIVini2oE0b5mNDcfQk5YyPLj7meD9Bvj5wcIpwjLjMHhENGlmmyFWgZkc/7EkLSHyO7guSE4m/f8d4rMiPBWc+4nrTMYYJ5aOS9EwyPSqstNQnlpt3t9Tdz9jd71mdHqmDZN/fka0XeLX7zYoBpR7O+mUsKaIG09VE3pEnCdI+hNuEOUA8AB0iakjTmkiOEFqMgyTdc37SoNaaVEWUacnGCIRvEW73cJb2muBBegHDiHYj+JG27ognS2Rf4eqBSHiQBq8VqihYzGO2Q8+3r/8a625JV3OOhzNevfb45jHHm4j9poXecq8MZ5/+Ns+XGnO45vrmnp/95w2bfUIdGzpjME6gRCDIgHMKG2vev7smjha0MsIbw49X53zzsuLrl4EoJEyXER98eIrfSI7rI6IcMUOguj9QnsyYZBkwcFw3lNqSG0FIItY3I3nmET3IwSHLnk9+9CnN2NKsK+anF7R9j5hf4FTGdb3FOk99DMgspyxKLj9ccbNpOBwtPhh8e0TuOjZf3PPStVjniS8SXGtpjxLjFT6tOHkU8YMXl6z//CX7r2rU0zliJhhuG4T0BA1BK8ZgQVgCjtF6SFcPSXsE21ogKPjiuzWHJiOOBQTFYD1RUDgv0MIjBKQ2cD47oe4DJksZGJESrOuomg2RqBjHFHP0XBiYY9nsbtBigo1gMslJZxJhe8RxgFGAsUSlRpHS7vboWOELRW88Qk6QMqZcnvDq7RUSR5w1lEngZDGntQM7M2CFIo4yCp2w3w1EekXs12TJgDwqtEvI4wKLwyAosinN1S0YR986HCWcLjA3N6wyh4tHjLUMo+MwCGSeIrR9uBYkK2Rs0Kmm9h31sSGK1QPOVGqcjXn/1RpzXHNaXLKLFSpKaWvL4tE5senZmJY4npLFBYdxQ00MQjAQsZgkuOtbhhEaZWjEgG86pEqIzAO0p3cDaRqz7hxSB4g8YhDsuhqSiHHc4a0jDv6BpjeM0A6EMSKEgB8CrfTchgMqH0gvc1anKcvJhCyeECenTHMQ4puHsGgmUDogVIRwBpnkhLCirT/AmBnBTbj7QtL+yRZ/aFFl4OACkyxjCIIxCE6exXz2hyf84//yj3i9vqM+rPFKkyQBFTyXqycMRvD2fUy0VcRBY1xMpnIGWbMoNPV25M245p/+7hO6tuLJ4wXb48i63mClYOfm3G5mXPcDXTiik4COAqMZaE2LSh8orQJFmiRIP9J0h4ftiSCRkcL3MFEpqoPGK2bZCUM/0tQDRapxJhCUJxGK+WRCkWcM/YhrHKtogpQS4xJuGphOznj93mHiBalK8X7kZFkiOktdDwwG0mlBqhKaY8fd9T3ODyyeaLzoefv2FbPH/5Qv7+/RrUFctRy7Epe0nHz8guz1BnPTcxJl7O63bHXFeTRBXv6A2Ykke37G6uycrz9/R/3qgup2i13XEFsKJcmtJR8hGVPUdo01e2yRgDrFzQ3DdcPU5VRDjbIap1NsLSg0VNHAfnNPkCX7m7eYIUEogRshkjnX6wN5eyDcdkxf/JRGLlhXW8Jc0lQdSkyYRhlxpmjslroLnPU33H5zy2a9xp/CR3/wh5w8e0Ey/5h91fHNF19wwccsTwXG10S0REWL8hF6viRNT0nLZ3jjeXzquF9fo9JzBJqi6rh+84q/uv3vmT59yo//yb8B9Yfsd1dU+29o3+x4fH7C9uZLMpsT5Mjm5S/IY0VkPadqiu0jzk8mHN09igjtYmIK7CjJ4wwXGYZ7S+Iscq3J1ZRMCLwYEQiCVJw9e0TS50z7PXfjQBIvGB89QM0af4d/NmALwX71mmN9i8r9b1YMhMESRsk8FUyLDhc5IgclHl0rzH0FjIR4JD7Pwd8RcYMOG6z3QEwiB0a2pMkcrMG7jrGJ6A4N+aJDYbF4lIpJxAQvIsqko/QtNzdbRDqFYY1KDSHqEVn4e+iI5bvv3vHVq1fcHmd0+ynbVtAdC5LJitGl+P01GSOXHyy5P458/otf8pfBUZaS426kr1KSSDItEzphkCLGu4HlZfowNXp1TR91LM9PiPQUF4786tUXXL0fSfUZyaxgmjZM00Ada6arOUOtqKxHiJT2+FCrfHRySRsdObz/BtvX6KEgmy/QUUo2z3A3IzIpObzdsB+O5NMYM2wwWrCXK15/b/EHjUAhtwMfrp5CYnh9d8fr9xbhc4YQM7YZammJTnoK54hmM1QJ99WBbhwIecWHH6x4/HSB0oZxMWOuE65cha8CiSvRcYQqU7z1GAQqUjQOBhTbThN0TNu2vH15hOiE7TFBqAjjLBpIkCjlcfKhYtWTMgwGfR9oKkUSR7jUkqUa9h3CZiiREicRSmm6N9col8MxYMTA6vGcSSlxhwP1uwo/SiKpQIEM0FuLVCUMEm8cITnHaEkbGrpxxFjHREMsMo5ty7PJFI57bOdIyjNiUxFsix1G3n17x08/nlD3niLPmaoUIQQymXI4dNjmgLaWw97jZEIFHA4ti2hCIfeUT5bsbxqujh2tnpBEEYfbLWNvcaEhOEGyKCijmOOxwVtPEI5h6Ehdy9D0MADiiI4LonRG3fT0o8Kj2VUgIk29HWiNwScRngHjB7AOnYIfe3priZKAcAGFQriayDjioIhRSOuQvicaJbkSeOEYgmFsBgQDiShQQqDoUbIl6BilU3rZUa7g3/yPfpcnZyUff3DO5ZM5Oo3ww5z/579/Rap6onH9MDSVOASBICLkNAcesX/3Kd9/8wMO7yH0O7av9sRpR55L4vMJLT2N68knJYxrPvsnKz74ZxKT/zHzyyVvdo6DXaJtIBWCsE/Z70Z+9auasVkymQlSKfFxQid6ilQiQ8tklnIyP+GPv3/HdnuHNEdSDMtpznoo+eZ6ZAiONJdYWzMIR4gMkQJnAqMdUBp88OhYk6qSuhpRQ8QqmTKYwONiSqwL3u9HdJHjVUUsHME4IuURChIdI53C9xG4iL6vSbORhILN7khAUQXJvjckSY8IlnmWErVr5lIz3K6ZJitm0wUiOiG8KFHLOb5as91c8dFPP30In+pXfPjjD+mTGd1w5KLrUf6S0Pac/9Rx3Lzlu7/8krgOqDQn/9FHRM/PiZ+VDBPJzrRsXIucP0NNLwhNg79bM40GCqMojSQJGWKeMVwpynHKoYbNsOdUn2HTwHh2pLvqMJFEdAPTUmMjxUBKKiZE4x5i6PsjiVfodMIwW3L+yYc065q7wx1xrsjKhGL2GZmPuP/+CjrL2dlj1oNne3/Fow9nvLp6R7/fsypi6i8Muzbh+lBj84T0ye+yrx4xvygYhhbrbgjRERGXeCM5GsObt68puo6LxYQgNWJakjz6GNn+klXpSUPNYfiaL15O+ODH/wXqzLN8NGN1kZF2NYfvBJPpCjtanNOIBiKTggjYoaZ612NdSuckWa8oVit2V3f0SjORGuNhGFpOgiCTKaWDJLToEJMtE37/D/8Fbmegu8Ncg7jaI9SO6YVEtJrb+4K7nUYiIBakUfKbFQNdZ0BpIjGS0BJFI1JKksGRdIrC9ChhSIzkuPOUHyvmhcc78C7CC4EKLaiRNA14AypSBDK295bHT0tsX6G9QLqA2FqEbdGsKeIElUlEbBGpwWc52/2U168t395G/M0r+Pr1Nw9JXveM3X7kaBsmkymu8JwtMg6tJQkRzlbo6ZSxsuRyzm430vkcFweiORyOLcqOaGn5+MkKnXj+7usvqbYt5WxKPp/j4ojt3Z5632G3muisYXQDcS4JykGiefO24u3VQLqY0TQ1cSnoG8M4NyxPUkI7Z3u3RZUCVISe5rhDTxt6LIHKd0yzExZpwX535Pxywtn5lHkac1zfUG2/Z3A14Vxy/3ZDGzzeFUiRY4QkVikfvDjh2Umg6VvevR15/+t7jOw5eaF48cEZebmkblt87tHLBRcffcDNl3+L7A54C1G8oDUdcZpSJJqDqXECnCypjaBv90jpCK1j9C3CerwYiLUkuBHsgJCWWEYkcUE6Dox4tnXzUGgIA7IMxJFgDC1WOR6QRpJJWSCaG5IY8iLFOkVfH/BDxO56jRglzku8GLDeUe0tTdOwLJdoB/t9S/LiEfXQkXBGoqeI6ID3A4ejhVRhSsXyySOu39xhxoah3oD1JPGM9tbRP88IWUeERMQKOXraoeGwa3gSL+m1Z7Neo8qCg5a4wVPECRORolVAxxqjYUShR8PQHUiUZnQ9rU053m0ICFKpkK4jSiK898TBUzUjidSIOCDjhmAiEisI65ZBjMQhJhjLWNekcY31LV5ahB3BW7yGIbK4EEAEdCQJxqHj6OGSRwxygNiidUKkFQudcmxb2l5SaoGWkOAIsUfEgWk0I+xG0iLnt/7oD/nhBzOeXZak+UhgYG9j0k6QJJKPP54wEQsOB0ckviEKAQgIPISSdv0ZL+/+AXeLFdNzSRxt4Mf3yBcjT+PHqOmSZ/MNb67+lo8+OyUqM0Z9x1G+RI+eePKYyC8Qw4K6D4xRhjGKVsSMuUUpzehqUAPlNOckVjx+NENqwYvTFb07sD5sMKYnkp4iSggu5psv1wy+QKQKGUuSJGEMB4gtVdVi3UCsFc5a1GgBGEZP3yvSZIK0mkUWwTjiUfR09MOIkgFrDCL1REnEYDuqpkKGCGljsqwgzTW96Wn6I92xQwpNIgIjllhYYheRiBLpDavlJbmY8e0XX3C7/pofPP89lquEbdthi4LTj3+Inyoen15y9Yvv0U3N4vknVJTsuob12+8op3M++Pg5xcfPOJ+fc/M//Alp45gvL1l9eEJXKAgNaRrjuxYlLRLL4mzKxdPfYfvVXyE6T7CeMRxx3QGdC6ZPJd/++kDoHYlQuFlAv5DIdkAfBkqdETpP3K6Y5hlD6hirHgUUIWDCkSz26HzOj//BP+PQCt5/+5L2u/9Iqh3Kl4REgRkJR8Grq695/BPL5NMIV7RcHyyia8mKe47dFxyylGSe4w+Sjk+pzJT1L2rEeM1SBNLpCntWwpCBecS0qEnN5xxvA1sXKB+/YrUaiZbf0JdblIdM5PT252yuA7/90afcHBw3xwPqao2PZqSXM47X99hakGclxSql6jeMssPHKTJPcYMjyIZN3xLignHswYzMf/sFh6sWadekeqCYTWEQuCQichnufkRMXnLy4jWTpSWN9nSxZLs+QUvHyWnB91+ljHZCtIzxevGbFQNjBfHCEylP8JZItEgcoZfIWlMEgR8bShKGbQLHPWqxR4aWQIp1Du0TUnJ0ZBlEivNgROD+vkMWjsjsH9LmUkGSIP2cpIVUKOwm4fu7km6/5Lt3mj//PPC+KujkkmrUyM6CziiiBV5WfPjTF0h3RIQ7lsUGf6JImox215KeW9Lg8Tj8aIgQGDUyP3tCPDMcb+7JlGC1mPHm7ffQKAgziKYMwO7uHd1dg1k3aJYY6Ri3I8WjS+7f3vPu6p7toUXKCJW1fHJ+RhSXIBPeXW/pd3dkxnMyf0SxnOBmPWPTsn71FmEEaZsThkCcWXbbG6w0/PjTP2DEYcIeZQ150Kw++Iir9ZFgMlKZMEaaMQicCPRecjx47sKI8DHN7YDrPT/5h5/yO//oBV+9/Y6f/ceXiHiGLhQXsxVHFRFlC84fzaiMI0tyQt1jrace95BmCBKclRxqixtBW8Min9MLyagi+hCQcYTwhjzKSAYLD1P25FLSe0HTbji5VHTHiNAOIC3SJkhlEEISxXBo1lxOFkSZhlGyv6kJIWJfj4hohkpzhHDIYOj2e6wBRMZ+uyXXAmcsXbVHxjOCieh8y+XFB+zefUfTG7IyYcTR1hXyUJOXD0TBMEKiFCLN6Jxlfn7J1e2avu9xR0vTelyQtO09BQEvPA2BSniCijk6x0wXDHVgs+kJKiPXoJoN0jSkkynSD3RGYm2EiuXD33TTE9meGEMsIlSS07oRaeBUnuIjjVwccM0Rpy1lGnDxniLr8aoF6dCJRnpFJB7GvkRc0oYIF8PZfMVxu2d1cUpgZJJmKKUQ1hPhkUAap7z9dc3xPiOIjkgNzGZzVBbYUTO2A8JEtPuWvqr51a/v+PJLT0vAdo6L80dcXJzwwbni0WrG0DUMIYXkHOQcpMaOOVcvV5jjExL5jMRXJHhk8Jw9OaFwA+OmZvHkgunzBemnJeXSYSNBux4IQ42PRmIlSeWIHh/GxLxSGDnjvt0iFxFVV1N1I8/nc/7w9z9mmjusyvl3P3vF4fae3eYrxqHCpB2jHMlEzPs7R2U0QvQ4Y/FFxNAfiUqItSCMI1IGrDPEqSZ4gQ8BO4zEIsWOA4NzhCDJsoLBDSgR8MGjBFjv0EJQ1wdUHDDWEZwnSoDgEFbTjw1axZzMJ7ihox4rYp+wSlImSUapJ7RDj5aKDz6e88Hph9xVFX58yyRV7HqPjc7w84Q+CkRKIItAdXPPu/d3lBdn6KBJyxes9xve/vFfc7JakCUlH/2r/xmJ6/j8b/6OP3r0T4njBUFYjvsj9+vvcH1POTuhHzfc62tCvUaFgl70iHbPeDNyMV/SrbeUg2Y6UzyaSHabNW44kgfLvnIUkwl7E5hcaI6uwbgUiUINAuXAuRpnA9ZY/upP/gziKVnoOB0TQlNxdG85JvKhTpzEDFXN2eMJ1/2R220gNBmuG9ncdjz6h2e8294SOYPQgburv8TwhA+en6LDnqitGAdDu96DFiSF5cPHCVn9a/KJ4KIPxNEbiAxeCk71hLdfO3q9os01U10yBHjz1ZrjfcsZK1TSMdiOcpIzzmOybEbdbalHQy9yrMmJ+gmRs8i2IwwJs7MV+9uvOW53tGuDSgL58wu4v8fpAU4V8vGC4e5Ayp8xMxve3NWIThG8oR0yBplTDTHOWRbzhMpJykWOb91vVgxY05MGx6RIKHRM7BQyWOzeEEcJYRhQcQxOE8uYzZs1zz+UhKBAWCJlEAF0fIssHeMuxavHhGgkqADxgMwGUEuqw1NuthdcXU/47tuBt8OG93vBq3tH25dsj0uOrqTd74lzj28OpFLj1RwxmyM2A33T8ujyFHGA21fvGWzGvqn57MMLDtsO2+4o5hOiNOdQVUyzgLpds1vv2B8dLz55TDeOMBieX04JOqXtO+qr7+g2FWosCHGKJqPbOZJ5wvLRgtevNkiZcPFYc7y3nJ7Pma5yRu04rDc07yvaceDkk3NiBSozdMcd268OaLXAyAGpJeeXc26HNXXvmE5m/PLnv6SuK6JEYvsOzcgkDsispu73KJWjJiVpkoNW9AfL/r6hzFJCO3D28SWb7Q4ZK/70P3/NFy/XtNUUUs18NUMWM7aHCitikoXi9u6e467CjIY8nxFHOaNUOOFJlKSIJ9TDiJIOO4yMI6i4wKvAGCSRignKIEdQMqK3DVJKFDmx0HwwTbHDPV51jNJxuzZoqfDC4kaL1jHeaEYnyNOUqfJIB7IssN5ilCbVEdI0KHpsP5JNNO2mglo+tPqNZXZpSGVElAbMdiBJM7q6wnSwfnlkHCUpEYmVuCCweULrDRYYXULTjZhmoOsEGE8SRSSp4CTOWNyuqdQ9W9uQ+gsSmdIOR77b9ugbi4xyJrMCXzjU3sOhxw+G2fmSeDqh2o4YJDpSaC1A9JyfLZCNxFnD2LQkuaVc5OzbA2cXKX3V8PjjU6S3qMxyspqy2x7oe0uqs4c//s4zHA+IoqCdzflmc4Mm4pOPXpCdzbi/26DijN2xwjtwXUueRERjx37fMUsLZJzSWsPmsCE+OiaPSkKhyT+YgpDsbt7jBkPwivtmhx0k3+U9nX7HB+cT/vU/fMrlQjE7mRNl/wCrBq5vO/72391z88uGYnnDh791w/a7X7JJeqwcaVrN/at3TOue8uwl//z/8PHDjv14jwuCWIIXnkg58AeW0xwbKu7qhN1BMxrHdh+xqTw9hjzPmeQlfX1NFFKuDwPfXTVIeyBSDfMsJeicY9ezGYeHkSA5Q+oGnfAQOLQ1prGoEMhEjxcZow9EPiZOcqQNTPOMOM6o64BtJD4ExqDQOuIkUlS7irIo2EkwgyWLl1R1hekNkywnEhGZlMxFjitSqFKe2BVlKfmqqZgXCz7MF8RCo1RKkRecpANJsiO9HNEX0LkRFS2IbxxJMWNMUpz39N6SPF2hz6cMO8NXn3+Nrg/81g9+h+nkMVud40TEL379a+wPPuOzn3zEVJT8+Tevib79Kx6/WLJ4dM6HP/khb7+8JilOyScZQ3WD2VccjefkyYrQKsbYcTx2pDvH4/I5v/dv/gWl2vGrP/8c//2GVWJJLyOOi57Ra8rpDE3K+HbLs/MZ/caD1lStJIoMwQ/sXn6LilNqtaE/dORSosYE7xIG0ROLgclMsr8R1AdH+bjng8cl95sD19WezeuOkOUcKsd6dySYiIvTHi3fMHa3tDhiF3ESxYhI8vnX33GrLb/zgSC/HEC0dG0F/YxSfkRzDf2bFrU8oZjM6Fzg3c2X2O6OyTQj1ilRGRj3LceN5DgMmLAni9UDrMprlF7gZETvDb0dmLUedg3pKHFWoG9GfJmyyzUn88es31+j3JRNKXn2W5Iz/SvWX+3p3z0GmxB0zzjAre7Zd4q+g/R8AqMijjy7+vibFQOm7nFpxzIukeaeRO4xvsOMEuESPJbBBUY5g5DinUTGEu8tIlgi7ZEhQpzUBD3FRgU2lAwmZhznvHsVuHmf8PLdgs+/ueTN4RGbRmN7h84iirJgvQmY0eLFhChKyCea3NfoJMGawBg6nN4zivCQoL+uWKQJrl4SZEJ5suLmpuLduyuavSbLIFloJvkcZ75nujRcnhf86uWWQ7/mxJbM84faUF8diWyFPPaoQWOUxM1jTs8vePVqQzbPSYopLtUMXrO/rUhVSkyCcwpTD6jR4roNWjnS6IysFKjFyPbnB7p9IIkz+gSijxbUmWHYapAJN23HZrylzFPOlnMm+RNENzK4I8uPppSxYapKlJrTjgl3VUs9BPLTCc9+9HneoN8AAPqgSURBVJyqWlMHwdBKrjaBN3vDtp6TRClxBqE3vH93Txgq5vMZzju8janrgSQryKYrjLPYYEA6bOjoTYVODRcnE6Kmo77rCBF4ERGCIjhP4x3zYoIeYoSIidRI3ThMBNbc8XS5JjtV/PzXHWJYErRiOB5JJgVSBiIREVlF/XbHgoQkTjHHFqUlZIFQOLqxZnISYbc9OrWcT6a49y39ZkRYT6JB55JmaInK9KGBMo9QZmTYtIhJQr6QSDtiMeg4InaCYCTbtzXFJKD2HRMJYrSoLEU4zTQ2PH6RMjanpPeG+3pDKA5oBL0bwHviIqE3ax4tliRiRr1vsF0Fu4ppuSJyHpekCB8o84LF+RLTD4QkpVKeSCeUZ46zH+b4qwOyB69yrl7Bk5OIadqTdjBcR1y/a5Fq5OL0lGmuKcs5Bztgqi2P5ytsaIE9AUHnGtIo5+zRU969f0evY755uYa64VF8QpwLdKSo7z2H+z2nJwXpGIFMMHIgpANBKlwNCE0cS9owMLR7Qj/h1fcN/+2vXqEYefyDFU9OJR/OZ0Tpx/QmMH3+mllhuNt+zvb+DfPViuv7G7rGEMye1qSEo6bb7piUA8EeGccOaTsi0RIpA3KEsEOHOa4ruXsfs+87LCPWCB5fLJnMNI8XGabb883bjpdXhlBV2LhHa9h0NW3VYXXAxgGrYFJ6vBPc7+9RWqCUJxIKfCBPUoxPcQG0l0x0TJ5OkGOOHx3aGyKZcL5YMS9SGlNhbg88jmcki1O+rDua3rHtK5RULMoZU5kiXEfqAyfxEhk0U5uxCBHDredJMaMXEd2g8bGi9w1JIpiVHc9OHJv9hih4ZtMp+25E2pS28g901VyRlHPGDLZ9RYhznn72O3TDHd9v78iqG7y3xNMz5mcL7o93PJMfUDyfoZ+cYa9yPv/rX8A3V3z4g9/l6WcvOLqaYnKB+V6iH8fsru5wx8AsXzGeWJqrt8xHyfzkhHWj+Qf/m3/Ldpqz+z+/YtU1bC2Ipzl5VHLz3ciYRHz2+x/z/quv0CHHpYZVseL87JxxHrMeLbe7K8welMs5+aBgfddTJjWTk4qgJJWtkR86lnbH/PyMPPoQ/+uR+y+2HPc5w3Ekm2fYg6bpHGnSkImG6Sxh+6rhRE45DD3TkxlZMqXZdFylL7h83HK8v8ETGMYYMzljcAPy76/jrumIVxNMtyWb7Oh1T9N0BNPjD4qhT9HzAmU9pjsQJRHBRThnyARYC0meEpWa9/dv0HVHPoLrLU4ODKtA42NcERMdA0k/4ekPSuo3v+bqS4Wfz1k9tkTGMYo9ST1SX2ccqymFiInnmqFvUeNvOEAoggAezk7a1kT+BgmoJENPLDKJwDtknDJ2EucjZLqCSDLWknbQNNuSzd0pd+9O+LuXmu+3Pdeu43x2wv/wZcR99zGNX9EOU1w6YZgO9OOGmS8wh8DueKCvHOkkZjLRdE4wWkidxNYdE6VBeqLJgmAq2tDwZndNScSkSOgaxzh4hkNEkizwRHTKc6jv+bf/kx/yo08vCTZh+oN7/vJPv2JsGk5Xk4e6R5YzNPDtrzrGsSGewmRyQjnRzC4CcVazvjtw3LUcqh3ZNEXPIvr0SNe0pMWSyGkuz+f4xDApA+WJ5ptvbri/lTgLZTpSXk5JV0sOhwObuwYrwMQapVNqPIfDhuy4RvWerEjR6xg1ltRjw/F+z2AkvQKlNJOzFfvdSDemdLYDIWhEwuhbisKhrCEiEFqD1Ip0NiEtl3Shwm4tvQtonePThOOuxgooSk2iY8ZmTRoH0mQkKwS3h0CUF4jeIIUmhBGHROkCGVJG02KDQ/qE3FSczwuqqzXpoxl+15M0Hp/25F4ixYCIPIemZHQNWRwhnEcFQ4chX51Q2TWJGNGdo7WCqJyzu9qztFMmqsRGawa/Yzwo5sUKN8SYY4+KFHqqkeuRrM/pnEPMNPSeabTAWkfve8bjiO8CZ8mUyEvkoWUen3M41ggpEbuB+MUFWaZp83f49Z6JmlGPPSQ9pg7IRBHPFPSWZmvI53OOQ/vgF++v0NGM6ekcFfVI2+LiFtN2REox9hZdZkTKs69HVLFkvd5iOoeUA9+vDRePFrRtgzWOTz59yjgYug6+ennAJpYsLghyZEg7uq5HxR3utuVujHkjrhD+jqFvEZnAi4bpVJHkkp4j1qU4FZhNC0Qk6IaRbhjQU4U0QAJh5rB+JM4KHmVn7L8dcAJ0LIikQ1jF62/2XL+1vBYj/4v/8jP+0e+fs3pyQrJ4TzdeUb+dI0XGu/uItt4Tuylv3ikmlysWy4jBOmIGoMbYIyoaEMIRREqmTgnDY+6uC461JESCOM6JlOLTJ484HK65erfmZlOx3/XYAFEOjT1incDEAa8CSmsSLYiCQ6YjpuvRpUS7iDJL8G5EhZgiW1LVA1IrZknMFJioKVurWM2nTLoRc9AUakJGxi/+3ecsokD5eEUQI4kDpQoenK+aXCZM+5xZekIkO5YhpdtZhr++ZnP1juWHKy5+/IR9NsOnkoGBqJjjfM1yprDN92TqPZk0aHlGczxBmkt07pgET4ai3b5Fhoi7rxsOg2Kan5DGmuK3f8BkkvDd53/KdLLj9y4+ps5TbADTDbz95d/CYDn58SfcX7/m1Rd/xeXzP2AywPv3f8fu9TVnakJYKW5CzWbfUm4UkY+oh4HNV1/xSgycfeC5yDXp6RPyfkDdjVyWT9idKbatQ2jF6fkFVd0gVppZUlKsPiQvCv7mV/+B59mS8TDhMI7kkz1Cb5l/0LD4qIHVln1YcfumZnpmieM9bex5+7ojyAOrS0noDVmiEFQ8LWdcNYrdesNicUraS+I8pjcD08clgy6gGcj0SL8f8FYS64TRObrQcxxv8GIgfaxp44GdPdBvtvhQoW2KF4LszJEuFfsO9keL3e54+mJGdwMyaJblhOqwpwhTZFGwmM7ZtQ6pJC6qULFgrDqyhef5H5Ts37wkEwHVKPY/3yK/1mzyjnY/Y/ZTmFz2NNcW0UJmHJnUHEZPte4RR4MuArnLfrNiIEoSGD2H7gThOhglzhYc9ylhmGFHzzgMDN1I2zqSbI79j5rbbUJ1SNm2M26PZ7y5W3B/nDPqJeZQgd2zMRqpVvi4YHCO2DvOFin3psa0R4wPBC9JopRBGKIQ0e5atPKQlTjhWT4+oW3XNK/2zGcl8zIlrgPWjQjVEcseKSpcGElyzXSqOHZ30C1x0jO9uOTdYWB7vaGqWkTfYVnSDIK97rl5+Y50bFlOlyTKcbfvWIv36ERyFi9o5Y6vv/wLfCn45PfPUCpidBFNX2OrHuSBrmtB1zx58ZgynfL1y+94/fKeYnpOMilY5Ce0tuO7797TDS1mcDx6+iG3bYtPUoxvaLo9SR44Wa0Yezi8syhV0DQtwRomC83HH3zIKBR1e+D6mzWZSvDVPYfdkckk5ySbsT/cIBnI45wsz6FzhMFQiXsOVFSHFkFJ30fsqg4RSVazKa158BCFkngz4seBtNCcPl3x6srTd4ZUSyItCA5GY2najpAKVmkBoqdt9/RdwcGuoDmQnkfk85xxiDBjRwgJm4NjXBrETBGtIrq2JhoyXDWizDsehQ3P7JKhqbmXnuMuRTdTmJyyv3vHPJ8iTY+3gmBi7GDQSuKCIYtLQmZRWjLuG2RS4LxjGAxYgxICLTRjF+j3hrNkRRAZQiRoLRA+YTg62oMkyIgknvLouaS7l5xcntG/ucNJzTgOzPWSIrvgXo30xzXBgRSBPEuYJxGz+QTnZ5j9LeEoaRuFdQOdsUwnCdYb3rxsyZYpi6dLJgw44QnDSJIIXr+9Z/1GU35a4saKbjOQZUvk8gHM0xpPXCbkZUkx11TDPc3djpaWYd8w9IZlkvPxZMWqzOidYxwdm/Wa7FQwT+Z4NSClZaETynnG/faaut0zT2aU2RlNlWFbSMsYH3mmccZY7XAiIlE5h0PNuol4/3/6Cy6ykf/1/z7hn/zPd0wnGzLd0diaDy82uBDwqqS/TTGmwSWBpluDPCDoUKpHEB7aPP4Fu92PePN2ya6JsbJlsD0egTOOn399RTcccN5z3D9wCZxyRLnC9ZZBBWKdMo4BPwYKHYEXdK3HDg5TWVSUYJVDSMXxZk86mVHKBDVasvAQjutGT9XETKYJne+I5gUukkBOWVywXd9x++tbXnyQ8PyHT/nm+JZsojjL54TRkycZc7/kTAeGdzd8/6ffMK07inHEf32P3Ut+73/5CV11i7NH/Fowrmq6dkeUNEjRI/wB0XXMooREz+h60ESUqWZaCqre8HgxobtquLyccbN/mAqWboqMVtxvDiSTA9Y64oNkOL5kd/cth2NHHQ5cFDPmccov/vi/Z1aOZHpPtHZUnJDFManOkSElKjSXj2PCyyvu7vfc7u94+e47/vXv/x63iUcjESFiuljyX/xX/1P+m//mT/nV/Q3HieNf/tf/NZvtjq6q+PrPvuVpd0HoMq6v10TG8NFn9/zwH77CZND6HjttqLXBjYr8dEqaGbw1XL23HPo9T38gYDtw4Sf0hwFnDrSbHu0XEApuX/XYpMQaSxHDIVEUqwLVjaQuwY8PtNwkT6j2B4KJaXZHjm88SkrE1DI6j1CK6z5Cup58UuC9ZR5HNBhqmSBtzGL1mJQZx/qAw5GpDEWBnAj0SUpyv8MkBrTGHC1BJCjvMdsRIRa4PhBpS5o5Uhtwu4fsjestcSSoog2TVckqleyUY9vt8HZEJAV5nkD9Gx4denW9Rkc5v/7/XNGOlno8p2liovgRxj0khf3QIbuOXEdMkg94NYeOBnscmK8+ZD0WvDy0uDBF+ohIRbjOYr3EJynTYkqOwNQ7YrtF7Q/kXsJwZDZbYlUgm+WIAPVmg5IRqV4gEXRdzeHuSCdKPA3DoaXIFCeXBenlklIC2y3Kgc4vGCpJumt4ux9wkebm1UBQPZ//3R3CCmQ/o48zZqXm/rjl0EtUMSH5ROK3EX5v8VKwP8qHoZM4kBcpFz9+jkoDx+2Wm9f3iDFmaAxNvyFIyx/9i59ye1vxFz/7jmG0lOmSD58/4uXLNe9utgipMcJjXM/lk0s0jm7fEGWgkp6JSjnRGe19Rd+PhM7jY5jOYyarEx4/PkEmE95VFdvrjjha4FRMX2cEDw2GyDfMTmdkImWUA33fkkUp3WGLaxQDitxJOtE+eLR5ynQy57A90PUBax4uRV5ItIhQBJwZqLctSZ6Sp5IgPFoVhBSGqkV3Gh8N/OT3B+LMUh1fU5Bh+i2z00vq+we+xW6Xc3M74KOS1fkUP4loe0PvI3wo0L7ho3nNeddiNz19GDhdzbjtDtxoQT95eKm7LiL1OcdDg0grtIwZbI9TMeiEZOnpux3FZSDMB24OHXmxIuoFqYZklmOUpLk/sB08cSro+goXzYgySZlpFicZ14fu7/MRA2pVsb2tOAlnhDLi/fHIMUhsFSNUII41o5ZEE83i0ZxIS8z4DtfmTGWGGTvGbsTiSGNJLnqKJOfQCNyNJeQdxTLlbtzy+PEl39+85/21psgeOv5ONAgNWzMQyxwzBoYmZtP21GJH9WakGxui1BOVgayYYFzNMo6QQ8/y9JRDvaWvOqZnZyx/OCMpItZv3uC6HXEao3PPMkgyCvIk5+rGsL8HrVd0TYUaB0Ji6bcNfSMxvmEMgsgr3NFjTMN3X275L/6rhtF9R2cDThwY5YEgQEUpZ/MnCDMjzxY09RadxwjZYwhYX3K3mfH2asEXbyzrqsfGPcF74ljTuh4rHNeHO8xgcNIjE4PwjiSVD3blGOG8oWl6gpAkKiaPUjZ9SycAVaAiT5rElIWiqx2JTsh0TBkVNKKFccBaQRNZTBioDg1pnKNEhNeKO1GR/94pSVM+zHkvCu7ELXqlyVVGrktOFifsbxuicgYqcPj2FWlySjs2BGvpB7DXLW///c+Jmgbbv+Pi8SXJs5jJjyBbeYSdooae4EfKFGLviSuPTwxNN5LZwPhtT/23HafLSz5M5xxw2HbA6o7Z7ITBJPz68y+JopbnacwnsxRzcsbw2VPuDltsVnJ1f8vo7qi3FWlimccF6XTBdr0jdoZVMsNgOX36MbJ5w/ZqTSFjmn6gGxXVLjCrLLgjTRg5HhoIjmeTEybJjPtvX0IcM9zf4tt7ru7eI0zH/PJDru+/4emnNcXylo2IULGm9xVpmuHvKtxuSuMlk2XK0PZEncaPET4RlEpSHVri2QJ/L0mbGRMxRWDItSBPFC5yIBMYBqYqRTFiTMIwJDQ6pbOGIE8Z6x5Dh0ktzWhobhTS18xWM9Lc44JgfQhUuSIYR20gzQp+8d09s6RAyiVVt0UlgdXFYzbNntevj/zWDz4me1Lxiy9/xbC3TOKYdIT2lSC48qFJ0N/CtudYzLk4f0EwI83G8jZAppeMOPI52G86JkOKlhDlAms3DEP6mxUDss+YnDzi6lBhw4y3txaVrsjUimN1R1ZOEDolOEVUFhxFTCY884szRn3Pq/fvufqlxPQ55WKGZMTSU85KnB+ZLpcsT1ZsXr/mfLoipsXfNJxOF7iopdQSHUls1FHVtyjhmU4fIXTB4FpiFaPihGlREuWOi8slUTzl6zfv6A6Gsfd4GyODZTKreHz+hOWLE6Jv3yCF45eff8HQXhHPFwQEVhuSLMNpwe2mRgpFdqG4+MGE968b7Dee2dkzZHLCod+Takk39uSV4nRxRlR0TE81prMc/BpnLR9+8JT9uuKXv7xmVDl62rL6MKcza7zcMg6go4xJPmG1eI6ygqFdM0t7sik8+fCU6u0GqpHF4xPiomDYrllMCiyKd682vPz+luUZ3F1dwRBhXM04NJiuxQND27KKz5FGMkjDdr8mTwtkkSAUjHVHcX5OPNtiZhFdNkHGktv7LcIpvNT0vicRltXZijiLuHn/lt3akzMhCZKotxBFZFGJEgqlBMfOYIXnt7yCYeSsWGKV4X4DNtky/TTlsNHsNhF2mqL1HKlixtDRu/DwrAVKOSZpAXct95VgbxJOy5w07vnofM5NsNRyRnvdELkEOzrGwxa/ctisJSoMspAgBlTXoUSGjyLERtP3lmKl8VHH5HlOyCLcdoMyGhcAO6JVhC4kT55MmJYV19pg1j0/enqKWEn+3f/jPVmUoouEdNzh3UC9fUVSQqE1IlswfTLDDCm77w54L8kyw6ArOtsTqSnTbEbiRsT+SNdEKJnj3UDnIrZjionm3KmYr18KqusTosjw3lQskohYT1HTFJUILj4pqb95Rb3vcSKmFwnoCdNYQn/Eh5FlkvFbn35AvzFoGXMyOSPyFd0yI8oKZAxPnj+ieV8TKUMSDMuTMxLX8e13NX/7n1qgIEtADB4/duSJQjSeYC1RFCFiie4HtJHYXnD7ZkN/HIiKHSJuKZRDjgHrwAdD5CoSYr7/VnO9O8OnKa2x7A4xmzplvVUMVjOMChXl4FvAEycpu2PPYFviLJBGMT44jLXEWQzB4gGLRMqUwbRMyhzvHFVbgwAZRuxomOSKi0VOkAHbjTxaPeVyeYEIEWG/Z3U+JYlTvrg6MFcFp5NzBD2pjGicY923OB0TJgEtEho1clJOiIxhcBl9G3CxJ04kaqZZ71qys+c0Hxz5+JMVt+/es8hKPv70gt3LNzQby/Am4duvr2hfwaPfecbqUcA6R5HmxJlivNdM7wWzrqQrPdEkY/96z8v/9gvGXlJND/zx9REuMubhOfftS+rdO1ZB8WwxxYSO7v4bel8SyRPys0c0quDq7h15PiMtl+y/qlhePuHk2ZzeDMz8CWNnuLvZMPY9345XfGQb0jJi3DZ89OzHVIcdTlRU44ZIBq7ffMvqzcBJccf1fcnf/MwSuYHUOMb+GiEdzTFG21O2hwMnj54zVF+wHyQyKWj2R6LZQ9g4m8CJTbF9TQgFoT0wiwJF1DJdFdR76OOMY0jwScws0Ty+zJiuJhhzT1/vaOuUKCQUuiY7Nei8ZdQ9GzlwU3maKkB6g3tqyT+wRGlG95Uhy+akmSaEhsFJghkgijk2HVky4sWA1xlNZIhw5NOY2UnEWXqCO+4Zvt8QVMT/n7X/WrZlTc8zsec36c1w089l9tq2ClUFFApNgCLRpNTd7AiFxD5QhC5BOtVF6BZ0CzpUREsKtWEHJXY3ARIACyiU2VXbLj/dmMOmz/yNDhb7HAd1CTkiR+b7f/l+z/P1lzestGVuYmz2oWu3OA+x23tCG1ACUS3oxpA7C5c/mVG+ekn5kUIWASrUaO3Zvq8Jq4giCDFjyP1vJHG6IFTh7zcMMO/woeB4PzH6EW8mkDXj6LGmYWolCoFpDbW9JU0HXv66Q9yHJKlmHDTxsiC8HTHNmjhLPzRplaDtD2zfPxAYiTOOtqtpXUWkJpLYc3r5gjgO6ZsGKfcc335FMZ0ho5HwxOOspmoGRglns4AgcxzevidfSOJ8RtP3pFlJZwT1MLF/PHDY37NKP/zwkbQIf+DHX1xRnK447jraRhA1ktffvSaJLNli4Cc/ueZ0IfjqqxGdzWnaATs+0I8D/kKRpXNu1mvcScFgAqL5JUNSUSwSGAY2x4r1V0cCvSIsU5SWKOvwFi5fXLC2d2RCgLE07Q4fSrw6cvU849mfvmAyhttfbHEbyeg9kfvAXz/Ud1RHQ6qXqFhRDw+UsiddFqjQEiUJU6Qwk+HQtEQKRt8zDB5DCDpgEiOd+TAmVX2PsJJ5eUIrYjZ1g3cQ6ohBgowS4vBI19zw3fsaN3WMk0aFHjEOhD7H2P5DGa4bEaNHThPEAu8v8Eaz2Y2kuiGNnrMzPamS9B7UU8+4C+lESDbXLNIz3PiANBGV9MSrkt2+wkcrHlXHIxl3h5iprskrgygc9iiJ5rMPI28kjXkgn2ckucQZQ902zPIUN2loM46NYvSOqdvAUbKY56hhoNqvyeKQceg5zc7pIoPMKuazgjLqsXZHu1d8/vQPSNKOXbcFr5H5CicMy8WcdmzRasS3kJydsXUN+zH8ILHxCrSiMwI5RXRThMlm7IYDVzGEWELp6f0OvVBIqanCkYNp2NwfeVhbnCv5gOwskF6Txg15oRBZxq7eMo8LytMQ+WTOL37zmlF8eLDkoSKKCz46XZH3AZu7jkHWWAnWD4SZ4vb9SHayIB4bQpFQ7VpCYrZ3FX03cP+oGHtNqDXG10TeAQNu7MmiEfSAGCWy9ohRI03OYDtmpccNjzjZopiwBkL1gS9ipSFUlrFTHOuYZnhO3bU8DiN3dcx650mihCwypOkHIyVC0jtL26wZTIMNHLtaokJBGEisl3RNT5hqxnqiNx8KValKoZ1IVUgsYwQDMhiY5xFZEOCkpxpgrkryoMDIhH6vUaog9mfY3sNgyQJP4SzeWOQ04nvDTHraXpLFKZf5jNUsISdnGxzYBgMzHaK05K4fEDEEVym7ZOBn/+I/4w/KExp7ZG0c1f6W4ErwSfwDomPDt//2a6pvXmHcnFUyZ9AbtA3wVrK+Sfjdf3eLPd4yecH88yXvf/c1rjsgLzJ0mVBlWwgCbm92TEJQV29wxrG+yxHTxFnu2NLz9v4V++O/oVxdo/Z7zs8vqYjpg4z13iGWhrHtsCYgiSQyhqgZGe47+uIMVVj8Q8v7f/c1n/7sR3z8x3/Gd9+8ZRomktGymA805x3Puyu+f1mhFoZ9fYPvBk5nzxhmCTqbMR6ODM2R/mh4drnERJ77Nx/a82Eco1WKyB2+y2j3Ge0uAOv55EWJMQMPG8uwSXCR5iKNKT8VrJYV1XpE7CqmgyGiYPWDlnR5x+p04tDu6IDNYHnsYvwCokhi0BgxYm1IFIfMzmG0EzIKUZHAb/f4RhGnBWfxjDAaUHLkdFmQZgn1cECqBh13bL4zlOEp26pi50b2E+hxzyyVxFPIcIjoKsezy4xwuiE6VIQm43GcCH/U8uRnd7TyG6rhQ8k6MBlBGRNfGaJnEd0t3FQr9LMIe2x+v2EgdZr1ww2D02gd8NEq5zD29NMe6QxmdDglSZcBJ0HJxVPFV+/ecltZGDWreIkPa1R45GSWI4RBa6i6BiUFcuw5PLwlL0PCRNHXnkE6VCQInMGPE0pZvLPkKiB1EcmspJU9ZqrprWd++Yy4kJh2jdsPbIe3nP0nnzB3S96/uafeHvBq+eHEKiuSLEIHHafnTxDtBXf3W75/e0egA6SIWBaKi2cClOcnP71iPp84Vhu2655dFWAbiwwkVllkFeMmQW0b3v3F3xHpBDeOHH3DECtCZYhrWDDHeUO7XpMnOf6YMAyWqCyIFfTbPc6MZNcF+bMZU5/RiJ536w1KSZycUMQ0hz11cKRMEyZTsVgtUP1A6AVuZyllTr8/oOKAzf1IOESMU491MDbQy5DGjvggQ6QZRntaIRhUTDiGjCZhMJ7KHJmkIAxjjNf0bmDyLcoeSPqOoLHkUUnjBSJJibRB+DWzleM0DYkywzh47tcGeXrGEA207x9J9TkmTaiHARcqkmTB0PfUnUGHBUZF7LqBpm0I+xFhW9IkQ0cpj5uOeHmCiWraVy2YCRmWNC7GNh9KkcHJSHpScrzvsYcMIRz7h7dMB88sPeF4iKER2LpDhHN0sqTe3pEOEiUCursjagiRskAmmtDnZCVkp5pMpMh1hZA90idsqpqhnvBZwcie9dSReU2cauZlzGF07CaLKULC0nK7aYkW55ixIclm6OUpu8MdxlgGofBjydZYorGh65sP0J9JMZvnDO5IN1Wcnj6luKmYhg4xBiTdjud/dIXRGd/+cs04GbzUiDoiDwQn48iP9DlGGnCW5tAhbMTBOLbThuYYUI8jzdCTxYZTNVIZz6vXR45NjRpHtPNE6hE/NohQsQovSLVDebBuxCpHpBXKghgtuQgR44gwhtGCiDxhCj/9k5hIPYKfcBYQoL0n1DB6yEPBnhBjcwZjGbRg7xL6MCOdKRZFROon6uOao9tBlNK7gb45kCYxBznRWUemQ1wICo0xLVJpVnZBYScoDMoGBKMgGCHUAcIZ9FiwmnKSIODdYQOBJJAheXJFUwfkOsMwkU4dRhz5k+uISxlQPd7x7u0Do/GsshlJVvLr32zIzs/5yZ//mDAOSIqEq7zj7XGHGRpaO9KGA41VRHHGp5884Q9ePOd5ElPVitfvat41GUIKBho++ijnH1/+F8R/8ddMcqTvJNYsad05j4eQ1n+MPn2J4Xu0N7hqINc7TO7ZWYtRGp84vOxR4ogTFp18kM3hOkIbUi4KRuvpug5z+46DPTLLFduX3xEed8x1TCtK2t5CI0EemZKJ4vME31imVwMP1jFcZzz7aE4/fc/f/OVL8lExsqDTHfNoSTPA6lLg5UBykhEqxea9YDArspMVyScxs4VE3x95+PY9U9AwjAVlNnA1P2MMT9gcH7Aq4TA4JGcMXcPF9QJ7u8HVNbEsyY7nuA1kuWAu4ExVBBvJ2597InuKdoouMHz0IkRfdOikpduO1I+O3lqS05TOjyiXsH7XImJHlFmC84nYQn8TIIkJlEHaievkDHNoyeojic3wQU/wqKje72mcoLj2LC7g3f0ApoGpRbqAPvCIMqT1HdMBkJLOCMZdxU9mKTo8Esma9ijoZAlFj1EfPChqnHBNh05SJCfYNmLqIozWpMuczv/D3vH/4DBQBCE+FbytHUPvUG5E4ImDCBEonJPIIEWmjtqN3B0F5enHrHe3GKNoncc+PvLZ6pKTZcHD7j2SAZFC5w9o1xEnM6am5+bNgaG1KAWr+YxICtZ376mHkb6pyXXBfJlwW6+p1gNBvqT3kqmStIeaq2WOyy3WOWKjCMsEITTRpJFjh0okycqjI8Mn1x9x+7Dm4W1FIDTPPn9KdBLQmB4bTHTvI6Z1C9EZ68OR/RbaViFyReA1QjuU9NjJIlREHJV01Z6pcFx+/gJxf8PjocJKQ7Y4JfQJfmyRGmJCFIoRSSiX6DgmnCkEYLzl0NwzjQOTcCRqxrIsMIscMcUUSYycd7THmtQr4tBju5F+a9EqRcU53oR03cRkoWsrdCRwscUmI00j2bsRyYTYTwTKMHQ1ToSMFqZQ0w0PKDESevhf2ICRb8kLy0xOFMJSZnOkzuh2I4iWz1+c8PzsDB1V9PsWl7R0+kh+taISiru7R56FBXEkGVNJPQDe4r1lMA3CJ6hoJJXAcUBGBZGNsf0tMvpgcRyyc26ODe1RkIkCZw0iHvF4pAhIP1JkyxkP+4Hbh47IBPQ3LXG2QEtLX/sPIS4OMFGDGzuUXkAeMlvOUHhcGxOIDB9ohtAx5D1pEhFMKQ+7PbrvibzFhJb3x0fmUYp0gvikpG8DGu1QpSbLFOYd+FjRBYqLLz7l/b97x2ZTEcUpgdbUhxqSU6yp6esjRTHDBED/4Z62diQu5rRSs9k+4J2jSHui3KMQ6E4TsyefT3z5V4/4XhNF4JC40xXN/ki+HgmHA7KMEEpSlDMOQvLtfYXtJkJrQCoCLxCjYTp0CKkYxYB1iskHBMLjzYTwlgDLMG6QJCgfglIYZxF2IpYRXgYYM6KCkSgPYQzotWdxVVAUG7S0OA/SgIpAOsBDEoBwkq9vR+4OmpaEwYSYwTKOhjgMmMaR2nua0aPDmCGIGAaI0nOKIiF0E7I9kroAPTmcdZwWBePUELuIF/k1Ina0vuVpfk7sArptTTi+YMeWoBL0lec0DFCEtEPDWB/JvWaZPfLxauRHzz3XlxnOZ+TZDPQFh+onvF/vycJLDi7mH/3zhMF6wkXAw75ittBcn57w/WbD+22Flx5nBYtkiZSes3TOdrNlsZrh+57muOfY1CxnM1aznLPFSBanPFkvefvld/w3bxt++idPiZIFw3GFtiesPoUH8wofPGBUiree5WJJ1ye8GhoksEgXGD/h8Vir6b0jjCGPU2QW09c9kpGl6ulrRdpFcLshTx06F0RLwUl8zjRryRaKrqrpqfD9gfRJBhOMKVz96FNU9w2p6ujXX6P/pKT9JkXEASpwxPGG4rJmPEswvuCzjw2RvOPkssfKN7TjFvQj8WLOd+8LBnNF5HvO4oyqnyOCiFjH3NQ1AknkR2S0RyzfkmXPabYHoumEZxenXF6nPM3uCIYDD29CsiCmXC7ZbDvcaNk8Vjz/QrLZTjzeD7hsyagVu/0RUcQE1lOuNJPy+GCDXmkyGzK9NRRmTqRSoqGgaAtOoyu8cxyGjiyJedxU6GhE9hlxGLB52RLUEukn0lSiiPDTSBJoztIVTd0xjCHMzng4bvl6J/nJ4pyTyyOvO8/rdzXPns2Zmj1aWnQssLZj8pDNn/Ldr+b44QIvHO1o6PvfcxhQqUTtJLILmESEkI44UlgEo+3w1iKFpJxfEmvH490d/dgS+hAR9JhwIFKOTELsJk7ziHRVchiObN5VxGFC3UO9MwSiIMoVhDV+qHj1+p7JQ0fIRMTqLOTHf/KCh3/7FX3d4ThSnlwRzE5xPmO7v8Hsjkx40ocjdqho9jsyHeCGFjMFSOfJY8H+5o6+2pFKidOGUR4ZXYjQIe/udtRvRxI341/9j3e0fUsuFfPTU3xqmY79h2vMUkY8RlnUTGCdxynH427L4/od1ljOnp2gU4fUI8PjBjmE4AJ0EZAVCuPukGlI3U6UOmEVzZgtI0ww8d27d6xfVxz1gG8V01Az+YnsrWZ6hDycISMBXpMXc3o7sHvcgkqYnCE7D5lyAWJCBw6yBjuORCLAjQPaKQLbExfug91veEcQGk5LKFNFnCjuq/doBb5qGTpFFi5wHTgnaeSIEBPPLnJOtGS4P/But2XTZcSX5yyfnTMdHXUzMhwUPlsSMLLZ73A6R04x339zoO8UU+CRXYvraoRRJDonVR/kxw+Pd7AAXZyzbSdGC9kCtLGgPY6JwQWIfEZrDSLVyOgDX89pCM7m1I8HrPHo1KNFiDwK4jBjbxyH2LH60RLR13gZMj0YTNUhNJBliEXJ/TcPOJug5Alvq4GuCBlVQ1QGONvgTxRuCNjpnjqxJJkiCCVxnON8wL7q0VmI6TpSlSCMI6DDtIZQW/LRov1IeDannUYOoyCKUsIgwklPqmakWcZstqBlw7GdiJTi7LKk2hxJoyVOjLTjgGotaXKJS2MW1xZzf2CQls12xBhJUM6RQYxxBtU0aB8ihCUQNYOzTHGMdoYUj1MGzESoJdZbJBPGT0QZuGpC+AgVaKzumaxELCBcBJxNSy5Ejh5imhxOfuA4vbrDjg4ZgQKEBSX/49MoBOkCxrHkcZNiA0PvGsa2JU8iUvlhnfl2s6E9HJHaoaMMpXIwoH3Ex8kTTmc7QpMw+IGjeSQLNLaV5PmSRRAxHQxJG3N5ek6oZrypvueL9JTLqx9zrO65e/VL1u8asqefkv/0KScXe54+Gbh+lpCoO6R/gxkTvnk/p5/OuTy/wrqARanIsjl/8/fvkGpBvEwJypDLIiULWk7LJZezhPtjQNuPRAheXC7Z7458en7O69c7fls9EFiH1JKnT5/RmJYgbBGZ5ug6nj0p6d8u2Qwrvr/xnOcV/U3Ly/t3DM0dS3uLyUoe2obKDQRTiy4E6bzEZoq+3eFcg9L2w39fl0TzGFxA6y1yFnB+MqfUAV9+9YjOlwRNipwU2SJDjBV9dUf4ScmoIsrVkqx6YHdsaQ/mw706KNYPGy7wvLZrsitDO2+oZkvmLzp0dEMarvhu7fDzOYkpybKWi9UGZ7+kOTygooQOy2gdfR8iJ41vQw6Phv1+Q/aiJBAhX6xOWDePuG7DIhBklzknYYyTGb953RINIzSWUa9ptoJ+OmdI90zBAakldCF3ryTZ85FRbimfeL7ZDry5BxONhLVgklsIRhAzbGd5sowRv5lY/nrGMsrpZhG+1vRhTKUM8bnDlZ4hSTDZQDAziM5hkoLdoaILW6S3GGEZTYcmYTQeMbdkRcPdtxMn2VNqc+Cu85i+IreGYnXOq7+qOPvokkgnlAtD/fDIMFzQ1DO6aM6U53ReMs0lrUxIZr/nMLA/Qt04ZOQxrgUf4IQnThyi6QmLnIM78O7lI6GX+L5DJIJVUn7Ah+Yxen5G1e7Yv96wuF6R4JhGy9gkHA8DRjjyxQXF8pqm2hElHdu715jKQ5JglUBEgvxco4ORyGgWekGaFZBLjmZDbwdSBMkspZw7Pv7hinf3R56cn9E3jvq+omkqlvpDO9xPKXnmGMKK+CTFxlsmJOIIHA5M2wZI8c4T5wHlRcY0DcTSQ2YwsxEfekJV4IWkczWDa4hTSe+OhEWISiKSOGNRakzdE+iE3o4gB7yUMBhineNliE9K2hAmu6e6cXR9T9sqqseeQ2YJGDGuwTqBUOBxGOFBSryB2lXsdzW2noiKEBXFRIlgdqZwpsHueyyCYS6p2wqdTqzCBNs2JIVkkS/wDsbOEAcaYXpKHxM0MwJnMDtDc9REVwFqFiPzmN42/NFn5zzJB8K4pzYjxS5kvHHUR0n7GBKkAhVWOKdoJpge14jcEWUZj49HNvcPjGWKXMzIrMQpg/cZaMMUHVCpwVU9ImqxSUerB1SskKlietwRTQHSC6LLlMXpkvXmwObtET2tsGbC2ZCm7bFZSBCFpHFOs2+ojhOxnHBxQPnkkrMfPoPpyOPxwOHwSDIVREmI8R23txum/UDkU8glexERuzmrKERWLd3hSBDM6JOBIbQI6+iqnjC29JHBBAnHo2f0Bp86hJiIvCI0I/t6SxoklOaD7XGsDNVoSbNLvLd01YiKQqTN2fc90TAgg4heTIQLTT8Y7t+0jMpgkwSDIjAj3m85f3LFk6ea28M7du8emXoBcUC/r8gSxVUWsD/skFqipCOwlm4aMc6iXUDoFIIPkxfVK4wD4TVaRER5DDKhty1x0FMmMcJCnmuWcmS1T/H7jsP2EXkGL/7zF6hU4bz68PKWfFCW/8cggBSIIWdo5xyqkeXlwDKf+Lg84W6zYX9sebjf4ZQmCALCQGKMJw4CsjghRSD2jrNwhpWe0A08TebsDkfq1xVPlheE85i//O++Aaf4/uSXfPL8c379q/e8c6/48cefcfPtA8VmTTE4Hl/+mh//r37Cn/7jAZk/4L1i2L0hylpkEpEVgt46dtuRs3KJ8DEv1xXNlKCmgFSmtHWNFB11Y7hr7xBDjXY9dhSs4oKx6/HOUm1v4TgQ5oIwSfnsfMbbY0vQGX58lRLqHXGYUv6w4Ozip9zfWSJiXv/ir1FVzWJ0lO1rRrumFyW2yJGBp+8qUhTR1uDuc+bJAoOhCSyDm9CFwJWOVo3IsSOaPGfasnAF+bRkqEKEKRi6HjUOpEuFVTVVE1LOr5BDTRCBnucfNkasJIkK9nf3zOYWNYvIP7rC2R4TVVx9LsjCW3ojsPILlD1jGhuayTGoFf3uS1Q6w5sC7VKK/JzzlWQxk7QT7O8FMviY4bFDHSTnXUqZjPjZE6Kjx+6OmCmlu7HE65Q5MVefTYSxp05S1qrH/aMW320pomccfufI8phFIRmTlpvJczA5g46YlKScrWjaFiMnyhAuzxb4dk+4L0jFBemg6EaPUQJ3YtleranzA43rkc2M1ek13ZuUIHQcxPfws7f0Xzi6TUQsS9IxY/PbGn1q2X3e8GR+RvdqYtgcCOzE6CIeRc6mC5DrjvArxfSQkV82yCmmeQy4WT9hdxeigp7s+sj9foeff44PZ0z/MObQPzwMNH2LCRSIibHzWBETa03kFVNVE4SWMPYEdiRQMeUsBzrEcc3pKsIeLD4oyD9a8bA9cDcZdg877ODpVYaeZ8TOQjAyyBZRBCBiusERSUu+CmnHkbMnMUXi+erXr6gfGmQaE89myLRl3I40+wnnHJqQz55/zutv1txsHlkuP6brjnTjQLYK+fGPn/L4+oaxaxnMgWgmKBYlTkfUfYubIAw01TBQFDFZIQguAyp7IDMloh5YPl3i84Kpnrh5X0Ns8MOB1BlCr3ASZKTI4wXn8zOuLmO6xwPrzRtkFBMkOZM0OEakdAxuYEwtwzxmrmdU61ucHzl7esEqGsjmC1S04PD+LaYO2LxvGc3Eo9uSdhGzZMVm8OwOCjEK4kAQo5g5OPn4BbuXd4z2hrbZMTuZcXl6RT9taB+PxFHM1ckpz1485xe/+Jrb1wNFrNG9ZZRHkmnGUsXE7gQyR68ccVJyHB9YXcAX2QPC7HCmZJ6foJM5r++32DFh/VVN+ZMStQxIVA/+HmN6OjFSvRlgLbkI5rRTwDTGRN2E7yb8haYNb1m3G8LYMdkjsRJMPehJE4kZ3XZHMAV4F6JCz36zoflS0dYTkdPo0FIPR+QwcKlm1Fiq7cTxeKTtNcbOaKTHG8uinLHbHqjub9lt9phtS6AXlNmK2hwZugAhHCIJsPGA3Y64wDHVPflK4BFMQ4X2IzMRkg+OqPe0u5bTZxeIfM5jVaO9QiFBKar1Htm0aOVJggxCC6nhaDuMU1j3wXAX+IRMBtSyJ9Uh3UOLCyTSfqB89luDUoJOdrSTY3QeKSOCGMSq4+3NI+vf7JEtXBQCR0/XW9ShRyUTiVMI/8H1EFiLbCeUMJhA4CaFVZrJKqZBEmqFVQYZSrJMEMSOPJkx2yuWLuVQtbD2ZMJSbTYY6wlzxeXnZyzzGdoHiOnDdSFA/C9TAQXexezWl2y3AaHy/MHHpxRzz03dcXx9w+lqybHW1L3FIPFWMEwTBzHQRR3YCCFGhNFkWvGZWnBdG/77//6O4+ueb8vfoE4Vc5kQJJroLMeONZ98Nufw7Tt+8be/5DRV1HKDDcFnKZdnBuEaoEDYgXi2xI8DkzlydvKCan9FdYgIRUwYRHiRsT/2nOQjH81zus4Q6hmhcpSx4soY4nLFX3xzgxQ9Xau5SBcYs+bJZcTAniCcsXnXEHx7y5PznJMfroiz8YN/QoSEuSGbGf71f/1X/O6//Z40FgR6y7k7IPqCd1+tiYKIq+UZD7s7XNUhmxIfekrZ8LDtSFYlq6s5nQ4Y+wrf1yQiJA0Fs1TSb2t8CIM0nJwWiNZhvMH0gqZtmWpPomcwOmw2I0wVXfGWMd1SpEcurwcurh0qkVTxRLcvOTkThPErRgxaR6iDRsgQFwc0lUC2NyTKkyQ52/1A3ypMbLl6MUPpDhH3PP0sZ1MlvP37gPMgYFUc+fTkkpdVy/a44+rymthovn5fI9cZOu3Ikp6t8RzShBtzQKaKeJoxyxNgzzwyNNSM1tGaHF/G+KBnMiO31bdga5IowvYNaaYxgyVYrOjzELM1FGOJLzTrcc/j9J4h7uijLXa1YIcjJCKcNvTzA/6yZ9PtyD6LqZIj+p1Ax6ecf57ShSHfvJZM8xDvBtxhou86HI5JKqLQEzQB69eO55+VMPVosWSzjWnECvlwi1QTgREEo0UqGKrfs5sgkoJ8tqJFUeNBBniliPKA4/semQVEWhKHinbomTqDt5aZTlive/qx5+mzc9peACNjdaQPLGNToaXk5Mk5HCqG5kC7acln1yRJQXFxwfDNK3jcE0h4cir4x3/ynH/zX7/BC8nJs5yPPj4nmyvqiyN3247tzRY/Brx+c/9BDmJ6MnNH12yQWnL9wz+gajrSVcRi5mGyhKFGDYL7dxOv391jpCeIIvJFiEga4lWCChW+hiiV2D4gdgLdasba8bw8w2YBm3c1T4sT+n1DNdVEeokdBAd1x/miJEk+4E11JDGuIRASgUD5AO89/XFAqRm1GBnWe0In6fwOHYMcNNvNA311oK57JpuQpUuayTKMjuNwQIoZAg+hYxQtwivsINnevGW33hF7TaoKTvScNFiythE3vaLpPDe7nl9/+y3tzmBNxqBHPj5JeFrE3P+2RsYBRaFY5AFVf8Q/arSK+ennZ4zVA028ZNOE3LyaePVtz6GSnKwCZmcFYwdFdIaSW8bmPV4bCCKCh45lFeG8ZZgcU1uSXK5AP5CfC359v2YMA0bpyZOIRSyo321hneNUhA7SD3hiPeDHhtIm4A2DbVk+yZG6J37YEIeGrNBsv29wN46ZKoh0QJ2H9HTk5ylttef7X2yJO4/yCXbwHO3A+Fij4wJrPSLbs253uMNELBdkUUqqHEJsCYeeaAzJrMQKhxwNca/oq4ByTOlMQ/f+Hbb/QCNrg4a56FGJRTqFcwNhYJHjkcjNaXxE5xpSJGkQo9uWwE+kPiTVsK46Ui9JZMTY9EjbEZVLemMh8Bz6lua+ou0fKOwHq2SqIs4uYtxwQI09+37LbpHiC4VsJiJhWGYZ9V4T9xIRKeRMo040Oxfy+rs70iCgiALmRc4f//hjTkLJt3/1iu7rRyobMkyaxk589J//iPx0RnqR0PRrLj+aUXwR47XATJIwBC//YxgQgI1o2z/i7/7DU5quQKuevj0SxB2+nXgyX3CxOuGrXx3RYUEcwGgqdBQR6IQ4FPwnpy9YaMdcz5Gi53m0ZLbb8atP3/FQ9+Dumc8Tdl2PLVKYHGO7IVQbZOfR/ZxRj7z4449ww0Can5PmZ4zTEVF9AEY5OyADDZT87teSX/58A9kJQVoxqhphBp6SchqHPL56h44kOwfzGWSB4NOLGXeVh4/PKWNNmkacpnO0tZxmBYqQf/83R/5v/9f/hmh7g5IT6//Tj/g//J8/QYoWxBnWNkS+x3cTh02AOlGcLkPstx2r/hThLNte4BXY+4SmFlhnef5PB2YfP2K/EvSypGsaojIE73DSIaaGNIso05LvvzsQDAvi84TsQtF3E1Iq7API2LA6ec0n1ztsX/CwmfHsaom/vmX+/IHEj6RiRORz7m4V0pQE64LZ+TmB3pAmz3j3vcLXHUoL7BRh5MRoe8p5gIotoZrjdyuO64jFjyNwPVJaCp/y3VuNa84pLtecnnraynP3c4Fpr1mPR370syPBdw63N8zPJUHxHuEGvvm6437siJc5o2uYuZHjouUYaQ4bic0z6txz5CU2S/GDw0+GWTAjGi1C9BwPOUV8xbu0w602ZJOk12DTPeNVgyl6nFI4FdNJx93xW8qnS7R5TXouEUmCDGdsmhGynvREcaJrxrygiZ9zP1UUfzgx/0QQ7Rzf/MUBcVdi2xAjI4QIebhp6ZqQIGiITgLSi4S771ryPsc5ybRrWC0DJt8TCvV7DgNBStWM7J1lVl5SNRak5GF9i3Ue7xVd1ZOeLFlen7G+ffhg8QJ6KZH+Q9LZty27txusHMnOFygpKGcxUWJoDhVlHCLajvGwxliBPNxxIia0zCAa+Cd/WvJHP4j5+f8UsDpd8MkXz5kvVwzDI33V8P67e7z9sBd+fm3IleCZ/CGZ1jRvtwjfE7UVx8c9SZJQtwO+8Yx1i/A1DsnpssBpD/GEKxqSeYodNsS7JZkqKcucvrojajLCMCIyA0K0VFtDepTEPQg0csqYAotddqhRM1Waw2bN0Gh8AMMwIXVIexzQ6sikHMaN6KlCRZLRKlznCfxEHMzouobqfkOI5vn5NcJrptqy7RtiGaCTlnbcsCwEZXFO02yQYuREPWF826A7gx5DknSOaHNevtuwrwdGF9GPjgGH7RTapxh6rs5KPvpBhrt5ID+3xE8cx77FEmCrnqm1XDyPmAvL7x7gW5vxy7uJvtGEpiTUA0E4oBwwjATHCN0GhPIZtevo25q2OnBRzujbHd4rZKY4nkdESY7OIXUF5lCDdERRRFKnTAeJCEK69pGhHXn20UdM7S1C9ETBSP3YoGVOUWqkmZidztDjSL+tEJuBk8XIfCU4jIKTRcwQLNk5/2GdbTfSGklgR4o0pos9tetQPiIJPNp2+H5DzIwojih0iDctpY6wXYCMQ8bDA/IIygdQ5HjlWO/vedgcibqQCxHSBAEHMREnAuE0h/1IOFmoDyjzwOWJZegirIsp5zPoN7S7R8YoRWUzpm5P19RIIugb5mXKtG2xkyJNc4xx2ERgNjVOBDRdxclVQTgGfPHPfgx9i/3V/5dt5Ek/+QSXRNibPaJtWOY5aZxQXp7iSokvJ/r4iLszXJ5fczE/oUgFLz664Posw9BQrFe4Ooa2QxvL+eWK+T+7wsxCjtoxNjPukMRDTTZKiuAC+huCeATh8X3CNP0Bf/uXP+Hl26f4NGE0Df0UEo0pZnzPs/OBYVMz1ZcE+ZIiGgjTM7ZypDUhJ1nCebpiNdMM9Z6b7+755ps3/MnHF1x98jlf/9uGpZjxyXzFK3dkt4dFXNBv13z2o5h3s5CdCLFzx/zZT2jXW0yS8PpG8+nVp6QRtP1bJDu87RiOCb/5u5w390uyQLFcxCyeXXD77luui5GkfuS+qpBixq0RbKVkEWtmuWCxmFHmS2apRjjL+uFItfccF5Zl1FBVAx//4TPsnYaHiS9/2fOfvgmZn0KUaX718wPzfcO7X9+D7oifhySFpP9VwXhQZKnBOqjbliIr6GrH+VVKEW8JFegk5Hg70O09Z58X+NEzW56QzyXT0NHtWppby2wKiKcI300gYmSaoLOe85lk+bTj+ed3DDbkqvkRcXhDGX7HmprjwRDFDo/h5PoZwuWcxitM79HyhPXWst0IilVIM23oK0EUdR/Q00rT7ATN6yXTd5f4ynPfOWZPX+BNSz/NuP+mIC9PsF1Nu4a7R0m9iZjSmOVlTTV/QPxRxfIHhlBP9OEr/HiONYJsEWPjHf4LeLz6HvlHnndLQx4ecUnOgQlrBEPXE6clTIYgWeL7EWWO1F3O99UBFx8Rn91TfqFZhinh/EA7bVDtE/xhTliGbOoDkxTs7D1nlyONtnTVhBs0syRHCIdJYqYxpLaCwdTsMTR9SBtomtLy5M+f0f62o1kHdA8w9pbWKEx+StPH3NaWl23Nlojs6gR1B2HXQr1DzgoGZ36/YaB3I3aKydOE3hpc19KNHcpUzC8XDNOAl5bj/Z5j2yEDTx4KYhUiopLmsOVQvcaZhsx7XFZg2z2zM83FZcz2/Zqs9MQWZuUSN8upTc9FvKJMthx3I4sy4MlFTqAMk1WoJKbdbvCD4Ls3N9y8eiCdxWgF88Wc/92/+Cf8q//fv+Plz18yT3JCH+CDEdEMmPs9t4c7Fqcp88WcPuhRUrLbPCB7iGYhXX3HSgl4bzCjIk8t5rilrlouPjlB2YLNm7cUgULVHXMVs4kMYgauHknCgDxN2VcHgiblsbpnCiyP24nGWlQSk8aaaBkgfU/bbFBacJKXJKHkGMYkcYITLVFpeHh1g/eeqEgIlKN5f4Pfj1ytMopIEsaO4GnJZj8yHmsi35OkHrF5T5ilSFNgJ839pqZpGoyEwUyoUBDp7kMzPRjxg+WT5wuelhoeK473PbMoAPOhoPfmfUsyGSI1cRknHNb3DJuJ990e7xNKHRBlhqiYkQrLVG15/uSKcGpphorKJngtGDoJUYJLRvw4gtH40DG6kMM+JN6PeHLksUdoSGdzYp9wmPY4a9BKf+hNTJ7AKfCeJPD4MWQzOJoHT2Qi+oc9UW0ZrCAfNNdPPXEx8vBNTaROeTgcuK8nwjBlrjIsnmJefJhexD3+MGGHPYOpCOyBoBsoy4TRGdrjDqpHcp3Qho7jZkPStszSiHFq6GxFspgzXUVwF6IGmBcJznQMYY+XPeVsRTtJ3OiInGOWKtTME5cRb28MiYRAxwRJQSM7rD+QpuGHpvuhJlAD0dhRBjGbY8PYVOTFBb0whIlgFsVsHxVdHJCVJZfzL2i239MOX3Edp6yuniAvr3AfbdhuXpEoSTeWyNOGWTHSW8WnJ9eo+1dcn5Vcff6CKJuYFwVNO7ARMePHz9gnR8o0JLIHpBxo4vGDRXEUTCbjrlUsjy1D8gQzDpzmc5zbMzUhbf8R67sf8fUbQXSypDYpWXhGHh4o9ZbL64ZltOHbg8LLF+jsDCNaQpESmoaLKeY8jhnViptjyyQSvvu7Fl8d+R/evqMIE8qgJaws09+uScYBdfqEIrTMvaX9xjLPTwmXEcsvrlGZob6r6A+v+PoXVwzmJyRqy9PPl5SnW6SOuKkuOOxOKKKCRRkxL5eILmMRW06SDDltmXUNzeDRekHdBKx9hWRChUdkAAKJ0ZpIKuY+5Y+fx/RpwL3oCX90jvp4SdoKgmDg9duSt/uYYg6r6z/H5i0f/2yGuv4t88sbnPG0sxmqL/DHB4YgZkxj5Czn8hPN6skz3r36Oe5Y4PUOghrcgmx+TlPdkp7HNG5LUcRU7zqkiUmEoiBFiwjjPTrIUbP3XF02XF5CGXi62BKpAFu3zArJYBwuHFGiISBitGd0dwtWi1Me+h331Rw6RfYkx7dH/L3CjI5S7yjzgpu1QbUlYr+g3wzEKsPfKdp9yM60THKALqd80nDxVHNsGnp1jlAOuWgpr254072nPbMYeUfjR9J0RMiQ8d6x7QRJKGinii54gAQ2LuTah5ja0XhJEp4SPu4QPoAArDCYsKdreiJrOQpHO+wJM4M+LxjrA6kw7O8MvPJczkfOopDapezNFvweEYSMbmLQA2kRovSInQBvObZbAhnQTCmHoeUsi4njBUNf4iKFPpMUvqC6aTBBj5yd8dUm4le/qXDpgm0YEnw0o649+rcKu1eolSfILbL8h+0W/oPDwGhbxjbADw6jRtLIYcaaQEqefHSK0oK720fuDgfcoePi8ydEImDYVYgWzp8+4fl8Sf3yV7yddrw7tigVs7w+o2hK+r5DdD068FxcRqhTw5c/v+FwL1GJZtO0rC5ibt5N/Po/3PP6u4AiKVi/vyOOO7TvubpO+ON/8lNef/+aTXPkr//yL7h7dUei5mgnCTNNESWI7kAwHTktNLPrmtXZxLHrqY5HlrpCICFIcdVIkXzQxZokot4fMXtDURiCNsdPjmk/kD19RjfdkqmU4/0GNzSkSUI79ci65lSVTNNEfbdh4yy9VehMML+cUbUDYSoR7cTVYkVle0Lbs3m3RqcxcRKgu4T+sWWqRlwCPvaMh0f0oSKXMcoOhFlEMo/YHR/hUSKDkCwIiIRGRYKtqxjqCeETOtfilQQRoJKRk4slPlT0m5ZEes6ezXl6sYDmFvd+x4lJ6HuolKDNOqasY7ZIGU3LdgqJBYSzNZ88SXk2jzmsD+jZxNuXe8JuhhIjUWSx0xFJhxtipJWoypNcLBGBhq4hDmIoJGqSRH6G5YB0gkid0qsWY0dc7Ok6Q5QsoGw4X80Q0iBNAPchkQvoA4MJLE3XIGzGLF5hdo/Qd+AasscQKhDVwMasCbKS07xgMh39WBMDfS8xzhH5EGENz84z5HFi14wIm1NtLWGqGFTD+TxhqRTbyVOuronCNxyrA24cETYmKAWyCMgfBHbc0/YTSsK5t5TGwGHHrtG40WH9HhcIxocjZ8/nPHrF2LdY1TOoHu01QeNJhCYcQzIZ4IWk2hxxSlEkJZPZc+geUTNBWuy5KHsuLmOS/GOGnQCZoynwTrM/HImThsuf5TBmlK8lzZuW4HTJxR++4ulizctva/JWEN29ozcFpz/6U/pS42WHEMAoidQF6SxG4ImKmmUYQ+NQQ4DXFm1zAKp+R9s3pPEMki8YnOa7O8fjWmMe18j07wmizxDTz0jSmFn5LatZxfr9V4TyhmQ6IXcB3i6JifHtjtI7ImFxr95wfO754Ysf8Gb/O/JPQ1jHTOOW6OIlxdk96ttLpgeL72vqac/J8oTEB6gopU8y/tGf/4yf/Nmcyf+O/Z8WzDLDu28td++PDGJADBM//hdPCKJ7IuaU8TlOWl48f0o/5Gj/kj8u/2f+8vATZupTzM0RGcfshWa33hOVPUGYYCgQcUSQzml78NKxMZ6/edMxjy2HIcaahCHTiHDgND1jXye41nLz+J50WqOajDC9IrruSJwgDRT5/3pB0J/y+Ood8ULj9JbZaYM8Aa0CAvkMmaSEQc0qddweN6w+/4zICtrpjiGeGMca/SxGPY5It8WmLSJc4bzj6iRmYOCTyz1lsiHXEf1U02y/JDYnvHo0zL94wbR/TXRasUxCHn+VoP0cmXekomXeahp7ybev35Kqjs1bR7E8ZVNtqKs5kpzO98zPE7775UA0BvhcYMWOT/+wIdIbvr0/cPrDLbY8cv9uoskawvMTOulIg47eKsawRYZH2nrCh4K741uqKKISAUPgkGOFDiKM0DRdwNvBM4oOH6eYx4ok0FTTDh8qkshz7B7pQ5gWb/D7iCIQBKmm6ydsramlQOqCdpCcxAH3m0dc0ZLPO7ra8vAwEs8kkxlBTFghscbhbc10UESx4HY64hYBk7XQjZQHRX0n2L/P0bHm9A8DVj8skecNf/nlIzfvDEnpSBYWo3vCuEQUMSo+kIURdW0w1e/ZTVAsJHEKq3DBpoeDmtiPHqUdd+/3TGaiGbdcf1YwHCdmSuF1w/wUUqGQQtKvJ4ZbQTQsuU5yxqBnehwQc82Pnv4IFXik7JhlMdPwyAsZQR4hpcYyksxLivw5v9oICAdkskN1H9SsUQ7l/ISf/9Vf0e0akqygmiwJgmk84FSAq4/IjUBEHeG2+XASlAEre0QPA7lqSOYDnQgxPUiTM96PRKnEBYr60DMLJGG3Y3hV47uYLEoIVgGiyeiaiTyaMQwGWxtmixnSgBkNcRrSCYmkZx5qinmMVBW173GkyDzCTA7T9ZgoAOuYRRozHJEDjE3H+cUpMhGUhJiv9zyLzmDo6a2h7jU3t5bOpURk+NaQlzkkHXbyRFFA50ZUrzg7PacbKtrekJ+fkIQBQRRhU0HhITYTauxIQkW6NMjDWzof8uBDZquSVV6Tz1q6ShC2FxzmEft7z3BvqTcbolnM6bngYd0ibYqymrt393hfo4IGbyp8E5LWAdl8ydDWdM7ixwZ9PDL2Oc3UE5aGMA/pw4m2M/STYWzuKE8u6AJIngakgWO73mFkwDKZIcYebVoiY1EqQ4QSi0fkAmk6AmfIrKLd7NBDge57lI5JZpr24Z5gcMzSE4SS1GFPpmNYebISVqZg3I2Iecam2uO0Y352QpJIbt+/IY5LVsuQrjXgFYOL0EqDF6RO4EeDnTp0ozBjx8fXJ1yOiofHe+xjgErmzPIF437H4GvUVcUsLHBpQO0bnJeELiOKGlLRsncDNpQkSUmqDeFCkBYGpUJkUVIuAvTNe7LxkZPTjPc7x9t7x6v7U9y3vyZ/SNk+RvxwvuR8cU39cEu3veMyOWUXrBhqxeOUkqsb1Pg9P7gKeH38hMBeYXcDiX6N6w8k+4+Q40CgviWIaj56eoevjxya/xLbZvTBEakikrFFTQcO7ob3uz2/+Pd3JPpPmNol9eEVufoVRX6PT2OOhzWnF5794Zb12ycU4RWtvmRqnnHmQpw32P4XpOLn6AUM0chIT5z/mHC+A/srVj/6jmezZ7RBypvDd4ST4pP/7R9SvLPcf/3I4uyaQkes+pR10+JUyBhWfPmbe0y5Yf6kI04FprlEHmJ0rtjeeW7eH3laNpwlOR8/f4JRlixL0HaBPP6cPFyTqRXHduBi/nOOR8cu+N9zbDw/vrhgijPeHUfUYIiSHj9NSD0RRws652jHR0rheZEKRLJl1j9wlhpyt/0w+t/eME+f8vB4jUuu+Ghl6B7uaKoDLlliT97SP1OgMmI3oJIbvI5prKW7fIXxMWXY4qMQY2vu3fccdcVke+oGXK+4uNKsrgUzc0RPXxEiea7hxbOPuEiuONtt2NCyHStOF9eMGEaxZXV4yrgfOL8syWY3iOQdz/9oi7UNctwxCwJCFdHdtpxlCe/uW0gCDoHn5m1OmGlWZ8+4f9dzVweYMoO5x44Nl5+EcP6SQ1MTuDPOPxp5qFrKC0/67BtefSvQtyVZuuPoHlBWEyqPDT54OMzBkBQa2dVo/YF6eTgMmFGRRZLbhw1FmaPDCZNMHMaetIhI0wRpQgIVMkUfvDHOG5IswsmIyhmi2YRREzYOMT9+xfv0e3QiP/TnGomWJW0zMCEIo5zjw8AUGOJEEBhN3YaUUYHWmigM2O8mvtnf8EQG7I97youYk08qhmwPUUY9Btx/X5GlM57NTghzz5e/u6VYljhd4/IA002M1Uig/2Gv+X94GFgW3P6uR0Q9QZLx9KNn1MENLpWsN1vodnz2B8/AOrr2yN36lstPV1yePUW0DTdfb+i3Nf1RIpKYPI6RScjsesXlswt82/Hu5QPv7tbENoDxgf/qn3/CN39/QzV4sjPH+fWKf/2vv+Ht25rnTxeE0cTh/QGdOoZQ0K/fEzYSnQVoLLbpmac599GR2ckpru0w3uKFY+w/QFniJsFvDWbtsfmKTShprMfqBH2ScnJS8Oa37xkeJnQRcX0ZoIcaOd4zDiH6/Dlv9/eMbUpbr5m2j5xfnXF5fsn3X78mmKW4VLN5+0B4FZEAs4uM8NCiRsuJUdQ13Fc1dIak1MjaELuIGMEURkydISsyFkVCZY+4uwOr1nAuDcemgjzivR149BnSanQYwdTQuhpnBgZjodMfVsSExg6wnK2Q7gBbQyscughRNmQCup2hNwNXJ3AiSly7IS0UnY25/8pxfV5wmtZY2zMEDxx7z24TocOI2ZOI7DqjR8FB0LYgpUHWDhWMjPmBfCVxLyUBJ8SRpWoaxt4hXILeVnTuDePJDBXkbI81ru+xRjGEhjBWXJ5G7N/uMO8HYh/gO4suAsbaMPoJoSaWMqNvLUPQMAqBCiZUJEhsgj0cSRJHHEI3CPAGObQs45JgaNFmYHIpKlIYGlKdodqRdnJEuidOHXkmaeWAE2uOG80yKNGRoDVrBu/pnGSMHD5UGBSZ1MyNQEpHJDSVjShtjNlvibsZL5Y5TgdoFbDXPVnUYY6PXM5b3q0DpiSFVJLPB7KTB+pjg9IXyBzG6jXOPlBc52y2G0ZbkJ78KYGWTLsPLehocAyvbpjVSzZ/8xeU7e/QXpEuVnz26Q8JDprjRvHLX3zD86uB+Q8/I4i+pFx8STN/g744cv3ZH/HtX03cbzY8Ux0X6Vvebg1L9SmN+y3r/v8FZwtqoVnkimz3EhUvSJRiV98xHL/jbvo1Nr3n/cNLomJGkSxIoxynthxlSJ7P6OQt4VlPvOxZ+pJX7+bY6DltZtm2I+mz7xnEt8zLv8HKnxOcLenbgcD1jMEtP7//e7yriNizGX9DMPsDYhOSWMe2+wXZvOCTf/Jf8ig+5v3DX9MU71nLA0YLPn12w94eidVANT2w/TZEr3/KbEyZhiPRas6brwVWSTbNL1gLSSQK7GbAdTWJkHSn/ynz2vMs+recR7/k5+1PUW5J4DV3rz3TVU8sM5LpNfN0Rxo27Na/JV+cMYsVSr5mOQ9xzhD492j1imxpsbLi8R2I3SXh/IIueotK/5709DuY7REioOk1QzLHKc/oRqb1kShyBOlTjDxSZjtGB4sh4/DVQCQTHg4OFidYO+CGA0JlHNsJn3aUhSfREPlHTsqQMN6hGHDuLW6CkYJ4ccFhvKFcZVTqQBgdkfGOLCkZxole/3u8yFjlJVauePV3a6RQEBmeXcdMQvLyXU3vl2zjlrtvJ8LqDzDHEn0yYsMbLpcgF/fopQMhafstu14zDgMizNlPD5jZnsRUNNMWI1sCEkQEKM1dlTJZTyAty8AjrUXmKY9dxUiL9CMy8lTHliwYIDBgQQlDqT3DIaF+mTImLUlmKCJFGGS0w0SUWawbwQs6ajgTdDIEGpx3kGgO+w3Oxogp+lDGL0usqzHOMO0c8/QcMYV095pwNmeq97jrgTW3JP94IJ4bjBCsZgF+Snn3dxJ/yBBKUN/ec9rHnL0LiRvNcYopnpxjG0mSSpp99/sNA5vHif6QYKKJODM8fPs7bN/R7mJQE8+en2PCFd99+ZKwEVw8P2P+5JLX9y39ux1iGuh8DZH88KmhM5jOkz25xKqE3/36F/jOkIcZkpiL64xgkTCeBcid5EQJqq/+kkQE/GyZMrVvGUeJWpVUY09hEvzhgGlCTB4TpRPNweCjGJkqjDEUckkTVazvd7gpxYYaEUXY/Q0nq5AHHAcVQ1ESlBLjBh7VSPSDghMV4sMR/dCQbCfs40CWnHF3DLjfNNA7QmpOrnOk7tjvb4jORmrRcmwTgkAwL0IikZEokBtQlUNojUkHGq+IdIJwjsMwUcQJjJbjboPQAUmc0g2W9uBJRUASeUS9JUk9a4702SXNEHOSG2bpgMsqopmk32qwnmmasErhGIGeiZwkNbSvHijTkkgVSMAZMGHOZvLYu4GTyJCHK2IZUmYBX39pSDvFdRbhtw2BHshShXgWks0NstQ0YsAbwzz3VHKPUJ75WY6Win7tSKsR2hC/inl8XHN43CBGS6p7EiG5NwNCBcRlhGpH+rElKww99QcTna5Iph3hfUAUlaRFAn5AJYppO5DOS4aqYggPeFK8iYmUQEUhkXVoL5A4gmmkNy2h1hQiJXETQky0zUhWaLyc0JkiONbsvntAqpwkT0mmA/HR4yy0hWQkoe9CxGFNWQqOQtFICE4C0pMchSJre4S1CBOhbEyeBWx3G8RYM/iQ+TPFk59mfH/zK6LkkWcvJLNZRHd3zdv/Z8pqMaP8VKHsDukb7MLzqO4ZnePkrCcJb7CxoHCnbL8TlOmADe6Yq4RICdpHT9KnCDuhxj3UOXUwwy48r9/+kvr7kXSco2xBO/UM1f/A849+zihfMXrDOF0SJVv+/D/9NceXLX5zS8vAtPgYmf+CsP0fyXJB/tEZjXhkEb3Bure01R3d1FIH94yfrGn8I4snAVEK3mfs3Xt2YU2TVMyM5+x0JHYd1VGS6Q6VOuQnW9bHkHFe0yYj1r9l6Ae24p40kGjVIbMVz4olIgzY2Ix62DKJnkRpbr7/HZ98HtBVA4dNgxBzHtuvqYq/p7z4t8wvjsw7DUOMmFvo9xRxTFFErNczXPIKHTiCq99RXG3prGAfvGE633Ed/GtUkOMNaDeR80Pq45yT9ud8dLWjvfsZzfjP0FVBUXWk+UQiHLG9Y9r+O9z+SHH1DBs0mP4tk/dcnN8wDTeYybIoLdWxxsiYyTu6xxm5+SEDH6MXoOQtY/COg5HsDgs8SwauSPKKVD/SFzAlI0ExUd03hNJyGEJefh3hdgvkYIm8p2522G4iE5BJS2A8q/Ipm90NvQ15cnZC33UUs4hYQqUtx/uAxp3zUDzh1XritPpgv1Tzp0zJij65pThz6KFj7CSVW/L6IWC9eY4aClo/Yn2DvjygheOj8zmzlaMN5shqxl6EmELwk48CTqNbWlr2/QYWR+bzmCaN2Yqepn2PSTQErwmznj6uSdMA24w8tIa7vuTLL1POzp6zffg7fNyRRBnrdY2QGuHcBxpoNmPb3GImxaIosGqgjDxX85Sv7gLG/RIzhSSlIU8rjtU9Q5Di5UQcLzBNh5h6VBLibcrke+YrzXHrETYgDAs8jiCKOaw7NIJIWoIxYHmh6YYt3aNknCZS1RCFjs5sIbPIImH7ynCah7h4ZLP2OBsSKckqT9CPNSebJeb9njiGLvC4yCLkiOH33BnwdBRFjM8jrj5/ymKV8Hh45P/zb35BO07I62tab0nnM7zfcKgPLPtrrM6YX5V0hx2b/g3Xn6Tsv9sQxwFV37K9vSFwhlhF6GVIuxHITPDxZymvfn1A2pLLT1uKo4G3NXI7kLmYrlYMUUpjRxKRYq0kdYLtY0/Igs3NHYgMLxtm84z6WPHw+gHnB8IkRp/HjGOLDmPKFbh+zUdnV9gKXB4SXsQYlXLcNsSJR48DWZpQXk3svwe1K+iHicd6z+yjpzTBkbmOsZuOpt8RLxPKq5L1m5poMUPbkcOb90RpRuIjtHQ07UQjWnSQsMoTmCyykJjO0bsBO03keYQyktQEDHULxhOrnjgaEfUIsWeUBf2o8G6iiFtenBuGmx2i81S9ZKEWqKsZg4zo+4jN3T0cJ0LnsV4RNBXJNCGSgaGbSOYnBPOEoFaMTLRyQhEREPP8IiDkEaUVKsjo5ZZx4wiWV5hY4kRCoHOiww2XSc9p2eNVwDB2aJ9S1AGLumTINd+4juE4kbgIJydCNSInj5Ilj+0Bu6+ZRSFpIRj7LcssYj/2HJstZ9OIbju0EHTHHUomkMYEEwThxCLtGCNH0KW4vkX6lrP0grg/MjiBsBF5oJF4hqZhrBXSNfjRITPF7EogN4KZibDDHcZXPE4ToUyJJoO/aZgXGUOuEbkni0DdNYzvLKIosZMiOikonn/C8O2a4y/X+Ns9p6sSVyrm/2xPMH/LOIYso4/xUYv84oZZdUsQHJmdrhhbjXtTcZXn7NRA0waM+1uWUcgYtfTpjipUHKKBzy9OYTyQLWY0h4Fl8C0KiRUl0QqCJEWlHU+eKISqeflly0nyjPxJSHX1G+LTCb+peTq/4dhZHrqefBNz/cPn9Os9Ws0Jihs2xy9pX/wl48ctN9UVo6kJyr9FuQ3L2RmFKumGmEhoxsHAfEervsG0bxiUpyxjJm0h0+i44O39Dt/ViEjydF6glcdMjsD3BPGcajAEcU04btmMA1v3nuXpG/zhg+Z5cqfo8WOGJubyD/4x799/g29fUyRHkssA5Qwzt+NqlVLpHWUokP5I5H9FGW2Jg4b7w8TRnrIMY3Q6kuqe1k4oa7CriiH9f+C8xMgtwfxI2ylQYMzI6UXLdlrTd55AeML8yKvtM+Z+RTl8zDfdBfflt0Rly7PVBWb4Bln9ilnmGcwNLjmh7u85tp7T5SXN7Q55uSRKWpRqmUSPDktitUQIzXCYkZaf8TBe8GhjhvaRE/cGPXUE5TOO9RW+WbI43xFxTxNMTCrDWpBOIaaCeDjhuA3xLiKZN8Q6YBhqdL9maGPSE01bHZnuQ4YhY30XMVUFL04usWceK2v6+oyhjjD+jONh5OTyAvl+izukxIuY6vGB8NTjrSa0S9JgxZdffs4vfj1yFi/ZNx0qUygN9Uby2WXO3fotbx87oquI+Scdr9xLzk/OOH3qifo9d4ccHZQExQYxNJhwhjEVNu0YGk1cSmQKjwSYzjJOgp3R3B+e04UB3x8fqQ4h0mVUxlIbaI4DZZmjohTbGmZRSOQFofWURcJM17i+Y9+EkCzwXcAXl0sOh19Qd5ZuMxGnMzoJZZnjdU/XWaLEIIRkmnr0VCIajRo1WXrGzGi6/SNNuyW7UoxmYpgq3t1umUxKcaYIxwZSCLxHOVBdROQ9Sp6w3ziioyCQIQkas+sJamjNhGksmpjUSdQq+dDh637PnwnOTk5wPqT38PV3v+VF9IInV+ecnZe8vnlkMEDVY9ojMhiIQs1hc6RxnkMr8b2jOLkmKEP8w5HJHFkuCqajY3t4h/YWOc/RzpDYhidZwF+8OZI8D7n8aMOVL/n1LzcUO0UsYgoumA4jY9dwiAPeHwfq0RK5EvvQU57ExGmP1gfcZJhah5kJrBXoSONWMe/f9rT3LaUciLqBqdmhghwZxQx3E/vjgVyUBIOjlArudtzUW4a3ktkQ4WWCDQXXpzfssejtJdMwslrO0N7w8JsHXJuShAOfflJw/O2I22awGTlWa1ose+Vo9+OHlUehKceUyPSYwZKmASs9I5SeZH/EDwmuAN0O9O2ElRK/D1HBBS++ONK+b1nUIeXjFjUoTBYQn1i8rvDacpAnhOUVYvI8fPeSeZGSxhFy1yNGw/L6jMPhERm2HKqa0GgqE1Jry1ZofNNTyIb8TGDtRNhaZosedM937yMWV19Qdz3zKEMeCtTgSK3ADxI7pez3W8S+RveGYRqYpEL4lMgqpBvIrSNMDHM868bhG8fUa7IwI/GCZA25XhB2kpXIyM87DqbGyRCsx2YDQg/I7SMXJvnAV6gmnE1glIxx96GYM8YEscIEFu0GImVYhBGyjhnkiMRw9+0j9mBZzlOyeg/CYpQlGAbczZGEHOEkCScc9y31dMOFF4xHjUgF2ekH/r+WZ3z2kwu+r/8e36bstaMf7ok3Lac/Hok+qbFhQrgPUfQc18/46SefsN5avvrrA+dfxfzxkwW/DWtergdSdU2UHjh9nlNNATfTW4rTiYd5ipAFo21I/jOB2bxCvp9T/uE5TdHQyXvkZy3vxj0iMcx/vKIdAqo6J5QLmnpP0/2G008ERZ1TfdXy8WchDbcIJGV6S93d89h3JBd7BiNoxkfiYs3s/JHtIxTBwL55S6zPcP6e8Fwj9A257+lVzO12RKWCeDKUZ9e8/tKzzGfUU8a4VTxunjK9jwiiI2Ow4+3ftmyrlqdPn/J+0xA/n6GnN9jGU+QaKSX7o6c+zhknxe9uXpOoVxTztx/6FVmKlwPOHHCyJ0paBD06mXO7rmh7mKslk3EE4mMOQ0kx/iWr5Y5xCvAiwQ0TvrjBKYdoJUWyYJgK2u6RLIVD7Tk0AhNZTuaS49ATFRYzq/mdOfLV/D1fR49MkaSMTpHdHea+JlKfIu0JRfAFt/U7pBoR1Za4E6wPmqefPGEa70jkCeN9Q25m7I4hTj5hkJrd3StMeEFWCgI/J+gkQdky1pJPnr7ANq9RcUs49Ki44M2+Y2jn2H6FlCXF6RoZH9g+OLpXIVJJTqMrHo4H8gvP/OqS+73BDTFqKFGV5PIHCVq+YZocZv+UccwZY888MRwfD/i7gF7e8/TF96T2e9wgUeIF+ypmHp6xvrUsr07ogjvcxYEohOpmoNnG7J903Ma39ATIyRE1jqooEQ8V+rOYJDihOViEsnz85Iyad3TGYGqBjEPiMqAZLAQpj3WPLhOOUvLmUbK3S+761yj9gW3jg5B26HHjB/GV9D2R25FFC+p+TvN4IAkL1luFLwJMFLKvLTLJ8XWLMgbRC9SYE6mOoXFMfUqzUyTFGcHQIvFEuSYYY4pkSZvuwXn6dU0dCaKzPYHpmYTD25JuH9NvRpJEEU0WHRU8PFQkkUR0AboMeH6mGZuMt38zMhPXHB46oiLGRD1NMlD+Wc7m2w35KiMJMx72DUooAn7PnIF67wlkwmFX8ewPn3Hy5JTvXr9mfpGz7TcchztWp1eIpCN2CqE8zjdMfUUYFZTzK5r6ntv+wBhNaOuYqhHtE6QQGEZMN2Cmhh9/csbt//Rbyr2kXxiW2Rr7WqJNydXHF+zaI6QLzKZHaE2uFKdnIw/DlmY3kBQZy2ctkWl4HjSIWnLXxdRK4ZUgVUfqo4TVGWKqWQSOoLU8xo4g00zqg4QmExvK85jxVwP+McEoTzsqrCuobcXUeWxu+Of/xR9y9/o9X3/TQeugG2kng/YDeRwQOsfnn57yVz9/jesmZBCipaNYhbgsAB8RLTICYQmkopABtjHkg+dE5ShTEWfQ1SVi/8Gk6GSLPViCJubJ5cS//L9I/v1/u+e7X1/jbye6xuNXM3oxkZ7OMcFIuztSH19S7SZ8GOCVY2oa0iQgWKRsjwPJE8V8WRE+RuxvYFMIhFYQtnz6xRL31S3h40SWTxTOEQuHDxvEYcbjt3umyLNKJHIRYLwDp9BtzFQ5VDXitxFjP+J9zzyasbo+pzkc6NcdWhsiY3k6z6l2AyfXC3IS/N6xu3f4ziNSsHNL4DU0LUUSgRVEwN27A0GeI98dUZFGbw2xBuPAZQH5x5LI56y/3SMCgXUpZ5Hh+UXN2XPB3/wbx/xZTCzuqcdrbtuOx/uaZ70i8p6zIODwek8aZAQ6ZvAW//YBT0N4VVLfGdrJUeSn/NG/+Anvdw3/4f6Opz+cWP7Lb2As2e8bksDRm5j7+gnPfcXy5Jz9+JZQrPn440+5f6043kvEdyvC6RTnGsyrhkV+QuMF6qrG8MBQKYqZZXd4T6VbrNecnJ9z+eIMG9wh5q95WT1wslwxmQeoJCYtUMmWB/2OZmrJwo8oHhTn1y2qyOh9R6ItTz61zJYHmn6iyB3aHUllTJEvSNMW9iHHx5DVs57MwXGICLMT4qXCTN8joi1JLNjd7LH2CSezJ9w9VsjxkVTFuNs5/fsZFz84oRruqXYG42J+fbMjiWeoVUS4bBFz+K4fqcUGvY2IvcRWEWHQoSLN2GaMzRntsGd394aPf5LwfitpzYr+MBDpEmE7BAYzRh9OhRvDoRW4cMXd3Tmul6xfgpQjf/J5gNDgg4KHd1vSfMB5xyQl8xON0BOqq7icJzy83oKNiJOAMDc4E+HcgJA3VNE5d3zES9a0YkAnNZXeoBODXJ3x7d2B8880drzhRM2QGEL5gItz3FHRbubo8shu6tCnp1RTQLfpiE8qKDbMnww8bFLiszVduEFkKWVWsuktM+O4uVeM6hoVLQnOI/JC0x3mHN9raB5JJklgYw7fBSzKc7JVSLupmZuAuStJpETtD0RaIUXGeZJzumzxDOzXgjAK8KcOLSsiNLHcMJaOj55uMfYbZNiQJBmIAGEkJr8j++yeh+OCKWwhzBjdkcX8AEzcFRmb2YEkP2dSkoe2pxsDrkaPNpZ6jImlRjQD09oSBprNztC2EAeSMAupto7aGJre0baOwzDiRcG+eku9rVC5xZWWbhwY2okwBuMmihmkoWDcTOxeRejZCZu3IcYKVHqOrTv8ZMlSGMKCd28HZDijriwuCHFTihhD0B4rQ3zVEoiQXnhkAmNyxC4n7K1l4eEstdzsJsajJPCS2An6sSJeOCbXEBUFgzEE6gnVOwhbT/Si4Gxe8rvXIa9flZyXMWEyUj+mDCLFhhWLdGCX7ilLT3Fe8q4fkFNKFmW/3zBQdZZ+X3GYjuy/+RWv7n7F5rBDCM3MNwRyQXP3lsRAHs6hMHTDa2TfEzJn1A3HeotHk+YxZaKRfsS3NTKMkTpg6vZcPE1JD/e4+zXZGDL1IVp06GJJT8y3Nx1qZkmu32HPG6RoMW3A6fnID/644F/93wcCK1ksFf5tzJOnhvf/845ofYW+iglWGrO5o9cV0eqSZIAlksYZQqfJbMowCZLAkEtHvXmF2Hti+Sld37PyDa3sUQhGFzObB8QBfHIZ8/CqpWt7SuFRkaLWCXsf4kZBFubYMkVfBKyKHDEFjGIkjkbU6OntAW9BS0UwTohWEHWaXh2Qh45Ya3o3oa8q/jf/x4lv/taTmx8y3QYcs57F2YHl1Ud8/+/OsVZztA9U60fkzEG/Yr8bGQ8j3g+UZUi0FBzeNJjjiDpR+ElyGDr+6F+mfPYDxV/8zUiTpnz2s5LH+o7dViLPcsqHgHzX8ySu0LJDeMneT6ix4XjToU7O8f17fvRPX0L2QFWd8fLnMZk88ix3/Or/7dEPM6atwg0Di3Cg2UFYL7HJAcueeDtybZdk73M6c2QyNXry6CTgaHqCh5GBjkVacGgc49Bie0XQJtg2Z+SCIUwRuiWaOUY/Mvsi5o/+q5xt9xtO70uab8+4/z7iz/5Jx0m65vttRlTCn/255dnHPX//V1uO24IkLkklmOP+g1ksjAmWK3waMNgjfn3gLJ1janAXgvhpwSf/5CcsP/4xx41h6d5w3P4d6qdrWPyakAW1GQh9SLp8htYJT+Z/xng7UXWGm4eBsTvyBz9cMkzvUWrPzes9oo1Y/KCls+/JfjawdRXBbEkkJHETkkpPU3vmaiIZD3TjDlVA4ybCbMBPB9RwRhI5wjjCqpx6iEiCjFm5IhGepDyi5QNCr0nPVmgZIIYBohc8vt+BykmiW1ZJRK4TLqJLDt3XRF6Q6iVh/DFuuiWMc6zecbKwrLcR3/31M66efAbrlyxKQ96lHL+7ImkL9ncbbHBHW2aEqyOEjzR2Ij89oUkE/e4WuoHq8Y7rZ1ecnj35j1bGW45Dzqu3Eb5vMG3NZezRQmLGgbGS7B+eEfmWP/+nPVO35/U3J0TBCfNSE/GKd7eGbrTkkeDiizsWWUo7HUiiE0r1T/ly/xdEHQwknD7viZYjvZuwjUDKgFmWsLMh4+S5EganLS62qGRgPwa82b3g3WBJkpAoyOiaEZ/k9GFEtdxy/uyR6/gEcyzY3BumWYVeOCKzZO0UvRq5MwGyFJzL5P/P2n/t2rIkWJbYcDc31+5Lr7223kfdc8/VERkRVdWZJYkukt0g2AQIAiT4yN/hCwH+AF8IqodudrMESmR1V4rIzFA34qojt9576bVca+PD/YF8iK+wYWZzzkH/4wKar+kcAzOo8D0f6Tyg+TWt1SH7O8zVmn3msrzsGJweolyd4nKJaRrUH1rab0NkruMPA+Ib6KdA2dE1HU5gkNcOtz9k9I5LRAmB6RBrFZYI0YqaZK9j708gqbHcis7XUO0D56/eU9VbDDMjUQ1S6ZRNQ7FZoKkRnnPHYT/k+8uQ2ceH/PD4wPGsIM83tIYiV5JtvEUzwHQHdIWE1iCTNfPNFX5Q4U9KVFezvdty/PwAvdxhiApLKNosYdITXH6IkYEBwkXba0RljLQ7xoM+6+yBLAPNsJCawFIt/YGBUdd0mqKpa0wJ9qAh6SJUGeJ7U3ZJQ1ekFPotYiC4iVb4h8CoRVYCyy/JyobClAhV405SdDdD+AaJqmm6/Mfm0ZOAoNzS6DFhOGb1oUKlGidHklZGCBmSbBRlBv7YI9/tCdCZDUxMQ/DhdyXL/Zhq5LDvtbROimmYmMJClQ4FOl5/ysAHx8xQnaIqNUT3R84MdE2Ga1RoTkdjVdhCMVaQ72KqIoO7PePpgEEvxLAU232KSCoCv4+WS/J6hxcZbPKCMAxpkgqn11EUHdLVSKKKw6nLs+EQ7Tc/0G8jMEJi4aDbNsEkwDiWKCfD/fwt7pfXZHWBEjbt7QilatyDGaOTjwkcyeSjjkTfIU8EtedQaQGGrDnpT0lKm9bYQqhjLnxsq4/WE+wK0MoNL4efsLi9RW9MqqTCbhQqvkLoHU29w263GAzIjSGWpWFZGww0urc1J5lJp/2oANVag8awoG+D5dD2R2ROiqwjZNGhpTFmu+PYOKAWglZaUDZUUU1b2ViiZDi0ftTIZi2B2SOtTLK5oliMKLUJ8kUP3bkj2zZ8/GnK5uqa+ZXNWAs4oUOZOW93jyzmNequwZceg2ON3YdH2sJBDy3cviCLEjzfQrQNV+93GO6Q8fmA5bLk5OPPaaYfsI8DBvsAK+vwNR2tTMg2KQoLvYBWaajDiqOfaIxmO1bNCqXBk58OGR1G6MLnth5hPYxJvkmxgy29Y53rSwthaOhun+HJIbtlS2D7bIsF/quYIFQUtWL9sIO3ffzSQzM3bJewKwxK06MyHXASJs86Ng81jduROzWjF0OMcMHZ5/B2saWQfb569RX3RZ+XTs5o+Af0rmB1rXPxmYV1uqftTF4dv+J6YCFGHWm0QY0E5WBD3depZwH9oU2v1nn/lxb12sP5bE87S1Ezi3v/A/MqQ3NDvKM13aBgYzhMXIOk25J3GmmcMXMMxr2W9eoPXC0yqniI8Hxq+cBrfkv+vCCyDepnJu1W0IQ/4Lsx5bSgiFyE5WI0fZwipd/vMx3GWHJFsnlkk+yYuAMGliCQHesiJy82DEKJ1ehkJT+CXLYlc/pcBBOy/HdYlobXk7i+hqKjLW1e/9Chded0bsHZ+IC2tFkuY6L5gv6kQRcGbeKjhE1vHKPMAs9qOXMsvr55wfV3P2W57Dh5NaRxFuwtlztvwfv9nu4xonW3bK0E2duh1BJhfsT13RpP/zGjEeCz3ES8W+3xvnrk+GxLbVgsswv2tkNtPNIfSVZORSENtGSK9mGETF+iOSsGZsI2PWAXe0xmJ2zyK3KRUJuKVq9Y7gyOZx55MidtgbKkrkqa4gLiFMs3MZv3WGZN0yr8QKFUCYeH1BsT14Jhf822WdOWYNkQTmOK7IbO6KGZLWkWMRr2WKcal48Zg2HAvtvT9B7Re3ekRkltCLyeyXyp2CY20gtYJzt6E8U6z3lmZ8g2ZhgcYfUrpk6GlrZAinQlwn5g8nIP8Yj2JmQpO5oqwo5aBoS8ME/5el/SswfU2z2O5aA3ksoB/9wgVyVttuXgArzeA1Fg0skVh68cPPkNTZghzJa0WtJtAqrcx5z1MAcauvkHhIhAB60CyzZJNi1mEKGVa/I8Yzg65LPzMYvKRTJlty1IUwvhpMjQZ+zE2HWO1g1p6xxzUFB2DQ9uD7d+oAtdrl6nWNUI+9BC9vYYrUZSNvT6I7JsSTBSaJ6JZZ6SRWvydIuSGagWz9ZJah1P9pGywRUJoSiJIklVC2qZYp4oalEgS0knp6jOpdjHCMMkbVKkbqLXLXnSoZkaVODoLikRUoPhxERVJaXe/Shky0ykY+LbgmAoKBY7apWjtBw5hM410ANIW+g6E1Uo4rjEsPsUu5gj3yEUNc264epdzeilh+uUvL7acjDy0KsUXcXYsUFRFjw/tPCtFL0uOZwccZ3pdMs/MgxM6ajaGFNrYGHQ7Gps4dI3JLrV0DurcIOYcKhRuBrV9R7RSGRd4QmXsugwKx3bjOjbGSMZksc5TeswNMY8PxZ8/FHA/S/fsIu2DIIWv5SMC5cmWaK9+Jaz/6YmaRWjT3bEYk2ylKhOMTrLId6i1Yf8g5/9z7hbfYMz/SumJzq7hwn7zsW6yPFeRNzsNBrNwjI8JILHh5QrkSFdl9aQuCJHu78n/8OcAoPxUUBZlnS1gcGOplRYlYfvWayrFNPpkZYVxqbCzjqctqEsFHVnYrsavq04u7Apoog0sTCCFLFqsXc+7bwmCAIMMyPfaXRWTdN1yL6DfTAk3S7obpfkhcLWdUyzQzN04lznq39h8svXW+yRRa895PIvJAdn13z0ymS1qPEKDaqINrZRCx0zNgk9/UfJzTqmKxxcz2Vy4hG0HoqWQq5xA4+kmCIMgyAUCH3E4fGUy7sP7Bc1584RWqjTHRyw0lKqqGC/7iHNMV2lYakMMwCcEqeU5LGiak2yTrFdFAxf1lQf32C+7JhNdIr1jmF+RHMjSS8zGueC7uWG0nlk+KzAOLzEcPb03TPqXx5RZkOq7ZaDr465+v0ax9DRDRM5Tjj+hw5itmO1SYhky/auoBpaGIOKN3aBvjxmedXHcWw++dPXOKbF/P4BN6xxgho53LMbplDZ+OMOzmuswKWxTe6bCH1yCdOORF5iDiVnVk2X/5T8dgCv3sP4ktwN0dyGcveWyroh93c0Vs3k+E8o9z5l+tcEhxMqpaOamCzuKOsP1PsNpjcGK0eza0pRUgca2yojdnbIsY9ZVphdwiaBeF/QiIwmEbiWj0mJFeZ0dQ7KQtMMPFvj3D4hLRT5PkXFPWbuAZoes7veMQiPEOEYyzAZjncklzm+cUIga7yeRtfUYPiYbUOUxBxOjsiKiOXVJYHXp38YEAxqykJQ5Ydk3T39ox8nhh0HNlvJ7ddTktJnfn9PN5gjNR2zhL2xJrEnFKmB7ths0zWHrUvYL7i/W5KWAbI3JK8W9GYRh67D6jFGBStEvyOvbZZrxfhwzL66pc53KEcxzxNKFdGFHaV1wKCWBJ3H48bCcXRsUbGMHhEHNk2RYnkJg0mfVVEwHXooqSiKCuPxNR85TynVkNPD5yz3OYZXsd00aHlDKyUPCWhCYzYs6LSOZG8QxR0fn88QY4/yXUXbxMR1gu1prNuGZSGpFTSdRVmN6fkNWfyGwBMI2wEjZJEdUzcCRzb0LIVcV5iNTzhWFHnHPqoYhmOaWNCmoNkljmGRxBWaZtCEPR4MhZRbxsc+wmzJqdDjBvmyJgxGXP2uIXQlWZUxOnNxTxPWdymRXOL2U4yjEi2V1CJGn47Y7d6xVS5l45A0K3xvys9+8S8JDxSNecXD1b9Hl1BVIDSNuujQJGhmiubUlC0omXL8rGT+9Z4wOyeubAw3QBpLVtcFnuzjBxVV3BBqJVJeIbSKedvDbQX3NztqI6TOLJyNwyDco1HSNgOKfUdn9pEdmNJiFW/JuwXCkrQYaKbOZlFgun1My0Ajw+1cikeNKPLZ1i2Ob1GTYzcVTiOwhw6b24x2b2DIDsvV6VkD6rxguyoQto3taOzzPeHYoND3mPaQPHGpVYkuNERposwfb+9mVtKik0cWUjgoraRzFVHX4BoBbSTo+xZxmZFvGsLGwep8zLgjiXLSwmBU1XhxSr1wqLQBjsgRdoojHPRCMOs1TIOComrI0pgmGiFL/Y8LA94iJ2w9NAXxNsGxNbqgQQ4DGhrMIkVUFdUypfMc3NThaNxHMxvy/QJfSE4nY7ZzE7WEqFvDDoIzB1trOLlwufnd7xAbC8WYTB/RCoOmTRn0GrLmO3ofGUSPJbUm0fUBw2HI5mFLZSXYRkC16FNWCZZ1h27/HYkT8nh7TO9nH9Ef/Hu21R55fIxnZPhU3F7dIaeCYAaFqmm2EwaDQ4rFPX5go1OSrh+R2oTUgvEri14yIv0+p6lSBraN5eqs5nu8e51uD9ZEIJqOzcZE5TVPv1B89NkbNiphPD3h7CcvqL+9xSiW1KaHKDyMOsVqwAr61KWGHXssbza4/oBWgKbr1GZNKhN0d4puH7Hb3zBzT/C1t5yf3nC/1Pnuh1Oi35hws6U+1NC9MY83Bmru4sePhF6NaB1SDfpnAxSCn/xsyO6yYBd5BH6CNCdcXka8+Aclw9EDWedSWgmm0ZGlOhzssQ+uiIJDvv1XE5K0pf9S4LQOxrJEcyFLSpKyoaLDDhtsPUKrOpraRaNDyUeco5rcMalGB4hBQXmfIh8KioMPRMYdlbbg/E+esXtU1NWWfZxy8OUB7ZnPfH3H3WiB/oWNW2mU8wH9QxPzRcZvd29Z9Vps1yGz+hROitBBNT0GRz2GBy7B8TV7fkPkScxnkmWh0f/5gk2Tses0ciDjdxRfTlCDglV9x6oz6fmSsZ8ROg7oBUvZEE/XqN6Ak19YbA2N6GHDvn1NYZRU5gOaWdDkkiZNMdwBjpxhWCVmCJ1hkpgOm817wllDUSbYvZCq2KDTkSmFaUmG3gWr/RZcg7xsKTpJao5pVxJDLnH9lMBSBLbOLhZ4roUmHAb1GA+XzQq4PGTWjZgNPQLXRoslRX7KwfAlzfo14vkWVECyCLBNSbTYYVkOcexjew1ur8FqOx6uO7yTLToxZXlKmQ/Yb12aKmZwtsaVFWnaYVoaUe6h3IoyeMQMTGJA7vb4joPyDYr9koqWXt/GK1z6jclnx0+I308o+gOS9hZn0CPxHKRbMhz1uN5lWLHLoDdBGBV1/QNNEWN7Douq4DGqUfaObOIwv4GL8wDpRLT+nL5rUddnmG5InQ5o8veMRorBcML8h5yh/RIrKynjFdGDZBj2SPWWTSHpmzoqV1DbtDk0uUPb+UTZAcM6wTYSaDWkK8lMnfdXGQNzRGT4bDYd9kBnHiUs0h3HsxFGneFYIbJ1qBOJa+i0mqRYT2iSAwx7Sbx6w/FBHzdv2S1idtnqR3DaNTTdPaEaEjp9tmuTssqwhjZK9NjoG/SpTSJyms6krnPOjmpU/ze4h4K4/Z76YEdmerTpEYUnqaw1mdaA15ANHyjCAt2zEVrDcmfyZxf/iGHQ51e/3WBSY7gFi+5fc3Ob0eutKJOUDrB6BkpINK1CeiYIn4oYQ1cYtqBJtpT3LWUyoz+bsIoucUctVb5G4pPuSmSYcHw0ZLV7jW5U3C43NFmK2TulKm3SrOb7DxE/eeVjyJjQ8GnLPbJT6MrD0M/48Ns1VThlHWUURY00fURZY2lDai3DHYTkUUuWSNLCpDMqHG+I1+U4TUoWC/brCmHucUctsoM2MdDFDj/MSWPJJJyRcY3mgu6DpdU0yRqvtFB6BZqJMktcT9GliiLXcDTBdNBjflkisei0ii5w0VqfNq7J0gynZ+BZNdDheM+Jrq8w6ppSCbbVgpfnfS7f7cm3LoYQBMc+wmwoVjmyMwmNAXXdsLzraEsT2/77OYz/3jDQaECtYSqTAz/ERlFUNWksUFpBoypevbwgWz+SeMf0T0Zsbx/YP6wxlOTwoke1qyDrGJpDurRgnxZoas3FbML+3Q8Q7rE0iRrsUJM+Zq3z4iRGem8xXEVldAwCiWeM2KQ2ouuwhcBzNUztALV1cUZrum5Lp5U0247Z6B/gnh1zt/l/sW8eOPooZXK4Y7MeYQx/xtkXMS8/ueV6ZaAen2OtG/pnI3IMHuMF+eKOJt3QGjnDfy4Ruc3NvqW/76OVip4IeHp8yg9/9w1i4GANvB+nlT2boCsZTxVaFSPrkItXY+7nDk/PLhifLnh4fIdWzNBaC32j46ZHLB4SGr1En0DnGcRWRqbPOXw2wh2fknYJ1iFsly317Irp7ArHv6FMZmxeW3hMGJlgVKDrPbTlHcYqYaB72E3LbpvR9FwOzg7pHW85/pP3dL6JmzzDc8bkjw2WTOnNSprhA21hoWn2j25x/SlVJ3GHNvfWHdGnNr0vW4KThKD0mX8jMcOOqvcHKvOGqrZo6wDVrHDsBne8ImnGaAXUWoEtZ7RNj9LISSdL1JFNJBPixQOjQUduLsFSqLQmGGrk1hX7nkQ9q0jqbymVS3/U47k44pvfpcz6DsVGp1AlXbvH9G2yVkPTTdpiR6tlfPTpnxLUC9q9S95oXJxIitUcN3TZb49w5XOkYZO3rzHGMblxg+gr7GrAk+NjJt0HNqtDohZ+SHN6L2qm7td4zpLK0Tkxx8x3GZ3W0nNP6JRgGy3wrIayWzGfWzwzj8jXEbUB+lHILlnQ9xuE3uF2JVm84ODsU4gSqmTBOmrZ53v8A51KVkStSWNU9H1AZNj+Cqc/oF718HYhYZswdB2eOOdEy4rtW4W2O6I/fkG51tDW95y4Y9JYYVxu8aY7pEgxg4KmqdDyA9rMJTdLlpfgqiG2+57V6jvSVcj04CWunfLh2ufg9II8WvLqiwF6esXrrzU0t6bOI7IipxpsSBYxxQqeDy0skbCLP6AFYzBTDFOhpz4X6gU/sXsEDznq+nPcp+fkxf+Hlf6GfSIZOT0C74i7dYwfZjj2ErfX0WYpTuEQmM/QAoP7bYSk4S5zeIiWvHzynGXzFUXxG1odPENHNz2WsY6lSfrDkFaziO5spDYjGPqU6T3LcoEZjrG9CZsmReu2dI2gES6eL6keD/HrAQ93PR4GE07O1hjdFmnWRPmOqms4HGTcXQ8IrB56WVOnFYbqsLoW32057+lU1RFRUZN2eyyvIo52oO0x9D1NY5CVMU/OHZpuT6VZROmeRoO0y7Fkhuqfc/U+wbNi9tsN0gyoMGmbllaXJJsc2w7RrJi4W9OFBnWzpXu1ZYHDLotx0DFUSTesqSR0ToPtmRhpRdZBFtWkmaRucjxToNx7js4azN6OPDIw8OgqA+ma2IZF3VXoHSjTIFs1eNUQQ3dIdiltBqZ2TNl2aPslql1DL8X0FuTenMapKQvFrCnBFWzTHEsYGKZD2wg0S6LCmLTNeTdPGY0Ek6ClLSVSswhmT/n6bUu6nZJWOhl3dKak1i1qPUM3FF1R02Q9ykqjFDG6LLCHklJb41otjgnxwmW3Fpy80Gj1HWUZ4QYhBhVFo2H2XVp/TadymkqhOomFhaEp8m5PXrV4joM7MNFciao6tARwSvpTj/17k67yULZOETc0dk0nM9yhjtu36MQGPzRJ0j3xVuE0DaYPvamBo2KeTeHuqsAbShA1ldFSyZZWOVg9yeZ2ThJJmrojDP7IbQK9GWM4JqUW44YD0jjDGZYcfbmHMOXmWnJ4MuOh3dD4JXX/BqkarHJK19o8LguKbc74JCTa7Wn20D8y+OpnS55//p6/+IuEoumjHcz5k3+8Jm+X+KaB0T1SyhTf7nB8hyQ3UfEMqaUIK0JzfWypIa2Wkbukrb6mrm4pGp08GfLi5TPudt9TxgUXn4YMzh9ojQ6TkPDVkn5oYwQuxd5EmRHGsMQZeOC7qLc+1uycXn/Ok6OCzr3G1s8JfnUCrydIsyBLdZJNgTqsEFOdqzRHM0wGhzW9dkeZBVR8xnbdYXh7Br7OysgYv8ywX95jIqizc26/yzA31/R+YrKId3jnLpW3Y9S/5Li3pOwgLiZ8ehZh9X9NZfcp8+dU3Sm365p4+wmoE+r2nrbaY7suxgvFiTvkt//9GkuTtKKm7imkZaB3NeGLiu3wA+orn5OBSbLKcJ9FXMgVyEMeFgWTSYLSTVopCEc95tEOIQ5w+hsO/sEe6pT7bYsTbxDlGWbacTEWpFqKris0vUBzeohApyteY4gCCslkYrLLU1bxIWG/hxVMKfSA++SBur1n2AuYOWds3D67/T2GgNqOyJ0delrhdgo/lDi2Rp294fTTCXf5HUHoUdSSVN3TiBghpqhWIcoFplOTdd9j90vq1GdxuaMa3+FZa0Jzxs3e4TodcOBa0NoI9YiuV4zDGSKLaTKdt2/6NOKc292e1NR4MavRnAVZ0oDms93csV4pxoefsE9rmvwGI9C5T2LqNGXon+JVfZrVgtaseWiXRFsXqU44cCyq/Q1CSrpWYXUVfaHYNHPaXFHlfdLSQfpThHGJdlTTlBm4DnkeYq19ToOa2WTEZVQStUtkoLFNBZVdYU8KVvmYZq/TP23Q3QWTbI2sduRti+G17LcP1OYBnnnO/OoHQrMjkAXR+oRykeBJj3TZUewL7t/UbBYVB6ceWVFiqCNs7RjX/w5dT+m6CK3Z0nYP1L7PzXLDT45HREnCbp/SGTbhwMXFYWxoeO2CPF7iezZ1ZdMzeySbN/ScQ6auz+pxzn6pUCcdtlkyrFNsaVGkNqLSaMWKTbYhLQLu9jbKSbmbX/EX8SODfodmOtRaQ6nN8Xs2Xdun3BfohkVy/wTz0yGx++8wPZ/x1CVTv8Fzn9IkJlFqsksFdpgQ+CaHT13iywPqVU5tVShDYgUBiZ7y4TrDMA26sqbXa2jiFjMLOS1HzOsl+XWHPzrC713wq7+6QR9ZVLogME2kb6Ml39N0CbquaLuUfSkwejOyJqBqJfE+IhAhwtQQmovZKWQ4p9AKClWwmJt0QmLrNqvNPbZ7R18KtDYlN03sYMo63lJoOa2+xB73KPKKMs3xTAepG2hZgaFyArdCTi2++/aW/+0/+y95u/kPNM6GXj9FyRTpaxiuh7BzPN/AakO0ao1yoc1MrIcp5l2HMCTBKCSOX7J9HGKMW8LDG6S2QI0irP6OxhB0hOT1hps8wugcMG3iNsO2NXyrQOU7NnWB0BsG5pAiFehGhWNfoIfHvL5acH/9iKmHpE1Hp9ckVUFe5ehI0ixFSptkvacX+GhGji4brNaliDJyM2N4rGH1TNpNgecM2Wc6lhQIM8d2SsrcpDBzWh+6LETTdLoiIS8dLMOi2ipkqxiMJMHI4fqmJZl3GE1D4A6odtB2OcHAxlIuq7xGjgz0MxtZVcTbHbbwcdwcrXlkFxtEmcMwMGnLPXkR4GkGnp1CqFOOHbTaoh4pHmqfv3xfcP+o4RwIoiRhX/79zvi//+jQk47U2iCObnHPM9J4Rz/cImd/QPRDnMErtHqPHT6jcVIS2TEaHLDQ7zn5SscLCwQN02eKzaYi+jrkmecyMS7pblvM5CNKveb5RxEHowXzdYwbOMwfPhD2HXzfY5UHrNND9HyI6+wIHAPp9hFOgyUiWvmB2upRbWK61qLnXmCPVujv/xNa0tIbzkjjJY6t02W3jHu3TPpf4FSfUS8dHG/HNrnFEX+K5Q4Yny5xrAp9mqJ7C3YrA1EN+Oyf/JzdccAm+Ybd5Gv+4BRon9q02wnuQOGPdoS9Pa5eUy7G3N5O2ZkeBy/uOPO+I2mfk6s+Lcekrc3oyQVh746s+57ITLCrEXsk+0JycWBTIYnnLp6h43k12AltOeVEvaCK77ibP2fzYYoftNjPY9zPoTtcU7j3qMMJnhYxdY5IRUpnV+TbS2ZfXdF7FlM6GaVToA8qZqpGNzWaXHG7SjDtEard0VkDRmGAneVooibvebT2iqZIMfyaIDhjeSUI6pK+baPQUV2PptZoizVdnJBkAhVa6FVFq3d4TkBZ2FhNiFNPUBrUWMzXFaLsE1pjrNJCbVvKbog/dCmrAq2uKTILbWPy7KDB7DYs0w2xWbCKFGkpKbslpciQfstum6LyggNbxy77XPgmQb3g9c2GNushq46hD11msPndAHswJWrm2KrA0jWyJqdJY2aDhmyjs60cxrMtVntFbzxlH3d44SuCmcGvr99yeZ9Q6wZ1vkHpgl6g0eQdD3crRuYFTw98/OwPyPGaD3cxuv0xL8NzvEYxPna53d7RLAz2V6DZDqGj8/JCw14kLCjICoFGhmelKKUY9Xpk6zE0fUI/ot+zUNkR5X7Nxoiocxv/pKAR12STmrI5ohnMyKcOXfaB2lhybg95XB2RVT1WacOF77BM79nKeyxfsrs9ZPXmjHhVII0Ca94g9R52UaKrgpMvTbqqINZ3eIcGUbFCj1IsoGd4lEuDvSkQhw26f0AVNexWe4zQIVACx3PJshtWco4Vlhycdbz78A7DqekJDx4cPjr8nHTznsGoxDe2tF1J29Z4liBw9xTVjmZrYBgut2924B4igpqovCM/SEhyyaEzRZgDknhDaXeUXctABFTNkOC4jzsMuY9+lLzYRp8yNdhk35JUDVOvRWQaXr+lcmuuy18THb1j8A81+uOKhzIFM+D6uuFuIemkwyqxqcwQ2zkiXd9gbDd8Fnqs5jYfHT5j82HH7XXDWHOZHdXYskK3oDmGeRJRKw3Ngn0GeaxQRYHKfexc0rY20i5RsUtoQa1aWtFjt/dQ8hzbgDK9ZTbcI+s9RmMxMDIOxIh1WdE1DkZX46Ul4xPBsjAwzT75fktj11Q9ia00XAx0t0INYtL0NV1Tgt5jV5TolY1lJGQsUS4Id0q61mlzDce3aGrF7p3GtD7DGLTs7wQfvjewc0lPMzBESaSvKJuUXDQkVUPQb9FLHdDQDAuldbRmSaYKmrphvzKIc0FvNqCIz3CqE+p6RWaf8uHNCsvzMIc+WXlNa/axNJckniNtQS0UNBpV1UFUoe0UhgvBgYmpFfRccLQG0Qg83ybwSyyRMHMNitJik2UI08Ewh4SWyTZdgmHRq0Yc+BkxJbuVIJ0HjEPJSGnYjxriOw3PtLD9jH7j0u1rXDcjGChu3+m41oB6mxGoGXq0Q2UR7Tggtyv2Zc2mErSZwcnUpWnWrKId+pnL8OmA1O7xw3yN3rm4vsZ9qVjkDtudpGe3GMOGZvdHhoHsVUTemTx/eUDb/47QaKk2t4STPQ0Vk17O4vcL+sELDj7z+Xe/vMYpNY4OPNzDewZnEq1N2BFjnYWctTX61T1qH+G3YyZWxqufxHz80Q9YvTW0KcrZoOjQRI5j9hDRCz68HfD8yYDQWyPaW7R2i2sbyGYDRkRbp0hlgneAJRVF8v/Gst5g6g2mrMlp2Fd7/KCh54Jm/JYkHwMh1uAvcM8LyrKPM3iC5X2NJa6JmxF6c04VdzjWhHy0wn3ylir9n6iqe/L+BTChJuX4QiHdkqpSGJZPPGyJk3e8+vJnJNm39If3EJXs9jO2W4nnduRcUfU2WNJAkyarRcVk2GfW6Ywnfa6+MXG6mMmT9+A+UhOjWGP13nO1+YHECwh+umc0KEnSmG2oqKsY6e9wn3zg9OOKOl9QqII8T5nJjMGJziIqEPUhfs8gjpcIP+F+EbB+vODoRONgdI/0dK43MXpd0ln/LUfPOhZxxb4G00rwvJq77RxtfELQJCjZscsVeQmOr2hljjFqkPoEpXQqvSTsu+x3GSozOR5PiNuCsG+xv9sgcp2kFTRai97+JZ8emozvQu4bSeB7rK9LhPac2YEkzP8aqbXE3Yj7rUWLxra8obBj6hq0rGS7XUAMnzx7gXW3ZyAWmMYb7OmYx9egdTWNKon3BQKfLukQZobdbNC0DI8c0jVSG2AXIbpvkFhz5HiHY3lUdxa17PHL+2u+XYQ08jk9d4tRLPG9I0L9nGwe4RQZpueR326JswXPPxfU+SnlbsrBRxGOcU9RjEjSU5rkGgOFIUZs8zsOhi79sOHyVtGKAaLNcBqLmeOj9j0WDwmfnXqcDUzGdk2yiRj0XZZxgu1nHH+RkFclN7FOFj8ynIF5MCbbROR2w1ZvMHKX918rhqNTHtcNRbogtgKKqk+6meEdX7BJ3+LrJT27w4w1tmsNZ2TRNRuq/BHDb9hkMW2REB7Y1FXHk2cv6P6iIghretOa6/SRu3SD7lp4fYOqyXnYbNDWMS+PT3ANG08PsfINXWzRG9l4Rw11eUN622G4Gl0dITSBbdpI0RC4Frt5S9e5VDJju25R+ZIzd4ocQtSllO45SWXRXN3Rdz30ygI9Y29C3EpSOWdVTiizl1hySXKvc3UzoXdk4YuUw2HLot6xKGoOgpBNlDJfL9F7Ovq0oREOhn6Ays6wa4vvPixIrQqGC2Z9gdqkhEcrxn7H5uEJbVqweOcS7l8w0NcExhVeZzDfZSTtHtseoic9gtwhKHV+cfCMqfE5/+E/zdkG7zHklolXksW/xnNOidSU7WrA3VsL4Z3g9odU84jJtEYIhdWV9ELFLomwHRezmLLbLJidW2iqokhLDofn7JuUrm1oJVjSg9pm/qGHF1hkaklZGbSWh1YdMuvPkDLlfv8DSfoW3c2Qw4bdqiRpQAmBmk5YrV5Rp4pC7UiDHNO/pZWPCD+iamOyMkfYgq5tqI2YjgLL7miKDXQa/YHHblchpIZmGujCoNyX+MEhF/oTHq4LslGKyLb4taTSQ06sJ9zOUzLVIAyTtohw+wbSD4mWEdLz0Dc+HgFh3dJUa/Q4oX9oIlqP1abFaCXdOqXX3zCcdYQC5o8KpZts1jsML0DXFHGV4Wk5WDFV5dIKsAYGnmZh3nn0dwoOKmaBzU8uPL75bYphHHP5riGPerjjPiJLKfIGUwNNDygbi41yaF0f44mJsatZuwllnGNONYzBms5uWN9lFI2L5TsU2wSV1uSbFMuUYNZYZoOw3D8uDJitQA9sglBnOh2TpjvcYYrlgSc7bt/eYdUf4Y1+rNMNLxzWH+4JDIHoFFFTodUf8P2E6fArBq7Nw2VD1xooLUXX52wWGebPUjp9h+G1NIbC1MG1FbaVs74r6XY2TmDStRG6k6CzxzHBthuKVIBuIaSNwEXWChFtafQ+zvGC1voa29lhKqARlLXErA6p1hm+vMIxFnRuRyr+jlR9T6feojnQa05pOofWbbEPSqLqryir3zPsbxlah1ytLNANphcOQV+jVRaqXEHXobUarz7T2OR/wbbumFlPwCpJi4yLZ+fs0rc0Ucyzk69Im5i0vGYQlhiazWTaoKsUs7LoTWpC+xalHunqAftkyFuuSPU5hvMD7kmCci3Wy2OqOmDgaOhDg6je0DktaaNQpofXxQymJZo9RNt1TAcarlVgJBJLfMpVVmGoEUPTRZZzlLbBTEpK1SKsDbpsaesBg/4nJHHO/bVO0ZiI3j37okMRcNbzoGzAzWiaBlf3UQi0+o7GMCiyCBkG9MI1nbPi9psGP5ckm3uORmf84fUjdSbpj15gdltKd4dnB0zGU6qt4uYupZSKrMzxXYFe1cCOtm3QUIjOx5It9SLCynVqJdAdybOjEH33O1J9wfD8I5aLliLV8fsBSWMjjivq7nuOzi4ZhXMuVwXlviNb+YjdGbI6x+8iYnLk6Jh419KuF2hTm+WmJC4PGPR7DD2wumvEvcNInzAuD/HSGq3qE0qI3UvQc7RiyPbRpdY3nJzV5EaNlD2On86JojvWqYc2sIlxKSvQW5+u1Tk5kvQ3Ib3tEcubipcnW9rsGwrtgPEnL3nIWt7+EOP5cH7RQzNWKHPI63+nqHWdqojR6x5jy8aiQWsO2NzOWL3vGIYn3Lz/lrJU6NNDdpWL8mx21R3xNMaQJfbM4uXkU7RfL5ln98zfdRx8YrLZrvH7NtiCychHU0/4y18HNO4Sd3pDrq4p1ooWgeO3SArM1qLKU/rCxCqfkBcHvP4LnWyzRMiSi+Eh4+mKbP1AutPR7Zrp1KcpF1jSAPokUcBiX4KrkF3O8NAEp2WVLHgIBFkb4fdjfKdCcwRZ28PKN3ihoDccoXKD1FPcrBd45hwTjVLtMUKLq0jw5WxATzwQW5BpPf72Pw+h6VE2a5SvcTweIPQLblZH/PqXBYPJEsQK3dBw3ABdxSS7iuDQ417s6aYFOjYkQ6Sz4dOjnOks53I3wDC/xF839LURUhUEmDwfTuipPXq84//4v/hTNvEz/vV3/w96wQON7mI4z9itRhTXQ5qdS9tumXgLesYtwU4QTD0Op3sKCsrIxMhPcNYF41pxIKZEty1Hhs/Q2JNbKdnaRctf0LkV8XKJ7GaMfR+NHcJwKaqALHUIjm2++7bm7uofktsVSj5yPCvonelULZS1h/YU4uGS+5s9rb2BiwU6OX5YYlsWq0sD1Y5ReUbX5WTpltbQsKSB0YGoTWTW4jUtI6NEN33arc6gGfFy8jHauqLdKAbPNRyzwCwEL05/TmN63Kzv+B/+9q9xS516a6FrCkyDfjalKjTsoYs90LB8i2hno+wCb2Cx3KdUgcS0FFaQMjnOUaZGupiw+P0Q3bMwDIee72GLmnykkVYpbW6Spgo/FBye67iGjfH9U47qElkv+NTpc5iu+XATUCUX6FGKdAQoaKwa3VDUrYkKHDJlUOUme6XR6xWMppIrscYe6mh2g9e2jOk4kDb3jU5dN/hdQ6DluK7NtqtoowJTGsj8j5wZ6LJjHFMw6z/SpAO0SqffO0XT9nRqj+NJNMuh1TWKqmFy2ifqMppM8PHTL8i5oTbHdJqOqoek7ZBeIBHLNZ10aHBRmo5m1TRl+2OX14ClD64HaZRz+T4jsE2y9S3jkzFd+4A0NTRZ0SmbMj7j/vUxpqHDXDAIdKqsgyOT0QSUyDF1g6poKWuPrBvTc57Tadf4s0sGY4fEaqmVS7HVcEKJZpWYxgN6viIcmdAzyBdL9FrhhC2Gu6O3+oz1fIw4SVk/bBmPXTyvwDYLHGGimpLVWuG4HxOlY7pyjWuBbkmaRR/Z2DjjY25WlziOjdPtqJMS6bd4vkvnpuiORrRZMjrekbUV6nDERrNQVYdX7xHWDtcL2F+26OWQjbjkaLxh4A1ocJFlSdoVdG6NVtVke41Jv0dPrEC3seSEzbsJThdQWglWt6fdx+hSQ9prtGGM6ZZgO8h+iyrmBPmIJA44PjrhOvk9O2/JQJjssgzT09GURRho5JViF98yCTo8JyTFZxuD1Q/48ENF9DjCGURIVTNfLOi3HoHhMH/ckgOrwiNUHm2WYXYK2jW17dOaY1KVsqtqBsOCtuxwuoCqrSnSJSPLQ+U6+7amrhacvnBIt9dIy0GYWz76eI3GI0nqsStyutlrVHxNf2rT1DuOJyN87Zxv305Jhxcsygh90GLLPvkup1jv6NlzEisjc1wmvkerKfY7nSejp2DmGO4lurenvRlS3+s0pcHLn4wQyiN/OGS3lWzqhp02Iy4XjKYRw9mQwHTYpPdkmPzuux2jsPfjYFBUIooB+fqYPJEYJGS7CCF1DDtgvx9z+/0OlzWhXlNcrzl+NqTuJoSthhXqyOYWfacYeIKBcEnuO/I0Zqf7/PrNPb6bcXwyROY62g7yq4jrh5RgGhBOJzw/fsZ2nfDyZwfE36dk64K68Dg4vGC9+T3BNEcJk9/+yuX/+9/tyExJG1X0+i6qy/EMjalt4wuB3PmsPzh0rUEmdJJ5n8VbjdoUlF3Omx9Kkl2BszXxthJ51OJ0Er3TGVgB390EXK8ky0bD0Sq0veAXX/wJrdjyzYeIKBpRGh2nk5TnYkQXerx5v6frzdHbil7ZI/pDxEEArfye1tlws9zQm2jMDgbcfdejCA9ZmQKleuweTcrFBc8/ztmVW6LCYfVhyvZhSlX16cnfYQznHPcEBQ6blU/TQVzl+PopSpkcHTfIPMOdPvLTn0ScTr9DGCVB9jEvn/+fmHw8xtRzrO6eupX89Q+v+c3yHf/4E43+8Z8zSLf84ySlKWrizsGwBWa64Ng1kKM9sycfeH66pUxjHh9acj1Bt2pEBVXtsU0yGnvD1DnBnrtsF3tqO6OylkwDkyjxaNKAyN1g2QFJsSFKC3TPQYiYqT0liX/LzTKnji4Q9+e4wQmZLxAnlyg9py46WhWSWBYftD9QnW/ArMi7LSXQs01cYTMaP4XcJ95fYewKdKOhrnR0AkRnoEceWb1jNOoYigZLOojdhEn6hCdPjnj37n9EaSUVHTEF474C45aj2QmtXnDkTNC2Lc7KpVlrFLaFphtUlSIqG7RZiwga0nqNNWnY0JG0GoZsGEwFAwcM0ZFmIfNvB4T9HoW3wu85GJpBlVVYhklteGDYuJaBVlSc+jYj0+WajpE35eNRwUl3g9/qnPtDsltBX+jodkTb1ShKLC9kvzERlomelRiVIBwMUXlMLgrKokBaBnXSIYVJXlZMewahMqD18O0G1S0x0NCXPmptoHSdwdj+48KARo88SgjFEGto0VUVUoXs8jmNKDFVzegAsk2M2rRovQc+/jhjFp7wh6//hmEvpi5sRsc/ZSxDigpUlyONFiEyBH2qNkFYe9paoaGR7DRGviKQcPf2CLU/YnRqUa8X5J6OY40IJhKj9lmsZmT7n9CfvWL57T3+4h2iXVHHCvNI4o1Atw1006VLFFIcUJQ2trElVQ8E4wIMA7oxddQjjVNG5zadUYJ+h+Z26J2g0u7pBRmVpuH1AVEgmgzZOezmOwYnQ1wjwhBgSkm8MsnijKD18aUiiW7Z7TImZ6d88/2c/bLH+TjHFN9z5EcYVolhDMnSFFXPSdczfP0CqxigNz2m9oLrzQ1lNcdxD3AGFp57iN4miLYjuss5f9agqxWmWrPf2whvjGv5UG8RWoGjdEbBELPbINqGroF4XzF/aBD9GEOvcHsO0TalqxoMrcAzK2ypU+Ytul6R7TO8xwkj3aXeJ6i14Omwx6hvk2y2HA9+xt39DtN6QyNyXNHhOCHrVCNeKm6vfMrRMdljiRtek5sPKN2lL236jgbiGm30SJNrWOaIsk4QBEwnDXf7in1us5AzsvgbtKBE5dDsx1jWEW75QKks0kZHNR1WbWE1Gu5kRBSN8IaS+fIHLHL8kU2UZeTqDtd8ytnZMW0Zo1XPMeo+1z94fHerY3LNZCLo9jXdwqbZdchtQDs5oShGdI3kYRkh2w3OWFB2U/a5YrfOaO09SbDj8LMtJ8+fIO0B263L6jLCkQJ5eE9wMqS8P0QaJu+vB2hqxGL156iiZr4zedzqDGuDUFpY1ZhC5LhPSqxih9IDusajlT7X7+4QScvJLEN2EX1siocGZb9gkpnMZhbz7Z5yrXjyz/sch0tWbUYcJvz1u5Kus0AfkCUh/VTCBw817xjlDkcnHv/02KXfVdRmTVwv8TyThjF1VOKPHO4fGkJDchsP+ct/G9AZkoGXk3cBUVrh9QWG1VJ3LqYR4OUB1XZG0xQ0gUaxWaK1Lm6jo5k2tu5zan/MNnmLQ8yTI4UnKqQ34/L7Ee/fluy7DN8UXEx0Sgfm8S2e6Dhsp1z+us/oI49ncs2hX3P52FGvQ17M/gw/TPnh3yRkmxb/uUIMYxzXIlppFHVFf5RxOAtpm4rWPqBModimhMcdRiiw0gmBtUZaK8wsJp1HeMcRidcjLiUjV1AlBmlU4mouojI4OHCYaCtm4rc8OU6x2ZBTIYRLUey5e/d3zINjQrnjyWSBb02IdYfvHgd8dSrQ4r9D5a85nmhcJj62M4D6N7w835Jmj/iLljC8Q3UVeauhh4Lj0wZhQltC1QneX0YIv0b0X6PZDRQ1qAGamGAzYceKOF1xNHXQ9IRwsMALT5inLXF9i25usUc7HvOY0ZMei2SLUZh0SYBofJoyQ+SCRWESFzV12XH40SHzux9wnB+toef9PmXiksdLDBljaTVHPRtT9imUgRIeu2sd+WAweCI5HnW0dYTe+uT1iH/05M+4/25BU5QkG8HL7hjbchkGPaxsgZ/m2NEZ3nd7jvYS80HjwJ7iHR9yGSV8qNd0WYvzucE6e0MoLMpUkecCo1P0DYFtdVRtRoXHZntKqnrU1iNaU+GZFmiKOHVZzmMMy0FUPdzaxQs3DFzY7xbUBwMCteTZZyuG7p6ePOXP/41HubTxfKBsqZOKrm/x2Cr8pyGDfkAV7bi5mWMrm32+x7JMzNbAVD6dpnANC9l1SFEzsjSi7R6LHbLcIRsJtxZ252KFOir6IyuMd7t7eoMOyxpRVN+Tr3+JKub0DjoMR1IdL9Ddf4XRP8efnoC4I85uWetvODiLcOqKZfwMv9YwtCvq7Zw80/HpAw6WJ9BdQGmoykBvbHqe/PH5f5Ox/HZK3xtT8z29aY7XCyiqEcUuxFNnvP87k2asmJ4/Yh3+Fsd8wIg+Z19ohIYgzv6WUdhSFi3V3mY4qgiDG6SAqg+ebZFGU+7fDMBU6HWGazXkNBhCoCsDVQxxjSH4P+A5BqZ+ypvXDrv7X1CLEd0uoz/UaP0CvXvB1QrqTY2vrwlGJo32ntFEsol9fv/DN5RrydOXHuPgOyr3P2L1ezTZOat1gisFQvRZPYDeXjIeOkx7x9y++xVpWeF3PvnqAcNNEc4Azwyptgme1WCE11iVQZ0PGIRjkswlmu+Yziza1iFwDvBtg3y7RYQ6niXQlE4WZ/QHfQ4ODlnf3uIPPWr9DkO2+KagbnS0omVouFiTc7JScnpk87v1AmVm/OMXEa55w2+uJ6yXfZJ1ypNnHVlXkxQCKcaouqQ3MDASwFJcfH7Kt69/wMokjvQZTgU3t++pOw2jDbFMg/v1HlIHuxF0coMQPmVaEYkhenFGkse8en6CygS7PEUUNYGAKMrJUoVSErPLMIwCXXig1kh9S6kLtPYn9EyLxLpCb30Co2C167CLQwzhsc5yuqDCFiUjMSWrXRaxhd/VDIOOTNvwq1/X5KaPZUaodocXSITeYDoeUW0QlTZVrTMZ+7inHllT0xUdJ5/fs3ifkVeSpmkw7ZqoLrlPHHxdsNmGtI85vqcjQ8E09xkVir4b0PhjqtUfCIIKr39EHNlkumLqJ3jpgPr7MWefj5gOCtbxI+YYXr2c8erFVzwsDvjm/YrlbYF/dkjgJ6yXFkPbxp8+wxcRw77O+puC+MGkySpCzeKfnZ7z1NP57teXuEcBbb3k0wuH2J1j2veIgY1/nlBJl+v3Ds3OwXBKiqrCMwK6pKKyS0xTsisk+9RhZHSIkzuOgz1HZwHGg0bzJkFmAb4RMHZtXh6OeL+/ZbVMieNrSnLSaMC79yHx5Q6rGzF+AVP9kWimUWmK/UZhMSawLX72yYyxUdEUksfvIBA6F6MD/vD7S6JbaHKNnRYyPrTZ1u/p6glCRAipYYU1pdrQtbB9DGmufIInNr/5w2v8w5S+nVFWJdaZjm6naDsPlIFRD7FLnYFd47Hh4mxEmi7R4j2Ts46BuEblNRqwyXVEPkMJm/n6L1m/PuJ/81//N3iDgji/Q+//FMt6QdrGGHVMU5Q0kU2emATHO/TmBksvadyKIAzRVUNegpAmtu5g6Tt0BYO+h3Z7hlYfkz/eYo8f8IyOi+Mvudz7XF06GFWHaeVskgWeZ3DQ7/C9FuElxHVGZ7V0QUVaN0hpUweShdwSOAe4SuBi4HmC7zc2aRpgej66FxOtEyxV0xg6hm5QNw27lUTUBdKvEFaBj2DqPWGlSy6v9zxzQrzZgv55y7Ztudn6DO0e3kWB3r9m/XaB1WvQIp8/++yfMD0tSeKID6stu8WSLoG+shmkHgf9IzqtZLF5T4KONF2EMpC5gbE/Zuj02HBDoccMXEGnCuK0JnN9Vh96xIsTUJCWrzHslrtFhjGUCNPFsIE64GgypUxTnh9aBGlEkxoEW5vZaZ/9fMXpxwdYAeSVxmAwxNC3hHqFRQ+/H/DntysMZZLucrS8xjEFTZNgBhbJuuRoMMDGJSki8u7HC5ahSkJXsn0dYVUKQ7l0jYaRguFaaG2DYTp/XBjwhgFC5LT7LZ26QZV78rbl0OvTthlWb0cTRFTpltaqqLca/fCcor7DGbZ4ms/+fcETbYTWPTA7XbOaD4g3A3RlMx6PuE8T2uYA048QEqqyQqgMoZsodcDF4RGdKEmUzj5PiFcaZ6aOa79h1Pe5MyIq/XumF98ijyC//JzG8bF7OdvKRqssjLLiYGxSxo8cXOSgeXTNIXE55Ne/tTHkGG/SYHYJTVphCI+ym2LKFFdqXD/c4w0MkrTB0gxef3/K0B8ihh84e9Xi20sqLcJoh9x9U+DZh7TOOdvttxyeKcwBhK5B1iQIe4CjrzHMBXW7omo7yvKItFQMDnyqvEbwjqef3eCE3xLdvUSveuR1BEbHZLRH2hYqc9DclNBNORubXL8xaMsh46DjxRcNsb7GswW+DlnVUJQ5eRVgCpc63yKZ0AuneGaPE+cVfiDZbgp6hkGSN9QGOHaL4SmcUqOoTKJM0k1NvkuuKPKAajdhs7UxZoK029HI73EHGYGjkcz7dJnDUlQIq6QXnLD9jzrL1keYGll8Tn+Y0gJNmnPQmrjbKV0icErBxf2evZK0ooDAoS/3WN2e/JcW7SYk94eomU9f5DhFTZQqhiOfLq/Q9yN8x+Tj4QHlao2jBsSrCEdM8XWJnr5gvUlps5jhQYdILYLVM9Jbi2KqSGqNodcy8yTPg5TPv/op/5f/+z3lQCOzaoobhTkfoA0t3HCAOFSYXUwSxWhNjZ9oNNcWT88/4l98fkzd5fzwfkFXPDB99Q55cMbb/Ih90REMW7bRDmHVVDJHP52z2Vdo0uJoNqWvDIa7a16efcFV1JJFDiemTrw2CLpntPc3HHopyb5jmx0wDhTSuMSQLgNHp5Alb25+g65WXLw4J7ccbpcNnr5keAzmh5Sr/bcceCluPEVrxlREtEGE01eEvQmW9hER39H4V5R5RbK0qUIIei3xekctT1lnHaPTU5yLhma158l4TFYlOJ1GWfqslimNb+L4Y/JIwy4TjP6KSm14eFswSl4iU5sg8Om6NdKC8WwDm4jH+yP6p4pGVMyeaqSlQ33tM/MXhJ5iVWrMs5ap2yf0ppRDnaOjgvLOo972CMh48tn3NM011eqAWTllN6+p1zk3nkKMbRpTp5YK3VfMly1+TwM5Y3EpKa57OKmOZxeY00vGPQPDHLJarHEDgdmc4SRn5Oz56ROLdzQ8zmOOxyG7Yk8TGbAuqL0aXQOBRrA95frXv2CzNpF+zbQfsrtZ4JoNm3WN27n0BwI/XKHZNSLXseqQwX6Ad7CnayoMA/JYp4od3JFACAh8k66paWqNPBcoa0pcHTOP+kzHMXo3xxVDSulzu/MIynPU4j3BmcbJCx/bK2m7FYZl4YYeq/kG1xuRGjZKV+y3d7T6e5ypj9VZOG0PTz4gGoUt+lSVga8cvC0YY0ngPyFu5qSaQ1B/gmV8Qd29JVpcE6kURwrqzMAUU84HOofOhkl1hWcmJGVImjxlsdKY2iXr3ndkn2xIhEY7MLhO/q+ku4SubKhly377BVo6IWw0dF9SSpvNekslW/adYqsJMreiaRRpbtN3A048n8eHGwZuS2OtmC9tchmQFH2kZWM0GzzdI2GLsCy6QtIUNXbr0mUW/aOK/pOUT0OP8HJA9V4wag74+ZeCJjbJrz1uIpMud5icu2Tpkiw3IBfYRsPw0GKV6ez3CSMnRGoNdd7QSUV4EJKVBbs8oq/3CJ0Ao9tSJAlJXOFZPfrCQRQ/CgBFq9NpFY3e0mp/5NEhZ6JjtROq+nfo9gavN+ViPKLWbum0DZpSCCmww5QiiVjczJgehT/+E2oteVQSNmuOT9a07a9w3Rh54lN8I3HqFtOrWW5r0nRIb1ijWQX5WmBoCiVt9mXGiN/CPkNPLvA9mPZ2jMW3OE3CXj/kYSNwjm3McIAIQsp5w8EXJ7T6NUH5GWb1jiY+oOt/xK79Gq/YQ9HQtT3eXfW4zyQHY4sqizmcnbHY70jzml2S8+zJMRezz+iVOdvoirb9QNX2qDlCDjWsoML1NWzPZLHpkIYgyhqs6T33icIPAx4WKYORiTvMSS5HHI5+juvcIZ2WfX5NYIwx7QliUODaAavtHeNBg23o1Lmg3j7B2B1Q6TekTYgfPuE4+Ihs/Rq9/IZK0xmbZ8TOcx66P0eTN8h2SFj71GzIHks6ITF8j/V+yIHQkN0tD3Odjpynz2pmh9+jfJ9VscS2VjQO5AosW2FbLXoHTZ3i9guKVsNUBsPBAC8DzGOuFzmtkdBpWwy3JdoPSB4ClKejqSUDq+B+UxJvzmkNG1NusXyNuErRwxQnWNPkt+C3SPGK/W2OnYFsJd0PEZY7YLQLaL/N8echlWvz+9akTVosv8NNQG1DZqbCfoQqf4YXl/gfPNp1i2hDOtWj8R6wjpfk7SXb+6fo2y8I+h2r2w27X06wrSOiXcw4jlAFXNgan01L+t03eK3F8kOOfzCi70miaoOe6hidy+QjE+MgIM/3eHpMdTui1z/mn/38c1wb/vZ3t7x/Z5NpFob9hOWDw1bsGJ6ZHNUWpuzT8xV/eLdifHKAYz3iD0ZQC/RK0HdnlPWeVFXIMMRsfA4ez+gynbNpwaGt8dgX7Og4murEWYztGehK0hQucQ4dLvW2JJy6KKPjZh8hXZ110pIlA7ZKZ2hYOLWFUVpEqqH2G/J0SVTk7NbXKDFl9GVC117z7v0APXvG7eV7zIMehtERJRrGUcNgDKNZj37c8sR00RyXv17+nkQz0LY5J6Ll1VnEONRxhEFz73NmHINl4KgOjWMmzgSGd9RexOt5n+vfRBye62hGydZvGX3SY+/f02GhnAldltFGI7rKxzUfsbQCzWsx9YaBfINn7oliB2PQYEWHiH2P5X1G+tZmNOwxmjmUTUK+E/iFgT/yufxQEe1cPI5ptneY9QTrQKIFjyRNgbY4oT+JeftDj33X41/+r8fkec58seNsekYoNnjHK5o4Q+QFmg+toWE0Lr72U7Tc5dMvW/xhHz2sMSf/gbTdEC2OKMqccpsiSgfVTtAUEDc4zZae09AVkq6DZGOR7yQff3nGInmPFyQYSrFcwXZv0HpD3l7fk3tbRk9NhBmSWz0edxUqLjCdR45fvGUy01F2H0vT6Twb0Rjk8xWh4xPVLbaacNB0HPoll8sa3/HQ1Ya4qFjuQ0z9gKYaYEYppv3AIIJaH/In/8UL4u0dRT3j2egLAi2g3B3yuFvw//zDfyTPcqIy4+yJg5ABh3aHSASGZnChPeVtMWRXtmRaxmp/S3AuafWK/vCGOsxJDIcjd8RmuUe3XqFsA+dP9tiuRtWmpO9uKTcg6iFVnlBrBnpqM+i7CJVhiAHyNGPXJRw5Jzw1hnxzVxCYOoXzFm2UYZQxTu3jaR6DwZTl3R1Oo7Dbmo/iPX2Z49yHNH89wVnaOLOMs75C69W8W9n86tc+cqox/Uwjngu273XyVke3LaQJ+a4hNKeUmy39aZ9KKpK2QGs6PKkwhcDICqp1RYdNLDI2y46DQcPxzEVbaywf9kjDQu+1MGnJ87/f0MDfGwbaRkOXDWWRMpo+53IT8/6bFY4Z8fS5h65V0HbUlSJLTJ5+/CV6m3N1nbDcpNjOkrjZ8PD6e2RoI6spj+8PCEY2xwcJN9tLAs9j/X6E5/bRmx2uK9AMRdtN6QdDJr1DHuI5egZTcYbGc+6XOiQlj9tzZNxgnxYo55Eo1Wk2kEXfEJzeEfaWmM6cdH6I6v+CXTwj/XqDY1nc1wveXGXoTsvxdMLmVifrGuzwmM36HUrAbPgFd48uWdrQGEM6aZBf6xx1JxSPcxy5Q1M+i13BNj3ieHSBLv7AcPyIYd1xcTqmbWxsNSLNTqlv4fijNc5oRWocslkY5GWPJNkR+gVlUdHkC8bHFlrrIsUxb9+nnFQusjxGWj1mF59gGh1daLC8PqbIJdvtEYPhR4jyHbo25/27kp/9yQFxGlE2QzB0ED4PG5/3Vykfj79Az3bg7XF6jzRmSpt/QrrXKCMDAn7cHC/ABKIYqm5Brgvy6oAsMjgYtly8ytls/oCpm2j5BUq9Y3K+R+0DosbB6SZEqxy37vH4zTGj7BOausN1S+RJxmP2gJARjVFT6QZmb0r7oU91fcFARnS3C8adgxQdm9ct9r7hZGayUhqP/RDMDsfoE9w+EkR9jhubwV1OawxAzOl+n1FPA4rJB7Sh5HD4DL2qaJKOd+9dzKmN1rjcvE+Jk4Du1qEpNnz505SSBreviNQKsXcY2H3Wq4CdeOCjn3fEYcr6wafTWqSAotRw8FCxjrae8C/++U+ZTBWXr9fsXxfsdhnVdMD6UqBXEmvikqxSbm5bDoYevunSLdYUjoHpjUkeJZ4Lrz7y8Zc26UNMVRo4ZxOq7Zh/+g/+9/z+d/+WaagTrSoCt8e40ZBdhmtV5K3Lm7ctv39TkcuWWVjzfFZw3tP4zW2ME5zy3buK5UqjlS5pXTMLP2G+3qKXBSNNYJgzvOkTfC+i+yuHN/+5o9IVBy9LvNGCstIJHA/puOyXD2yLlMdVzcqBm3evcXKNiB2/+LLlxSjk2yRF6DWfvJA8MZewNjgd/RP+pnHQuz6ffTLi+/f3BGaDEr9ll2yQdkvbZVwVPq9/WOIHEh8TbVSwyEJ6qUPbdKgkIS4KnOKGSXgFsUObHZHsFE4oaNuQOu0htZTgi9/y8VcH/Jv/m8vjuzOM5zpleE0UtZwPZ7C20I2QJN3SU6es9ga+pUFVcmAayBpaO0U7UPgHGWpiwE7RVZ/y5oe/ZTp84OOfzqnbFWlR4c7ARCNOIU0lvvURd7djzl+EfPFP96x4w0bbsa326M0EJV3i1Qe8WiLx0fKSdu1QXDa0XsJw0idfR8SxAYsxKjNxvZbDoWQT1dgCNAu0tiDbP1A3Y/yBhvAtGuuAmyxmWT7w5NmeZ8cRhsiRzhHX1xvOAgfXdsiTBL2uQNPJFgqyPfVScGI/YWKP0AYNpnMJwuXd7xVHnxxR6jl9t8DoYlxrzGLlMDR1BgeCxfqR2/V7XL2muA+wtU8xFhN0reNkZvK//ORTZpOQTfSOrnW4mv81cQE9o+Rw8gTZpvRFQpxuOH82wjM/ECUFle6RtTVpoiOnd3THoIL3rJsV7rDC/7yj+foZ+W8F0lJYrcLZOnxy0aPp5lxtS9oDi0qv2GqPjJ5ZvPR7KLXmYb5FNAeY1pBtKjkgoP1lzVA9wUbHvg34l7NPeXhzDWj4MqVylhRGyL/6bwU/ffEMz1c/BkOHiih5wC9MDoyK4olOp0XIuE+QrhnbOwLZoGKLVd8i1wqGIwevLfCFycg1eXhfIewhuWGg9IbJ2KJ/UJJrNVqvhSrGPNDQJg6b9+0fFwYCOSCvH9juMrbFIw/pivGRxfH5IZ2W0tU1VZqyWLncrHUwP3A46GGPpoyCLUGjEf9GIj/5KWXhsizACwUHZ78H2VLmz9CTAeuHgoNnn1En32GFAqvXZ7/8EnvyFco4ZLX7jsPzkvsPC2rZZ94+p9MVg2cjvPoW1B3po43ehZjSxbLn2MMV/uADZZ1iBi6rZUT6XtHVDWJi8pg1ZEnG049i3NEdq0hSeyW2e4fhJFANqZM5i/UbXL9PLQIs+wVar2C/umM2sXCNmKJ+wJY97HrEch7z/DmI7pL+MKfT9/THI3YbRdyEnP9Uobs/0HQF83uHaF9TqJQ6W3N0PCCrtkjJj3YrQ2dzVdJvG3o9nWhpoPKWKrnmfrUnW0rq6udczTNKrSbR/pYL9wAzfkbpaIi2T38cskvu0JwOVECQK1QQIYwlttuhOS66XKGbBdt1jN0bE5c+fq+gURllpYhaE2E5GGXFUAhS45B4r7GrFMJ6QNRQiy1RnuM5AqsbEVUKzztE122yh5c06SHB/nPcoqa1v+e0l5ELid2bQt7HqLwf9+L1hs6uGE86ptOUuGjgUqHlKVMjxp11dE1DUXrg9Ggjh/1b8G6PuDg6IbmcY1SX+Cqhjko894D0g8JzbSYTqOoF/UCQbPcMxIbN6nP+7f+wIV64iLahrStmBx3/4n/lsNwv2TgWXTolexxRLEssV/L8xSF9u8TvBKYz4OXPnlLIBByN9VKj2FrYlsFgaBPFCZaseP7c5MPfKPZ3CQeNy8Q+YJGkNFaDmXZ4gYFqYgahRprX6FXN+MCmJzqkG3PX1JyGFi81hSdyPnkZ0DavifZLHuscY9zHHt3zbKxTNdmPAzpqyNuvDaKs5cuvPmas3vLR0YZWlVBLhHnA3WpOmtvI1qbpDGp1hfsqI/4hJywE6kGwvxthWZdYbYO9tPn9f2dg/h8+wRS/AecNbdKnjlomI49knmK6DSYtVt8mq/Y0Xs6mjrECl6nhkN56DPnxubNtjyk3hwwvWvp/sqEV78m2LmenkpX5a2J3SqcqZDin1i2U7NDcBmU1zM03mLqNaEz0fcxM2Nx8EzOQMHmxp0kPUcUrPnxYcPF8h9ElZGsDgwXDZ3N875Gnf/an3Pz/DLZRiH5vYVkx/mTPLjZZblukMhiOcuoPe9JqR23fcNaTdM6W3oGkaj0ydlTDO/I8pC6OOZZwlQS42i1Z3cDawjxqwGjRa4060kmFg3i1IkmuuFqn3GSvGV6E5GWKCvuotuJY9CnSDFMUqHb9owMl91EDA1XHKNtAFE/Idhre+BGh7Smilqq0wWtRTs3hWIP7HM8KaWRIlTe4E8m2SQgmDT4ZumNQVQFx0dEqnUE4o65v8TILoT/n7r3k8Ejj9Dzmw6PPdnlKh2AmOz55kfP1D0vWDxbGwMIZFBjOIzQ9qhyU6MjyGzTnLYXxiNNvcIsZdXTM7Cf/O5wf/pr9/Tf8z3/+c04mKfv6gYf0WxxxQ6/nEGkt/+jTkMAv6dngU6F5T1CiRu9/zA+XOx7iAscMiCqDntxhNVcY1QrNV1S2i7J6iOM5rtnx5fEnHE+Oef3wBjP4hrxQTMyKroKNVrHQYqJOp9IyZuOSl/YFu6uXhD3BLLAxbzP0x4TJoYWSBXFqcCQ+pgmfELsrLmYxyQ+/ZVkWfHvdo9NMPnm14MTUMcOf8PF/8RzfH/Cf/91f8E1ywz43MKqaFy9t9PgSfZPguTPexi2O5VOv1riTFtVu6IYOom7Jq5SyqzCPNMJDhzi9w9KtHyfwyxZlVpQYRMXf74z/e8OAZUgsI6KVd1TlDXqiM/RGHE2nLHcf2M0fAJeinXK3bNHdDZrouBhL9sk7Vg8rgkOXyfkOrIoaSRpvuN/c43cvUcE5htOhDx2W6YS6KmhWijIZcR/1sb094+Df4z27Rga/4IdfBezcBw5+fkEUJ1SbOYf9Pvsbj/OPfeTAJdLvcIOGZVRi2BaedkjbPiV6vOPp8++JlOD3HwJOjl7w/Owt/vgNu3aHe3CKN7Co6z5+b8zq0WZxe4cnLJYLm3XhoNuK08EdRx/dcnR6SF5mFF1GX7Mw0wWypyGddyTlmrBnUrcdZZmx2idU+xy/Z1LKis1WkTSK3tBHirdU2RKNHgPfZbvL2C9zBmc9mpWDsbMoaXEdjfAkBLkkmcfUyVNae8j4Rc7t/m9wzxqc7ifMdj+j/8RAtys20Q69XbAt1lx+eMv5wSmT0Z6ue8QbB8RRim2M6aqcZJ3x1T8c8d3bHg4hqr6DYouSgrr7MdiYx31ef98wND9DSx6YfjHg3f0HDJnTtBV99RUf/srDNDc0ncAOJNNpiKp3fPwLiea94fA0wULnb38TML8ZIAsbe+Lw5b80mHrf8z59w5PzPfrkjkb02aZDhjcOhn1PKVYI64ii8KC1iH6jM+CY8eAW7epvsfZH2KHN81Gfv7rccFnOMUXAi5MRTm/ObQaT2YiL5zVBdM3f/GWIHRssVE3fhK6OcUKb+SMQ1lhuQmnYHIYjmv9UEfoDJvlzrAZevHzN7rqlq2LiRcl0JAgb68cu/0Wf+/s5+l2LN/GYmhqzxSFG0fHp54eIx4an/hMmnwjmdw9MTqf86u1fYtQ7ZnJCT0yxK0X3sEHoBkPHZ9qPyZYrdt8ayFdD/vYv/nuKdwXbA5f2qCQ80fFmW4TICUyd9c7FfJhyetRyJBZ4eUa1mbOtXVb3TzCaBlVZSNtFpSn9sGZf/QamNm06JbmUGI8lj5ePGNaKg9kjpipprz7i5q9g+tLDmb3Dn3os3jXEH0wOuqdMyxSzXtJjxHUOpy8FopdghYLD3CQxT9CaHbobYw4bkuiKwcsOOXpkmew5/K8inPMZUaLIH3wCo2UgBYutIggtpCqp0paJfoRRw1B49DwDP225n3t0h32yzTHa+UeUmk8ncgzhozlX9J5c4ZgZpmwolIZCx9Zc3NjCGh1iujXb+j31OKcZPOeHrzv+6UcCV7+j5xX4A53hqwnp1uP+jcXBVLGKU4TrYZwl1OaG4sMzJiOX6OE7lPTwLIlnxhh2S/4QIvYfc7eD81e/x2ZP3gWUnaIzchorYVtuCQa31NtjTp8+MjRimrSP7jXIkaRzE/JtROPZNNaI3vk9R19uQCtpa50iC2nlnrb2eft1SFd9wrCeMJn0KaP3bO8f6YoCVZZ445CyqAl9F9KSac8h392gupSnhz/jN7+teXvV8HwQ0IxSjj+30ds9D8s11eQKa2QynHTc/uCx+sah91HGMJQ0W8liK2ndhpvHd8wON/RCMHSdtshBf+Dtu/8RWW/p1TVV+pbb7DvSumPSqzBFR9Od8Psrl9uVoK/FDD6O0ZwV2yTl7eIOQyUE/gCtWUIvxJRfYOQdYv9vGfYTCttjX+lY5KiBxunFUyZBzXr7d7gHS4S1xKkE8yhmMpiSpiWloSj0BsPTyIwQ3e1wRktmkwHtbs/zfya5/X5DjsmgmJHpOVf7Gz7+0zOe/OQXeFYLqxHv3t/z9f/5Fhn00ftrjKnk85/+l7TxJd9c/msMo+ZQhnhNAVWMEwXs31kEykHXO55oHlXi0Y4kdXGPJ3S6VtHaBWlZgWFyeDbB80yoTKxGQKeh2h5141CXGlKaf1wYsAdLhGzYlQWup3N0bhEGgrvVI10XU1Ytphyyno+olYkjOqZ+gbH9nrN2ijEYo515/PL1CqVgt90jvQrLDDCajnT9e372i2Msd8y7DxWrbUIU3+OcLxicNhiipA5+R3MRk1SH2AdfMbSO0Ypj6u0djfaO7a5ClxonTyN+87sl+8uWYLxmeGgyf3sC8ZeEx8fI8FfYg47FI5wefkGaS9Aa7N0AzZyyTzTWu4jJMCTKO7xjgewN0KJDNsuEd8s1vv3ITz5aYRf3KC3mIZEMx2M800J17xBdRFneowsL2+tR1zpNPePtb0o+/fgIXe5IUp/1Q07ldLiahZ0X6GnI6hudehwzHTpYtc3d/YB8H2DqPpXm4ngeN7sYYTjog3vk8Bvi4gmdpQiGe7DXVPXP2Gg9pmPJfPVv2SyuyboEa2xyOp7RtwZ064ZOhtSZgdaWpMkGLA+78dmtOspoyt6McPoNlmOBZqFKDb38mO//pyNk74hKjziYbLB8Hcv3EKqHa8Zo7QqtCRkOvmL3/fesM4fmeUzgpyzyP8ezHolvPQzvC+TGYehVOP//9u5j15r0StDzG95vv/fx5vfpk2Qmi0VWVatK1d2CpFYLgiYaSCNdla5BgAaaSJCA6iqWaTLpMpnut8ef7U14H19owBvoAYEGxPNcQQxf4FuxliHTyh2yqKKNQ54dzUhucmZfD1nN+oT3FrZloyQmkjYkKiSs8ycYWQ8FicmzIc+sf0HN5vzT/6Hz139hM9Bm/O5SYbVQMfY8WjdFH8eYgYkiychVwfZuRLxqCW5jnjwZoNdjppuA/jOHm3cR3qFLurEYD55x9WWFdikzeGzhv9sAI3TNoU0qmsWUcafCkEImomEyfo7bNaC6R0sCwvmQCzmkzjwGsUdne8pmN+fDH51wcgKP9gZ88Y83GJHDqVkT35YQSZhyn751znufWqS9DMmVOexfcXTSpV4FbL5sGAQ68k3B1tEx//UhoxMZtQixzD1Wv+5yko1xtmuO5w2GZMHgKd++VClvDlhe+qi2/IdFTezQxAa9LUhiC6Mc0lYTZD1n7yTFW89IMhu5gY6UkuwUkp2BMalozTnG3imFrxNuVM56pzyzB8QXCvnMxjz0Gezd0Eg1HfWUu3kf46jCGQ3IlILYuGFFjSan9A4K0nTGuowx/Kc0b/sYmcTzAxUtyJBKCyGnpDS4ko43ivDUazy9pY4sHMWjUVoyy2Tb3rCr5zjnEsP3C4RYoZghqpDQbZPlUkccKAw+S9BPLnCf+wSBCnWPTCvRagNPtxD2Bd0PU/LNAskYUEoFRr9kYsoo+RxdzSiShkjtUikZotXp9JboZsE2KtG6CZrRIhowgfW1w0re8umfLZGVjF3gc3Dq0EopltMl9Q1G8hbVveWzH6okb32md59geSbDA4FmFpRFy3bdEuevODg4YLcyyeoaUKiiLutyyN1Vj9ffHyHJj5DyjM/7IVnmU5kj9uwnKKVPsw4x+yoGJWqTs0yXxOWCketRNwaKs09FgFQHdKya5ug7annD0SOBaeRcRj2Ec4Lt9RH+kD1nH72YsdvGhG9PGfxAoOsyijSkyCLSVqOdH5PtQGoueOINeSU1fPv2t7x3uKQWczZFha7uc3tdkmQpg7GBLYXUVkDptsiyRBFJCCEj7JR8GSGXGyR5y2p5w1C3KCUD2x4Qhjv8xEeoA2b3U6KxjpSlvDicsN35IGtUVcqw10EqIixLJ892NKGEIj3F7bccHjZQ/BylB4nr4X3sYJkynaYiby36H7h8+FcGdfUbCimH3gWj44j+qMcHTz7Btnr8zX/3EcZQIfMNKut9XpUyap3xP/zFnLe//4KV36fRZHr1AX4U0loWcZDiui5ed4+RIZPmCb0A8lWK5tl0hzL1bovwBeQVRqNStgZVIZMWOaNO548bA62xIpcc0uIzTPstRXNLWyYYeoapV7TdAy5edvHXXQauSWeUY0kblHtQCgcfm23WEhenZInDkZfhWSHd3pDhZExwuGC5vULf7FMFGqf7Bygfj/ju9juyXcbYTdjGAlqHvCrYsKIpDJ73JYZ9hyq3idcNklqTtzva7pAaMLrfs/805+vfnOHZz2lEynDk4Hb/knYt02ghWRnTLHuEscPhJ2NKEYC8xjBS2lpBpA7q6IDCs9iUr3EsePrsU/zwFkPfJw5M8s09rVeyqWaYByEoW/KiZDI8Ig4tSM/4+qImD0rS+CU9q6KcntLJJjR+Sd+T6EgbTOWIRNPZXe3Y3TxGtAZp6KFULk2bQhwjoi2GnVOoLbUhaNWcVPmWqomx9RBVqUiDEld3KBON3bsExd0gSpPl6wGNonHy8Q36pCSYgyLHyLagiUxINQ6He8xnMobyiCJ7harVSFoPz3xKXli8/c0x+Xaf/qCHM44w+jGX93dobkbdVmhiw3jURY+OqG8dVi9dsspFOjyj41mYmzX6TGW3tameWUTVDsmVSUWGKtdM+hbh4hLHSJHkPm9+Z2NKQ/bPJdz2msavKPVHWM8swmHMgd3hveMpj463+L/ao5NZdGwXWw7wp1O8eoiklsS9GTdpTZkIRBtg2l2MokS3ZUYHMhdxy/Vmy6k7wnDh8XlNLoXU9RCUiKbucn+n8NHnI5bp17SVoF6lOLaJqwtaT6WcOhwMjkhvW3LTJlnVGJqNmmy4frNGOXuPoxcq1f01VVFQty1mL0DSJdLtHCOt6V71qRKXfl7SeWQQbmpkv8viq5Znf+FwOXuH3ZWxOzPU0gDh0WxVLCVE21jIdyMGoqQxKtL5Ee31Ht1GYM3BNEqGHwi+um75p39oUYwcv8gpBgJrFNCxI8Z00OuCjtTn/OknfLltias5L37isnl3wM0vBV0jI2tiiqaDHhvojUddFaiiw1V8zptLCc81efZIJ8t3eDpooYNROPS7LaHvUOUqhWKRKRqDcZdk5mKlOsQpy2UKusbhQKWTGpT2KUZ2jkqLpk+JamBXQC9kl14iCZ2iydF7xwQzkyyX6Esy6uMV4eQNV1GLlmr8oB2gaxqq6eJaT5guFNa3+8SWjvQXr2k779hKPsMnP+L1pUB39tmX9zHkl0jtEq2J0GuNOtQpgwHeXoXulMQzi6JqULBody5h7w7nfMNgMqVsW/K4pdfTqUKZ7TJFzVrKdcNw1CffPiEXC8BmtwLb2iI1NY7dIEoZ1dyQ5jKx1HA1nRClY3Rrg60dUo9HJHnN+5/UVJ015DVyC5NOjb8q2bafc3Gpsm4mgExPLLCyNWZhsLw95fSDE376s8+JwhvGezlCbFjsXH776v+mabe0ks3LV1tE8wgTn9NJztFexHSxpJEyNE1h6wtKeYKsOQx/+AZ/+z2FLuHaKa0xptiafP78jJOzDts0YDWtudvqeP4xXU1HEwqWtMRTa3bLlOXSR9MTBsMxl9cyS99B8ca4cs6j8ww/0ajCAWFacWI/oxD3dI2auB3jyY8pghZnIKgigTA94miAqbnExY6s7hBHfX4XFOzLfT6bDEnKmrCcYVgm82WILOsYUoNc5ajujr2hjus42GqGoX+IrQrUVqaSU8LIRxEHfPBvPqVtp/zLz38B7pLOPjhyQJ5CxxvTdTTsgwBj9BuW2S29/j63U4sb+ZRESRk5QzZqSxz06WgHhGFJVdUUwkRpBEZicaRInA5NDOsxsyQgy96gGTaH1TmEIXJlYioxtecTLu6Qq2NsoTAYt3/cGKjTlqSuUL0JQfwN+0OTOK3JGwU/s4jiCbPNiMb2qIuQeLOlaBzu3o2opD5xYZC3M46HLscTmx9/ImFYPW6vajb33+B1XJJ6ij58zY/e63FxmzDb5hwf2JTlik43IE1jzKbLKvBQTzw61ppG/J+sdzV2c0IbdymakN39AfnU49nTiqZfsgp7rG+eYY4MpEYjLhSi5ZJw1UdxF5z1VXrHKrvVH44H7bYeci5YlRo0EqLuULeH3Ac3ZJlMWzZcvP2eqV2jSzVWP8Pt1XQ0lbLY4BkZ+dako+5hS4d8/1uJ3dQDxebIjDmoYfvKZLt5yi6X+PRccDwOCYKCymjwc8Fw/wVBdYgsfo+pb9nIP2Ng97h++3eMj1dYY4uL7xsU1aMWPt3RgKEYIpcNHjGihOHEQ25XdIYWZXuM2/YpCwund8lh55qsySl2Dron01oKVXSGZo1oNEG1KfAOnrGLrigbHV2tqGqd+5sRqnnEk086oKzp9TQUI2PpJ/Qli+V2R88ZYtpnbIscz6iQlD5pmqJmMadegDeuyUbPuPx/SiTfQ4kylFpll9a0zg7duObgKCVMWwrHRBzCtoKPPwRve0UjbN5KLpPnNvveiqTU8OwZbZsjtT9ic7+Hmr1l8+Uas2dS7rZMjk7Ztyf4r7dIex0k7miLhm3Q4j6v0NUdQ8fg8uc1TmeMdixjqTVVKIEj0yoO09uaZZDw+X9xgnNfMRNvEcqCzK8YHTiUSsssNqjVA7ZhgtkvkY0pVRWwu92yTEY025j+HkwchTjYUkUquilRS1OsTsN0tUMsu4hVhCcbCKng+H0JJYuYfmHQ6gGnHwiy7T6a5mNqEf1Tk/iNTdZIKFJGp5Uovk8IxiGG1ePk8xixumH7OxnD3UPdj1jcSdSMkHCQ5BI3limCkMPuMWPJQdr1KToZQfkOpzMhw+f7VYNsDbClArey0SQNoYT0rS2epJG1LjdLk5sMdv0cTInOYw/3wCJoE/KiJG91siCk3YXYzQCjsEkCG9dSaOZ9WOR0z3MOuiqRJNPTLMZ6y6q9pzzfo+zp2IpFEd5wOHDRlJjf31fcvJLpHHeoKVC7HaoXc3Yk7EuC+2VJmOucdiWGowl1miDpKnfrfb65KOhKFr3Bgs6pTbBV2cY9KucIqXSRcwuVmsneGsXYIrKG/SOT21cJty8zxsaH3H19gdvvUzQR/rrhyXiIpXyDdhQSKx47vyatoKOeUadr2koiKFqUgwVPn+9hqw0iE0iSSlKViKqLoph07ArdlaCeo1uPSO0K6flrhu5blFIlybqsdhV5WKP2daTekiaAVtJR6pxOW3K1M9jVGjEpnqsitylFmWEmsN+eoCWPkGubcb9HmX+HYWvIrcOTzk+4Xc0J1zKO8oKodJmMGz48LRDxBZbaUguHbRzTSBrbJKJWG3rdKYcnKiLPyRoL49Ep+a+7hHnANvyGSpbJF+ec9UZU+hpLdpj7ARUJwhFoRotLB9vy0fIcpBNeXzrU9YQfPD0kuf8FZeNS0cWxVTSh0ZXnmI3JVAyY3uV0tFtcbwNVhmakFIbDTViTNTqmrXN0OuTtqqYKxiSLjCo28XMdQxtTljJP9g6437yhjFv0TocgTUjXAuGoOKrDoxd76NKMm7s10tWIrvsZJ2enWNYvWP++REoknvae4m8iUv2O0Q/f0X3yIc7je5b5DZUYMV26xKlKWC4xKCjnHYr7Z7TmmNj0oL+gFAJXcvmp0+Hjp33+7b+xsbR70LrE/+pTvvjyz/j5v7zikXNCXl+yXM2YLjPsWqNtFJRSRjNbgtn6jxsDYVAQSNfY8gLXylmELavQZXSwj9HKvP62JNzZKGMJU1dRZl1M5XMMw8OXG7TsLR+ZK8wqIVyBhUy4eImnushqxrC3T53luLZP2eQktY0sCzodH1lJ2OYZrXC4v+vSBB/gDFS8wTviBcyvuow7J9xel6hCQ1fXKNIfTji+88dImc3hk2Mss6aSZa4uIqxeytOPpgjhk4Q73EmA0pX49jsDsT6h1TrcrhOePjtnHZR8faVwt1YoxSMOj1wk2We9u6VOUrp5RefIpchbLGyctsv37zQenXbYXum0y0ccTj5AMgWO+A16J2V6F9O0dxh2Q9aBWWUT64+4f2dwMzvi08+ec3IUcKLec79q+MeXJX/1FyNO9YCqKghrODgw2cQb9EolfXOIrR2Rpz4TpyWIDMLP/wHH/hcMUzBS/4bILGnyX/P0/SVCVMiyhrtf0EiCpn6f3fZ9ZncN3WGK3JfZNQsme5+QZT6qc01Wx2yFC70pe6Pv6fcy8qpBbxQ63oim2VHVOmr5A969fcxqOiMavmL8ZzBqAz560dKKnGXYpcxP2HYD9J6AgU+eppDGjJ1zyrDB17rIUo/tXKZ2XT755AdI6pzr3x3BzqF974z+hyNSP6NYTNh95fBK+hLr7gbplUG+3nJr5hwPCjp9kzQpcEMNJXFZvss4+dQlTr9F1izUvZqBMke97qHpE5ZRxPFjFdVuYdqlRmPw0ZBF7LP/Q0GU/4LdYkcoZFa7BNbw4U8FZ5/bvJ3PUR5rHB4ssbomS7EhXgRkksIqqbC0AjOBpLTwdxJxXPDu21uO3i958/uat982jG2VRi6xOn1SobDXCZgcpKx+1cP/rc17T0E0MdE7h4yAx//K5OWFjfHmD2/gaZCRrF3M+gXRqMB7dEfoe7TihO1qyOrtiquLmraUSJIQtApzrOMFI9StRZgIolWH4Y8caktHlcDYs/hGDuiPG/ofSZTf2Zh9n8MXIc4wp4h0psFjvvmFx7RJ0a2Ws1Od2rokdRQsySD8VYeJ9xhHXHCdFFijhq7xHhezl7TKgjp2GVkOji2ojBFRPsSfbjlQD6gVg6V3x+C9Puv5ht16jf/uDs2e0bQFHd2irQoW6wxZn1MPI9SBwyLT2OYV8U7FzCHwVbrmhywWa26mKU63hxQu6FsdlvcmnvYjpsuSWh2wd6hR+XPuV9dYVsaToxPiYolu18hNy+Zrm6sriaQ44fGPddBDKq0kyQ067RNU8VtQSxzVwRmYGIZDI99QY7JZDVHdPSbHj1CUHZ49Yv6qR1nklIbDKohoZwJVkvC8HrrVZ5amKGcvKY0phjwk2E4wJhpdX2B1TynaAzRVQlErNnFDEHTYzG1MqaKpCpI4YiKNWGwV3DKj2O64zwpevL9Al18TFpckO4dvvrE5sUATDUFmQitTSDv2hyVVENEoJlUjUYsYXZFoIh3L8ghL2CU9stikjEwc8wlx4WA8S9C8N1TtAsXo4z35AkUvEI1Dlf414RKaaE2pj5A0k938hn43hWwI+cdIrYHeMXh8ckI+/Q9YvQMO7Pd5cmQz377hfr1iV2gMjj+EzQzbikiLBqHKtMToig+xSaejUkkhebaiMzjF1TsYbUxxd4fQTU4+/JCylakIEO6cZh3Qad5j3HuCUHYoUsRmozN/V/Hh+WPy+w2WplNJb7BUiVa5RtZ95NohSXK2eYau5bTejC/e/JInbkFauVjWEbZzQhZUxCuHUV9j0JvRsxMWnR75SEWPZf72xb/nk+f7dM0rqloir5aISqYoQlr9JT/98SFPTj5kTz7ElA/4p18b/O//2wVVI2EYgtaVycoYS/sjryOu6wgpEUSpQzM0OHBKNCSktiFYDwinBkbtUWxntE6L4Z7y1d+XFCwJPQk38hGuT6q2WJ/9CKwd1ea3SCKha4+4XzV8+/sxn/zw39DqNUV+Q5DMKAtBng1QLAl7YPHFy5qubiAXbzluU/LdPst0iK9t0M8FIwVOTwOGZsQm65Cs38dyMoR1wa644e1XFY1oeKpUqPWK1krRuyVZ1YX6nKLySLQQzznk/kbwy2+/pXvs4fYcgo3gxdEjjg8H5Ot75GpErX7P00OLft9jMy0YTbp0OwZ25zWaA/46ofYkMtckWP6Ojz/2uS0UjCcq69UvUBtB5vQpwlPWmy6pZHP66Jxt4OFpDcejDot5hVKfc7fYQqUxavYJbk0kaYdOg+e9T+h16R7q7LtjgtdvYOmgRhvaYI01HOP15mzuN8jJHF2RUK2WOmtwupC1KnevDkjsA6I24aR3hnDW/P4qgU5Bre5wJAUpGWCqf82ymHLkhCh2g1G37HYqu1ClqFVaxeUmrkhmC6Q2Jyk1tN6WrrMkwMUJP+bb3yu0wsHSC9zBDvmDa9pG5/r3JsJQ+fZrmZ/91U/ZxQlSvuTshcXJicblP5cUyqdc+Bt++uQxdFp21RNKQyGtdV6+6iG9bXjaLqBuyWUHbU/FeWxze6kRFDMOBhLRKkGqcwxLIc4jYj+jjI+Yr0wSu2Wu3fOB+4zXby6JNg774wGbYk33cAPxLXLcYxt0uFs0lIFGmzukGbhWzZPHOtJui/9yyi7vk/c1qpUHgYNqGbiHOv48Jqkg1WQiO+arm5i73OXl32dUoYVfxqitTCRKDEySpELqR0xOWu5mMmkI3lHN9DsfrTSou2vUR4JuBPGm4uay5vhve6jSiq0/xNm/YJGkDEZP8Hotv7psmE9B7JYYQkP1OhhOi95NqXcZyXpI3vbopCHSpiS6S9EfOVzehESjLkPjhEh5gzgMwbum25N5dbfHb784IN6dobUVA5Hz8eiUOM9IawXbTph0NVZva8aPTtBrDWGUWPqYZtMnLLc0xobYh7zo89vriEo8Rt4959w1yRKffjxg+8qnsBL0XsG22nJyYDK0E8JcY73VKG2dskqRdYtNlpM1Gbqj0Zt00aWEqr1is5DISwNFaqmjBUVa8M2XAndPo/W6RJnD4jdvGYxSHp9Z3K0bTOcMWWlQ3JTaaeieTfj+2sGPC7qnPepaRtPHhMUU2TglKyZ05LdkYoVlj+nvHZOkPbbrCJRHmMMhiIg0D+k5a+qmIFMUAtnialaxCgWN9ZT7pcORa/Lvf1KB+BohVhiORVs9Jd6ZPHqe0yi3LNZjstqgyip06xFy9pSXL0fMMouOXPOR5eH7M4byAbupzLbUUIXBaLxFVr4ml16BA0J0eXb6YxxL490ri7yume3eUU5MDvSANIPu6Bl12BAGr2kcFUs+I24Mtv6OgVegqznC65O0CoWjMfng12TGBabWo0pNZLkkL2PkFraLFF1eYz1KKB0VzajwOjsaTWa5lUAzMYdd4kKmySsgxhEKR4MO8+hbimKKikwRdklSBVFl6G6fHIdcz0jLED1pKGMH17Wp8hg5L7GTlKaaUe5N0Y/fcn7yjLBYMV+BY62wmw1Peh1uLzOEOkWVrxiYH/BfPvufeHF2wNT/R05Pf0cdFXT34Tb4mj0tRj69pC7H/PpSxnIVTEJ6Tw2kumSRtIiiR1O1ZHLKoqoIpQand4Ai/Ybj9+bM6XA/M+nnE+5mDrbzeybO1+jaI7IG1kWNMCO63YSyuiAt+8ybT+mWEmg6VXrI5599zGSQkRsqpdJSZn/kpUPFso+YD/GOUrzJhr6dUoYh0Y2Bnjzix58+wo9v2F2uye8VtAOZtvSxXJd0WRDGMivPwZQrRvIdrZyD0WO3qujoZ9y/6TC9dxD9ilpZEwYSfe8DRCz43VcbajPDPnR4l24Zu2/JNzuU7gsc5zGp5zOv73j/xEZrarTQJF3JLHdD7t4OGJ/klOO31F6G032C3kzoaIJg/g9YQwWrP+LqTuFmecw33ydkosIdZKy2gixqUBc+cRDw7PgFYtuyiKaU6xQrnTDa/xGb72JWzj1HT4Y0Zp/fz0KizOXNyxzFUFlUAW3wLQddgaaNKXcKtmQx6gru7i+II0G5jDDKQ5K6ZBZd4XkVWhOzjR1Ee8CR08WwQs6eu+yu71HFEzocYRgn/Ooi4N//9R+GG6ueSjnRGFZPePVzleypQ/eHdzTyV9QFGKZEiwECNF0gAUnQYRGOkYcG2mjANi4I/ILyRqd99Icb9Vr2MY7650zDiOOjIZo4JvdVbi7m3Nzesy03WHqPcW9A0saItmHv3MbFpVEdvr3XmcXvoV97jLTvsJuEqDpDVc+JE4Wec8Df/Nc/RDJqVoufs7jf8up6zvuP9/nB2Sm3t1+ySWpCR6L5qKJ2fP7h65L+8Quq4jX9rk0WerRhhThqUR2NrbciHR2RzU02usphR8ePt5Ral7aoCEWOaveIr21evTojUXtU/RLztCKcvEaEOY074Lv7CEP49B6/Q4glyUriLvCIcpmOphAvSnbXDWa74yePlvz2l+dE/3JIT3OgW9MkMqmUgdcwD1fERkKpZgg7JykC5plM/92EJO1jNFAVBZol4Z2CyIs/LCTxNKI6R9Y7xHmHppZoewWLbUI1P2R+kdNPZNoallvBMgk4OgroWufoqsQulnHCHfffpESOSt9OqaoMhIlctohSQz2EstIodwad7or/5kclFxch20pjUzRY1ZBgVXGxqjnQAU1QCoVFqvP1zOEm1ajkDLVR2PMm7C4bBr0XWO4b5P736C9MXGx0raLa2EjZmI0fE4Y2yeaIKi4pTYcoKLnJl9TeBc4456VyyFYWJK9lkmHC0WctWWRhKGOSrxvixibzVHwrpm73Mbs+dq+krRs0v0YOW44eT3g8jhDNnFbLiFKFUhyhMWG6XnI8GnA0HnIR1ISRT9yExOstxuiAyeMDer2caTun1ROySrCtTOaSTKr5JMWC6WKfPKhIQ8GHP3S5Kt/iiQLF6lKGDqGls3uXkfiHrIIOvZ7H4WhG3TZczEyCUAGtpjQdWtVlGrylkTKivsROafDjhGSdIVd9VKWHEhkMZYEUpVzP4B9+O0e4LYY+RrKeML+ukXIXywxwtYgfPHnKL/8+o213VEGNJHWR1BRLhjjwCXybTvcZhd+SJBUXm5KbN2Pi/j3bYoUR9whMiV3dR9pJxHGJ5X7C1aste4cn5E2JahgYUkVV5iSlxS5aYXBHd5Ry71tYvVOSbUqaeUgdga0IWvt3mPoOoRuU8ohKcrlICtZbA8kaoNlblMam8S3W/i1FOcBf6WyTtxw92VBrWxTXYDmHWbhgfNaie6dcTUNKdY1qQWWotEOZvCko4xRb6aKrLVml8Po2xDJMZvcBWitzPnpEK7Zkos9ylfNMchGXBp2jv+F//Xf/M7JisvFT3v36EtWQUaQ9WtcAVyXlCZt8S01IRQyxzvjsBLIxnvMxw16fss7Qa4+LK5O2uqXTbblZZ3ylT0j8LrdpimJ4zCqTevYrfvCDEtXxuY2+ZLNbokkKbazBfYKiaRx0/5L9/hHmJuPp0KU3fEsrCxb+ltl2i4FKXlZ/3BgwJR23Y2H1JBwlw5QPmC1r7q5zRr0Od80tP/w0wLGOWMsWp88PMDsleShQopTSTSj3JVoZ0vSaMqkQksJ8I/N3X9wzXSWIRqeYBuw/61JXCqnd4de/+447f4Ost3haTsCKnX+DrKgE65hk/gWWZePpKmX5jNe/B3/9Aks3UL0eyClCy2kNjaTMcU4K2vk9wapPf3wCrcuvvqj4bhbhmwG3xR2OKzDkBiXRearvMTQcloFPm0Y4pUnZdpnOY8b7OUZvDxmJwWMNp7sjTEI61n9FpqsI+x1BdUFa7HPoWnScPaJcQVgJq2VKFJ7R1MdsNzptElOpgkxIRJogqt/wsb3H7fyQrJ0gp99jJ0vCW0G0kDg9fYohbK5ex/TkPiKWScQIZVchG10S/ZIDrcPXPx/y0+dH2EaOOXnLWhXISocis1CUliKtWd1qREkPYfW5vQVtX0XumFh7ObmZkwV9enxCmR1gqyHFXcl263EX70jmOlLzmK7TxzArsnKG5o2QWw1LcihNg9tVysvNGWPnkA/Hr/nJySU3O4dvvpK4fbNPtpGgvELqWPz4vR/TGj/m4sv/CynTaWQVzUiZLtZoBydswpeYw4Zfrb6hiY5w+jv2zjUu/mHFwDjnrpnxJsswXQXFHRINjlk/XzI+GeJ5JtffF2yDkGetjNJYxDuXZfiIr6+h6SocH/Q4O+lyV/wz1kQmjQrCTc6AOZaVsLw2SBcDdo0AHbIoRygqhiZ48qSPLF2iOoLKVYiNjI4+pLvvsNbWvEp85tEOujWls0FUgjyQ0G917jdbPMtBQkItDYRcoA/g7GmOIkVoKsSNSutZfPn6CrPZoRgZ3fcs3nwhkeIxUHJMsyKpYLepOXoukKWvEJVKmj9CSH3UnoSjFjiaSnesEu9SrD2ZzA3pnByzm9lUcsF//xOBl7/k5q6LKQ7pEXKgB7xZL7iewk/+VmXhy/zqbh9bFcz8kti8pdXXmJ5LaG1YVSbhO5cPX1gYdUZhrGl1mV0t6L9vUM7fcR8syJKa+J3OelqT9FO2+py8k1PLIZv2HU11gmkf0WtMPGON5Gv02y5uJ8fqv8/v/nkOzZauXREHGYgCWbHYs/epNynNpiXduOz9ty8otr8gKe/ZJR6xbFOKFNMY8NHzj5Flg7/7f6fUeQJGQ9vqTN9tGI8qkAqkvokkqRRGw9fLW66ThDxRGKoW8S5Dsx0M12B+/454vcRSNayujCzr2FFDdt+iVhayWTDaLzCI2ew0ltGIMnMpMhN3MObdt69oWzCaCjla4wP3SYHuZJR5xUT7nDpwibfXRCmssyEbQ2G7ydGNPVpHRu71keqctJhhORKycs35YUKV1sRLA12KGJ4WCL0gRcXwJoRBDpLLbnvJIoCV1RLFKmkrY+cKL0yFpEwRiULHG7COBHG4R29YkwuNRj6kP/y3ZOUKy4kJlq9opIrFbMjTj37Kep1Q+EsulxGPno2J/BCEoJJUhKoxPjJo1TMuky6VUCmDACNbkuwU9p3H3L5esYyOePTRAe+u7pD7Nnsdm9mdxOLOIdnoyGUHQc3tOkNWFNxJnzgsyMstalNBVaO1fXIlJ2lXTDcVh3aPJu+RRSPGyQjNLeh1/4xvp18jLI2B2eWjkz9HVxLe3P4dW1/Cm4yopH2K5DWbsE/dfMB1MGXUPaOuv2edBZjViPH2EceDD9j5Bkni4yATxjm1OOLxicShJGGqQ65XAZn6iHmRE+8yrMGI47MeuvqaKId4VyIrNkgW62XD4lbm/MMDRGDxyAG/bCiDOT/8C8EsuUGTICxi3Aoy+Y88QJiX9yhGjid3ID3gbXCAZJ+zf3pDHd3zwXsVZ5OMf1zZpJ0D/v7bFeEqhniDI+v81Z99hG9cs0y3DLU+L98suZ5JXL3sEGSHxK1ALSFbavzuuyUlFZWxQLVq+oqBolWMKjiVDtgGAcIWFOywOy26VGBbLik3OMf7hLXFKjLJNlf84OOU0blgofbYLXPcoY9GRmfPJS5c3r7t8R9+uSJUVOROiSFpSJuQPcWmqzrEywRHNnhk7lGLGMtp2czu6E5SxucuqV8g5JouA7IYgijh6tsv0cSAtF2RKz2G6inaxmfXdri9yzjcO+bu8hX3iyX/6qdPaBudQC5pTUHk+1RySc91eXs/p6OD6f0jH53ckez2eP3FU8Jdxt19QFjOcCub/cEel9/dMf54DKlDnVfs1r/B4wNa2+PiXc3jn3UJgwpdsWkrjazok+dD/LTLctZjdashkik7X5Dr7yHFJyTiiqYSyMYL+uNTMrFlNKyI4hXr8J49aUXTXeJYH5Nm71ESc5+8ZBPk9LuH3Gxq1llEYcrgG/SOoTT2+Gr2MzYrjyZeow4T9ME9XUPH7r/mejclLgx6fcEuVpDVhOni96xCndSK8CuHdtOSrDKO7RYtXCGGEfZhSrFW6HzkMfqpiqSkGIMj/mMQEsgztEFArh/wpi1w+xlBWhJvDtgkE+K1y8lY5eCgz2o7Jf66wurtU/kpmzqEtkTT+2ymPeqtwXYpoxsJhtCh7tE0AsNJKPIWQ3sfaaLhP0nIZIXpMuDEHJLpNk3mo2wD9BSG+zaNbJDFFcpMQpM85FMwDJX6dQWRzM1XCT/+WYnurFEVB+NFSLpI2SQhyaLH+LDHdLoj7WtYf2mg3Z2g316Qvc3I0h6y66CLbxH0SPUTAsPDrL4jzQNcbZ/hB31eXV/QjDJaq2VxH0Kl4LgJbVMwn+1RrMY0RcV7P9mnyivapUmZVnzxRYQw98jzlFaEeKnK00rgFxGqKZAVjahMWW7B/1XFX3qndE7vEWQ0rUqdgkJKJs8IQ4XKkmhEgVS1jJ2KuBZkOxW5LXHUgs5T0O9f44iSsTIibyqU9hPCdZ/9sU877CBLOZVaU+Yu5s2Isf0ZF29fYVUy6kZBuj0F45osSxFlB6mzoTtqGHQ+BaHy9S/uiKcRpizoaybdxmHPdrCsit39hm1loPbGiGGIltSc93Qk08UwOpRVSLhaIGs66fUCr2cRK4dwojDuenj0yesVSmlh6Dr+coW951ALnSD26Wl7+IlJeFmhvuriDWTcVubp8Jgky8je5JT9AyxNJXi1YGSrtHqPVZAT1wK5VHALm7JUsFqFJNzhOBJSo2EUHle/y4kuLfxC4DkySrNjdGSwzky0+q/x2keYw5C79d+TrWukQOasbxCFj9DsAW3dZ9+/pHMkI2USmlARjYdq21SkjEqdUhL4iwzbPaNabln+05r9J31OPn7O7esNwfYd0nWf0fMeIr1CNrM/XMOtuqiKQ5YOuHsnsawsjs4HiFylrHNs1szeFdiKTFC1BJuALDN4/Y3E5P0fsfpqjcMhcrak+j5nWe3oOiZlZsHGRokKbENGUwU5GXVV0+/2SIIWSdtHFhJKmGFLBdttRH1v8fyzPoPTQ5J1ytlwwMz/nl98eY9bX7BdB5j9HkPvBbt2x2ozoxK3ZEqMszckXBzj7X7CvnPGT04/YxkHvFnsWExnPD3Y5269Qbd0vnv1S4yeS17KGFJFk7dY7gf8j49+xkefH9BRVhipRpAf45kJby6/QbVbsjBF123WF1veZV9x8fOS27sUxX3FZz/9hA/sxxy4J4TBjtn1jNtXf+SrhVZ7RLGNubloiFIJo9/woz9fMR7MqQ+3yOKSXTLm6+8kFkmFGEaY5ymemuH0engfHTHYnbH58tdcrneIdcsucBGphSc5fHB+iCtJ+Ks72lzDwqZYp4wkk17XQzdNtDfQ01xyMycO1qRdCfY00rwkUFOU/ZqsLBk/q0j9FeenV0wmc7blmHS1jyUUXFLcgzVhKLi/HiKUMY/MPlJiM1h0cTDRSpkz+Yy7xT2Hp/uYVktQC1ZpSj3PGLuPyYsAdyVR3pSUsYOqPCUQPnW05hNJYmzlfL+xUX9whiqrLO8SDq1HBIWguttxHHo87k842IIqVIrqKdE8QotMqATS1MDwQvojUB2Jjj5Ey/do5x3MtkWfgiyZNLXC9/dTDHnFv3uSoUgCPVMw8/fIUwunsnj7/R3/Ws4wtQZJl8nWGq38lHB5yPLWBF/lxB+yWbc8qgrGI7i66nAzP6DTOcIY6uhFwGS4pJUqpKaheKtQVR4jS8W0W3brGzy/z3vGp7RWByvuka53uKsQ7ajD0BtiXmu8u0rwR8/J/IrzZQPrFokJ3kkPJ5IxmxpbXZGO7zGNI5xqxOY7jcNVl51c0lEn1EZJVTm8OOgxdGo2SYo12FKMUx47H1Hkl8hhSHul8/HTp5RNj/XsLR03wipNzOI5o2uB+nJNTxyRFyqGKiNvI8ZxhSgN1OgA6oZnhgq1oCNbaF2NxesQe15T2QlpAomvUiYNj09rjOo7rqcumTjGVA9wyy7HJx5dVWV9p/Dx2sMwKowK5FmLNbIp84agbMkLQRIIAr2g1mrqTENsdb78dc1P/rbDLsxx+wVKKBguTsnmHu1dgye5YLbQahQz8IxDPDUnWze0/i2mFRDGJ1x/GTEZeNRhn45hox92CW4rPtKekApIigIp8ahTBSFqtrcj7NMho6M1jXvDriuoK5OTxzJts2N7LVN/7XCQ6Jz0xmhyiIgVokIjSx2ankF+mLEbF9y+Enz72wE/HsT0uj3SKGeQT6jnFWqqMkn22G0jdCGh7zQmKwtVtwnR0TSHJPZRTJCcPnEYEAQSlV+TRRuGHegoBu52gNI8pYwDJL1A6fRYrGt6VRdDt6iFzOpyxSef2KjbT+iGY2LxGuQdY+ec1TdQvZZ5z/HINw5uc0Av99jbdZHx0SwZJTHRbR1hFBRKg211EaUgXDU0aYzp9RG6oKJAvqlJ4xx96YAEZi+lNg1q2ePdZcHgySkdSWIZbrDmHkbP5Ehk9McO49fPEN81GEGLrhl0jlxMJWEXDZFsiKUN8qhk2Nh4yYCT0iZLFfSdQa7llLcmrTLA7nm0moOrKLSi4qK4Y3Aw5fCsILwN8G860PwZT/b+F5QsIdlNuf/HGP8/3jGxRvQ6PY7O9/Aa0M0xyf0UWYSE7RVSkRK9Kxl0BI86U0wkbr4bsl5OGT2d0Nx1+TD7HK4z6rimK9kMjRPSoKD4VcJu5WOfR8j2Gl2zkErB99930OVDDqQl+7lPXRrMrwp6YU2bpIDKmbfHYDnBXgUM+xbTVR/plUOb59iljN2zsObHHJ39JbvVjO03OVJmQ1ei8lLmzHD2GrQgpa9+SCVnDFuJqLlEcpZY3RVJV2aZveT4WcFmklF3Y6J8RmN8yDLaEtVLqiRg//CUd/cmjVbTqAF9e0yxldBuz9kr/zV9raV4uaOe7dDfrNjDol4W1NMFoXzNoFtguQf8drGha7gc9Vt+9sEpTwenNNEVcjGj9XWU5gR3V/PTxiPefsU21bmeWYgIxq6BgsGe1DJLj7i43Ofxo1OSxEAvJvStGnX4R34mEFcmZmFRRzFd0yT3l6jpFMMqmE1b9vaGJJlCWtZglxiWjKJqZFlD1rq8vV+z+809e4MD4lRh01TIkY7ZqFhlS/jLe1ynxyCQUSOTgeKwx5hBaOLNB2jdY6CGpQ/CAP0J7a6lvsvIZZnLZoX75wJp746EG9qTHNNS8dfnrCIDmxJdl5EKi0bdZzsbkMxUxGbFcKHzrHfOidKDoKUtGoxEw4lcyt+V3G8jxj2PF/0+WepQrjUsz6QsfFRMzF6fg6RDsrZphYPRRniGyWOtpYxD1qGGORujqCrGpqaaNpzKJqejCUxrkqiCeYknazRlSRRVqIZCM+pj5CPc/inh1ietHVzhUJU5sl+xC0J63TEtCZWt4YhjNqt3tH6LKDrI8R1iLeF0BIqcoeoaaWvSZDL+rmI3TRhHfcS6x/0641QdUpQaklVx7Dmsyzmjg2MUU8O2VwxG56TlGxQpphs8QU8dNL3EsDM+dgV2rFNsDkk2Eo5dY8mCj/VjLOUpm2BNvKr5kXqE4q9JRAVLQRu1HA5O6TqnWIscUV1jnniEwkGRBFLc4H8pY99UHA/GxGVF0zeoW5DqlMW9T+y1KPoeAyYUqU6nOqDNZLraAOO7gnbb0vOf06o5p0KQlS0D8R5ecEWb68Qh9IcDMq3P1TaijbcMjiaYiYVR2dRxRS84QDUk7BuBEZWEpsMmz2hFSWJWHJ95rOY5eTTBFX3sQkf3JaTbGsttOIt1BtcSXWOMp6k0eYW6damtmp0csxYRqxsNu6OylmJqU8OwLDbvBNGnCoZmki90+pmEHRgczfbhskbrHJPVMZWigD1g7r/hyPbRa594l9JTBK/+uURbq3gIVProdY63chlWEnJRkK0ccqul1SwkHRrFJp8lvPEj2l6N1s1RdY0njydcvHuJvykYdM/RNJmzzKUTgI6GIUwSqSYzFNZ1wdoqqd2WkecQLxvSTR89sZD9HGWp078eYwYGSthlL+4wliQG2h7m1kS3hsj7PUQbEdU5wjPJspSNsuNleEu1KFHrmDYPeTI85YBDpLLGr6akJbS6gdF7hd6p8RPBbi5o6gH6Zp9R1Mf3bQbuPsvpa8JmgWfaPD8oqDOLfDVgEMCBOmAiDnBUk9qPkTcJeq0j9B4IDckzUU2HepdTGRXarUxjq5TFjCzekUst7cZAyILMzNhqXdpuxRkNuqGhNhptIdPfjTF20HUqlKbgSWjRVhJ126KFHcpcYuTIdMwTYkugqC7twkSXLbSog7Rp2K8TzLiPpClYRo+qCWkCjdjw2FZz5Cbl/LBHfO6j9BWkaUK21kiCgPXNEiVZUd5NsVcSH/VO8eQCZRuhFRqd0T5pE2Nhs7zxySZjSi3F0WcY/Q1NOKdNbQz/BOVNj+0q5em5TtVV2N6rqCsVqbWR7A5aPkftmVS5wBY9dtEdhVYgmS3dropZVRQrnd23Gbads2/qlEUJQiK/d+k6LUYaY0uCbqtTvZXp1h3aVlDHOhQWUj5CC0ZYN4LTpkRRJbTSg7LDIy0hq+csoy22JCNZXWSnQ9PayKio1ZqYjGC9ozYKUGTKXKIzsFmEd9x+18VxjnHsG+L0a7ZRyv7wGb6oQGlQcg3HcLi/+Q11oVBPB6TLku5coGUtraagc8hKXtM5OmdbwZk3ZpsWBHHD5etbMuULBnbLpF/RrCSyUtCrS9T1Jbu7EabS4blhUQqBnGmYrow82uOu7HN7HfLq5gsuFJmTg6fMwjvc/I/8N0FR76jyGsnUqPQd9r4A3SUWDpgRpVxRNS2qYqFVBkapUxUteq1SBIL7ao7erVmZC9JNyGZRoEoNmulSShmN2jCvS6Q8Qs1LUkVG7ji0Ro+CDCtZYuqCuichBwZKIDDdLrX7h+eIIqt48+uK8iBFGR4g9XVap6ZWXG78BaKecvJ0gl5ZFEIG94B4PUXMYoZ1TRGteKtcIzclemki7bqoio5z0KEVK+L+ikBTyDUVwzFYV1uq1mKVJFjNnKJjIbSGUs5I/RJbOOyqkkNJZzbfordj3r58C+2K86MRspHxbr2j3ZnEsxxDTuifNKiSTlY1ILc4/T6ZYdAUMSqCZZJTaDpyRyFu7tBOI6q8Ils29O0jgrDLtjxnky1QyoiO0aD1WkohkQcHJGFJXjxnm9ZUSo+sK6FKK5q4pndmcxfOCXc5RlAwPBxj2ymVdMf1W4Xxjx9xe5viWB2yLGIWbNGjLZQFk+MhWdHQyyt8/Rb90CAqSqgrHLXCv9ugTWrUXo9FsAG/oulUBJ2Q2hGMPt8nSFM8o6GuEzYzB6HtoXsS83LHTqrodTxiMyIRNlWh4dcZuquQNT6anlEFE7htEekc2hRdklhubjE8FTOq6MgGWk8hlmvioiaM3hHeb9g7+cPZ5VVdMF0UrO0QubIwS4meVRLKAVVWk0U5VlYR6Dt8vSHWVXbkKJ6MZSnM5jZl8imFdkaoFxif1YjZgvBrm7qW0LQUOg2VtiGzdNoZaJFFqUDgRVRjGVkyCOKQtFOitDpxm2LIGlf3Jr09myrwWM8i3CqilWY4SkMbb8DOyGqFZGORmQ3lpEbWFCr/E/x2wNWdRWwFiA9aFEch+H1MGEZ/+GVQKslDaEsd7A5Cl2gNCUFKblRY0gh/U7B/dMib64B4O6YKBL5cMXwacllesc8AsaiwDJPczgnNglTvEytdaklHVgoaHaaxQtYYNJGB4Vqo+w7bxEeIW7S8xOsekBpbShc8O6NObyiahCyVqFubUM3w5YTK+sMAZn+oYCoVbXvFNPERhqA1UtZLQasMSWmIZYNdBElRk/kC8ahLOY4ovYrV0iDJhmhdE1E1UOe0mo0wS3IrIFDvWVQv2Y8n6FmN3aloGhnDHiMnIAkLuTaprRKpk5NlKq1pk+Y+QbGhsGr0PRO9axLMS9JaIQkj6hFEbY1fOsyiAC83CROFTRhzOoKFvKQYFZQ6dJlQhIJkKiFNWsJdQtomYNaMnnigQby9ZzTIEVVLFiiMXBnNLQjyBJ+QXC+o4ghd20eKD4jzkEb22LYJWi9DaN8TOVNEd4e/t2BTWrhqAVqBpWhMdwOoPAq2GI8tUjUkkG1qUZL8WkFK95n0R5TlHst2yzzdYFYTNKkgdHJU06LVIY1C6hzKpU4pD7GbJ1TaOfpBimtJKI2JXu5zPbum0x1TqCGm1eIHCkJ0icIaa5TSjmsq0bDUNrRji8aQqYINek8hLzSkQc738y9QpRzrREUxTdppgPLGQPdskt4GeZRQShcgm9S5ibbfUpBRNjZlJlOoJV3PIt7IRI0Nkcu436W1dMyTPbbJBmUhoyX7lEqKMzRJs5r125peA1Vb4qsew6MT3q0u0NQC3UgZ7cl41jFVYBBuE3ZpTBhnSKaOd9AjExW3sxtmdYb9l89pHJmbzR3RbMaeE5GoXXxphax7yFWG23ZR4hKMhFStKZUdtYBVGLHazXEGOnXwnxYDUtu2/2nTBQ8ePHjw4MGD/1+S/3N/wIMHDx48ePDgP6+HGHjw4MGDBw/+xD3EwIMHDx48ePAn7iEGHjx48ODBgz9xDzHw4MGDBw8e/Il7iIEHDx48ePDgT9xDDDx48ODBgwd/4h5i4MGDBw8ePPgT9xADDx48ePDgwZ+4/w84Ydad+5PGPwAAAABJRU5ErkJggg==", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "from transformer_net import *\n", + "\n", + "device = torch.device('mps')\n", + "style_model = TransformerNet().to(device)\n", + "state_dict = torch.load('incomplete_sunday_afternoon.model', map_location=device)\n", + "\n", + "style_model.load_state_dict(state_dict)\n", + "style_model = style_model.to(device).to(torch.float16)\n", + "style_model.eval()\n", + "\n", + "in_img = img.to(device).to(torch.float16).unsqueeze(0).contiguous()\n", + "style_image = style_model(in_img)\n", + "x = style_image.squeeze(0).to(torch.float32) / 255.0\n", + "print(x.shape)\n", + "show_image(x)\n", + "\n", + "import re\n", + "\n", + "# remove saved deprecated running_* keys in InstanceNorm from the checkpoint\n", + "for k in list(state_dict.keys()):\n", + " if re.search(r'in\\d+\\.running_(mean|var)$', k):\n", + " del state_dict[k]" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "c0a08a21", + "metadata": {}, "outputs": [], "source": [] } diff --git a/demos/video/chapel-webcam/model3.ipynb b/demos/video/chapel-webcam/model3.ipynb new file mode 100644 index 00000000..2a39ddd2 --- /dev/null +++ b/demos/video/chapel-webcam/model3.ipynb @@ -0,0 +1,153 @@ +{ + "cells": [ + { + "cell_type": "code", + "execution_count": null, + "id": "15d710c9", + "metadata": {}, + "outputs": [], + "source": [ + " # with torch.no_grad():\n", + " # style_model = TransformerNet()\n", + " # state_dict = torch.load(args.model)\n", + " # # remove saved deprecated running_* keys in InstanceNorm from the checkpoint\n", + " # for k in list(state_dict.keys()):\n", + " # if re.search(r'in\\d+\\.running_(mean|var)$', k):\n", + " # del state_dict[k]\n", + " # style_model.load_state_dict(state_dict)\n", + " # style_model.to(device)\n", + " # style_model.eval()" + ] + }, + { + "cell_type": "code", + "execution_count": 1, + "id": "3e8dbbbc", + "metadata": {}, + "outputs": [], + "source": [ + "import torch\n", + "from transformer_net import *\n", + "import re\n", + "from torchsummary import summary" + ] + }, + { + "cell_type": "code", + "execution_count": 17, + "id": "ea674d56", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/var/folders/jy/f3n1221n2pq22hjdxzygxggh0000gn/T/ipykernel_54166/3029904648.py:3: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.\n", + " state_dict = torch.load('incomplete_sunday_afternoon.model', map_location=device)\n" + ] + }, + { + "ename": "RuntimeError", + "evalue": "Input type (float) and bias type (c10::Half) should be the same", + "output_type": "error", + "traceback": [ + "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", + "\u001b[0;31mRuntimeError\u001b[0m Traceback (most recent call last)", + "Cell \u001b[0;32mIn[17], line 11\u001b[0m\n\u001b[1;32m 9\u001b[0m style_model \u001b[38;5;241m=\u001b[39m style_model\u001b[38;5;241m.\u001b[39mto(device)\u001b[38;5;241m.\u001b[39mto(torch\u001b[38;5;241m.\u001b[39mfloat16)\n\u001b[1;32m 10\u001b[0m style_model\u001b[38;5;241m.\u001b[39meval()\n\u001b[0;32m---> 11\u001b[0m \u001b[43msummary\u001b[49m\u001b[43m(\u001b[49m\u001b[43mstyle_model\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mto\u001b[49m\u001b[43m(\u001b[49m\u001b[43mtorch\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mdevice\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[38;5;124;43mcpu\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[43m)\u001b[49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m3\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m1080\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m1920\u001b[39;49m\u001b[43m)\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mdevice\u001b[49m\u001b[38;5;241;43m=\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[38;5;124;43mcpu\u001b[39;49m\u001b[38;5;124;43m'\u001b[39;49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torchsummary/torchsummary.py:72\u001b[0m, in \u001b[0;36msummary\u001b[0;34m(model, input_size, batch_size, device)\u001b[0m\n\u001b[1;32m 68\u001b[0m model\u001b[38;5;241m.\u001b[39mapply(register_hook)\n\u001b[1;32m 70\u001b[0m \u001b[38;5;66;03m# make a forward pass\u001b[39;00m\n\u001b[1;32m 71\u001b[0m \u001b[38;5;66;03m# print(x.shape)\u001b[39;00m\n\u001b[0;32m---> 72\u001b[0m \u001b[43mmodel\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mx\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 74\u001b[0m \u001b[38;5;66;03m# remove these hooks\u001b[39;00m\n\u001b[1;32m 75\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m h \u001b[38;5;129;01min\u001b[39;00m hooks:\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1736\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1734\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs) \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m 1735\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1736\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1747\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1742\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1743\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1744\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m 1745\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1746\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1747\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1749\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[1;32m 1750\u001b[0m called_always_called_hooks \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mset\u001b[39m()\n", + "File \u001b[0;32m~/Documents/Github/ChAI/demos/video/chapel-webcam/transformer_net.py:30\u001b[0m, in \u001b[0;36mTransformerNet.forward\u001b[0;34m(self, X)\u001b[0m\n\u001b[1;32m 29\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, X):\n\u001b[0;32m---> 30\u001b[0m y \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mrelu(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39min1(\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mconv1\u001b[49m\u001b[43m(\u001b[49m\u001b[43mX\u001b[49m\u001b[43m)\u001b[49m))\n\u001b[1;32m 31\u001b[0m y \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mrelu(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39min2(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mconv2(y)))\n\u001b[1;32m 32\u001b[0m y \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mrelu(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39min3(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mconv3(y)))\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1736\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1734\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs) \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m 1735\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1736\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1844\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1841\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m inner()\n\u001b[1;32m 1843\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m-> 1844\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43minner\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1845\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m \u001b[38;5;167;01mException\u001b[39;00m:\n\u001b[1;32m 1846\u001b[0m \u001b[38;5;66;03m# run always called hooks if they have not already been run\u001b[39;00m\n\u001b[1;32m 1847\u001b[0m \u001b[38;5;66;03m# For now only forward hooks have the always_call option but perhaps\u001b[39;00m\n\u001b[1;32m 1848\u001b[0m \u001b[38;5;66;03m# this functionality should be added to full backward hooks as well.\u001b[39;00m\n\u001b[1;32m 1849\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m hook_id, hook \u001b[38;5;129;01min\u001b[39;00m _global_forward_hooks\u001b[38;5;241m.\u001b[39mitems():\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1790\u001b[0m, in \u001b[0;36mModule._call_impl..inner\u001b[0;34m()\u001b[0m\n\u001b[1;32m 1787\u001b[0m bw_hook \u001b[38;5;241m=\u001b[39m BackwardHook(\u001b[38;5;28mself\u001b[39m, full_backward_hooks, backward_pre_hooks)\n\u001b[1;32m 1788\u001b[0m args \u001b[38;5;241m=\u001b[39m bw_hook\u001b[38;5;241m.\u001b[39msetup_input_hook(args)\n\u001b[0;32m-> 1790\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1791\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks:\n\u001b[1;32m 1792\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m hook_id, hook \u001b[38;5;129;01min\u001b[39;00m (\n\u001b[1;32m 1793\u001b[0m \u001b[38;5;241m*\u001b[39m_global_forward_hooks\u001b[38;5;241m.\u001b[39mitems(),\n\u001b[1;32m 1794\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks\u001b[38;5;241m.\u001b[39mitems(),\n\u001b[1;32m 1795\u001b[0m ):\n\u001b[1;32m 1796\u001b[0m \u001b[38;5;66;03m# mark that always called hook is run\u001b[39;00m\n", + "File \u001b[0;32m~/Documents/Github/ChAI/demos/video/chapel-webcam/transformer_net.py:53\u001b[0m, in \u001b[0;36mConvLayer.forward\u001b[0;34m(self, x)\u001b[0m\n\u001b[1;32m 51\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, x):\n\u001b[1;32m 52\u001b[0m out \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mreflection_pad(x)\n\u001b[0;32m---> 53\u001b[0m out \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mconv2d\u001b[49m\u001b[43m(\u001b[49m\u001b[43mout\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 54\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m out\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1736\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1734\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs) \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m 1735\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1736\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1844\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1841\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m inner()\n\u001b[1;32m 1843\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m-> 1844\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43minner\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1845\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m \u001b[38;5;167;01mException\u001b[39;00m:\n\u001b[1;32m 1846\u001b[0m \u001b[38;5;66;03m# run always called hooks if they have not already been run\u001b[39;00m\n\u001b[1;32m 1847\u001b[0m \u001b[38;5;66;03m# For now only forward hooks have the always_call option but perhaps\u001b[39;00m\n\u001b[1;32m 1848\u001b[0m \u001b[38;5;66;03m# this functionality should be added to full backward hooks as well.\u001b[39;00m\n\u001b[1;32m 1849\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m hook_id, hook \u001b[38;5;129;01min\u001b[39;00m _global_forward_hooks\u001b[38;5;241m.\u001b[39mitems():\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1790\u001b[0m, in \u001b[0;36mModule._call_impl..inner\u001b[0;34m()\u001b[0m\n\u001b[1;32m 1787\u001b[0m bw_hook \u001b[38;5;241m=\u001b[39m BackwardHook(\u001b[38;5;28mself\u001b[39m, full_backward_hooks, backward_pre_hooks)\n\u001b[1;32m 1788\u001b[0m args \u001b[38;5;241m=\u001b[39m bw_hook\u001b[38;5;241m.\u001b[39msetup_input_hook(args)\n\u001b[0;32m-> 1790\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1791\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks:\n\u001b[1;32m 1792\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m hook_id, hook \u001b[38;5;129;01min\u001b[39;00m (\n\u001b[1;32m 1793\u001b[0m \u001b[38;5;241m*\u001b[39m_global_forward_hooks\u001b[38;5;241m.\u001b[39mitems(),\n\u001b[1;32m 1794\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks\u001b[38;5;241m.\u001b[39mitems(),\n\u001b[1;32m 1795\u001b[0m ):\n\u001b[1;32m 1796\u001b[0m \u001b[38;5;66;03m# mark that always called hook is run\u001b[39;00m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/conv.py:554\u001b[0m, in \u001b[0;36mConv2d.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m 553\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;28minput\u001b[39m: Tensor) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m Tensor:\n\u001b[0;32m--> 554\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_conv_forward\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mweight\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbias\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/conv.py:549\u001b[0m, in \u001b[0;36mConv2d._conv_forward\u001b[0;34m(self, input, weight, bias)\u001b[0m\n\u001b[1;32m 537\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mpadding_mode \u001b[38;5;241m!=\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mzeros\u001b[39m\u001b[38;5;124m\"\u001b[39m:\n\u001b[1;32m 538\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m F\u001b[38;5;241m.\u001b[39mconv2d(\n\u001b[1;32m 539\u001b[0m F\u001b[38;5;241m.\u001b[39mpad(\n\u001b[1;32m 540\u001b[0m \u001b[38;5;28minput\u001b[39m, \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_reversed_padding_repeated_twice, mode\u001b[38;5;241m=\u001b[39m\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mpadding_mode\n\u001b[0;32m (...)\u001b[0m\n\u001b[1;32m 547\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mgroups,\n\u001b[1;32m 548\u001b[0m )\n\u001b[0;32m--> 549\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mF\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mconv2d\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 550\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mweight\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mbias\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mstride\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mpadding\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mdilation\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mgroups\u001b[49m\n\u001b[1;32m 551\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n", + "\u001b[0;31mRuntimeError\u001b[0m: Input type (float) and bias type (c10::Half) should be the same" + ] + } + ], + "source": [ + "device = torch.device('mps')\n", + "style_model = TransformerNet().to(device)\n", + "state_dict = torch.load('incomplete_sunday_afternoon.model', map_location=device)\n", + "# remove saved deprecated running_* keys in InstanceNorm from the checkpoint\n", + "for k in list(state_dict.keys()):\n", + " if re.search(r'in\\d+\\.running_(mean|var)$', k):\n", + " del state_dict[k]\n", + "style_model.load_state_dict(state_dict)\n", + "style_model = style_model.to(device).to(torch.float16)\n", + "style_model.eval()\n", + "summary(style_model.to(torch.device('cpu')), (3, 1080, 1920), device='cpu')" + ] + }, + { + "cell_type": "code", + "execution_count": 7, + "id": "debf5db5", + "metadata": {}, + "outputs": [ + { + "ename": "RuntimeError", + "evalue": "slow_conv2d_forward_mps: input(device='cpu') and weight(device=mps:0') must be on the same device", + "output_type": "error", + "traceback": [ + "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", + "\u001b[0;31mRuntimeError\u001b[0m Traceback (most recent call last)", + "Cell \u001b[0;32mIn[7], line 1\u001b[0m\n\u001b[0;32m----> 1\u001b[0m \u001b[43msummary\u001b[49m\u001b[43m(\u001b[49m\u001b[43mstyle_model\u001b[49m\u001b[43m,\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m3\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m1080\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m1920\u001b[39;49m\u001b[43m)\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torchsummary/torchsummary.py:72\u001b[0m, in \u001b[0;36msummary\u001b[0;34m(model, input_size, batch_size, device)\u001b[0m\n\u001b[1;32m 68\u001b[0m model\u001b[38;5;241m.\u001b[39mapply(register_hook)\n\u001b[1;32m 70\u001b[0m \u001b[38;5;66;03m# make a forward pass\u001b[39;00m\n\u001b[1;32m 71\u001b[0m \u001b[38;5;66;03m# print(x.shape)\u001b[39;00m\n\u001b[0;32m---> 72\u001b[0m \u001b[43mmodel\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mx\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 74\u001b[0m \u001b[38;5;66;03m# remove these hooks\u001b[39;00m\n\u001b[1;32m 75\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m h \u001b[38;5;129;01min\u001b[39;00m hooks:\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1736\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1734\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs) \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m 1735\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1736\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1747\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1742\u001b[0m \u001b[38;5;66;03m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m 1743\u001b[0m \u001b[38;5;66;03m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m 1744\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;129;01mnot\u001b[39;00m (\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m 1745\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_backward_pre_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m 1746\u001b[0m \u001b[38;5;129;01mor\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1747\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1749\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;01mNone\u001b[39;00m\n\u001b[1;32m 1750\u001b[0m called_always_called_hooks \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mset\u001b[39m()\n", + "File \u001b[0;32m~/Documents/Github/ChAI/demos/video/chapel-webcam/transformer_net.py:30\u001b[0m, in \u001b[0;36mTransformerNet.forward\u001b[0;34m(self, X)\u001b[0m\n\u001b[1;32m 29\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, X):\n\u001b[0;32m---> 30\u001b[0m y \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mrelu(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39min1(\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mconv1\u001b[49m\u001b[43m(\u001b[49m\u001b[43mX\u001b[49m\u001b[43m)\u001b[49m))\n\u001b[1;32m 31\u001b[0m y \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mrelu(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39min2(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mconv2(y)))\n\u001b[1;32m 32\u001b[0m y \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mrelu(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39min3(\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mconv3(y)))\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1736\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1734\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs) \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m 1735\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1736\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1844\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1841\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m inner()\n\u001b[1;32m 1843\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m-> 1844\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43minner\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1845\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m \u001b[38;5;167;01mException\u001b[39;00m:\n\u001b[1;32m 1846\u001b[0m \u001b[38;5;66;03m# run always called hooks if they have not already been run\u001b[39;00m\n\u001b[1;32m 1847\u001b[0m \u001b[38;5;66;03m# For now only forward hooks have the always_call option but perhaps\u001b[39;00m\n\u001b[1;32m 1848\u001b[0m \u001b[38;5;66;03m# this functionality should be added to full backward hooks as well.\u001b[39;00m\n\u001b[1;32m 1849\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m hook_id, hook \u001b[38;5;129;01min\u001b[39;00m _global_forward_hooks\u001b[38;5;241m.\u001b[39mitems():\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1790\u001b[0m, in \u001b[0;36mModule._call_impl..inner\u001b[0;34m()\u001b[0m\n\u001b[1;32m 1787\u001b[0m bw_hook \u001b[38;5;241m=\u001b[39m BackwardHook(\u001b[38;5;28mself\u001b[39m, full_backward_hooks, backward_pre_hooks)\n\u001b[1;32m 1788\u001b[0m args \u001b[38;5;241m=\u001b[39m bw_hook\u001b[38;5;241m.\u001b[39msetup_input_hook(args)\n\u001b[0;32m-> 1790\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1791\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks:\n\u001b[1;32m 1792\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m hook_id, hook \u001b[38;5;129;01min\u001b[39;00m (\n\u001b[1;32m 1793\u001b[0m \u001b[38;5;241m*\u001b[39m_global_forward_hooks\u001b[38;5;241m.\u001b[39mitems(),\n\u001b[1;32m 1794\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks\u001b[38;5;241m.\u001b[39mitems(),\n\u001b[1;32m 1795\u001b[0m ):\n\u001b[1;32m 1796\u001b[0m \u001b[38;5;66;03m# mark that always called hook is run\u001b[39;00m\n", + "File \u001b[0;32m~/Documents/Github/ChAI/demos/video/chapel-webcam/transformer_net.py:53\u001b[0m, in \u001b[0;36mConvLayer.forward\u001b[0;34m(self, x)\u001b[0m\n\u001b[1;32m 51\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, x):\n\u001b[1;32m 52\u001b[0m out \u001b[38;5;241m=\u001b[39m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mreflection_pad(x)\n\u001b[0;32m---> 53\u001b[0m out \u001b[38;5;241m=\u001b[39m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mconv2d\u001b[49m\u001b[43m(\u001b[49m\u001b[43mout\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 54\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m out\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1736\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1734\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_compiled_call_impl(\u001b[38;5;241m*\u001b[39margs, \u001b[38;5;241m*\u001b[39m\u001b[38;5;241m*\u001b[39mkwargs) \u001b[38;5;66;03m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m 1735\u001b[0m \u001b[38;5;28;01melse\u001b[39;00m:\n\u001b[0;32m-> 1736\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_call_impl\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1844\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m 1841\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m inner()\n\u001b[1;32m 1843\u001b[0m \u001b[38;5;28;01mtry\u001b[39;00m:\n\u001b[0;32m-> 1844\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43minner\u001b[49m\u001b[43m(\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1845\u001b[0m \u001b[38;5;28;01mexcept\u001b[39;00m \u001b[38;5;167;01mException\u001b[39;00m:\n\u001b[1;32m 1846\u001b[0m \u001b[38;5;66;03m# run always called hooks if they have not already been run\u001b[39;00m\n\u001b[1;32m 1847\u001b[0m \u001b[38;5;66;03m# For now only forward hooks have the always_call option but perhaps\u001b[39;00m\n\u001b[1;32m 1848\u001b[0m \u001b[38;5;66;03m# this functionality should be added to full backward hooks as well.\u001b[39;00m\n\u001b[1;32m 1849\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m hook_id, hook \u001b[38;5;129;01min\u001b[39;00m _global_forward_hooks\u001b[38;5;241m.\u001b[39mitems():\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/module.py:1790\u001b[0m, in \u001b[0;36mModule._call_impl..inner\u001b[0;34m()\u001b[0m\n\u001b[1;32m 1787\u001b[0m bw_hook \u001b[38;5;241m=\u001b[39m BackwardHook(\u001b[38;5;28mself\u001b[39m, full_backward_hooks, backward_pre_hooks)\n\u001b[1;32m 1788\u001b[0m args \u001b[38;5;241m=\u001b[39m bw_hook\u001b[38;5;241m.\u001b[39msetup_input_hook(args)\n\u001b[0;32m-> 1790\u001b[0m result \u001b[38;5;241m=\u001b[39m \u001b[43mforward_call\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43margs\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mkwargs\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m 1791\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m _global_forward_hooks \u001b[38;5;129;01mor\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks:\n\u001b[1;32m 1792\u001b[0m \u001b[38;5;28;01mfor\u001b[39;00m hook_id, hook \u001b[38;5;129;01min\u001b[39;00m (\n\u001b[1;32m 1793\u001b[0m \u001b[38;5;241m*\u001b[39m_global_forward_hooks\u001b[38;5;241m.\u001b[39mitems(),\n\u001b[1;32m 1794\u001b[0m \u001b[38;5;241m*\u001b[39m\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_forward_hooks\u001b[38;5;241m.\u001b[39mitems(),\n\u001b[1;32m 1795\u001b[0m ):\n\u001b[1;32m 1796\u001b[0m \u001b[38;5;66;03m# mark that always called hook is run\u001b[39;00m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/conv.py:554\u001b[0m, in \u001b[0;36mConv2d.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m 553\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m\u001b[38;5;250m \u001b[39m\u001b[38;5;21mforward\u001b[39m(\u001b[38;5;28mself\u001b[39m, \u001b[38;5;28minput\u001b[39m: Tensor) \u001b[38;5;241m-\u001b[39m\u001b[38;5;241m>\u001b[39m Tensor:\n\u001b[0;32m--> 554\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43m_conv_forward\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mweight\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mbias\u001b[49m\u001b[43m)\u001b[49m\n", + "File \u001b[0;32m~/.venv/lib/python3.12/site-packages/torch/nn/modules/conv.py:549\u001b[0m, in \u001b[0;36mConv2d._conv_forward\u001b[0;34m(self, input, weight, bias)\u001b[0m\n\u001b[1;32m 537\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mpadding_mode \u001b[38;5;241m!=\u001b[39m \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mzeros\u001b[39m\u001b[38;5;124m\"\u001b[39m:\n\u001b[1;32m 538\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m F\u001b[38;5;241m.\u001b[39mconv2d(\n\u001b[1;32m 539\u001b[0m F\u001b[38;5;241m.\u001b[39mpad(\n\u001b[1;32m 540\u001b[0m \u001b[38;5;28minput\u001b[39m, \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39m_reversed_padding_repeated_twice, mode\u001b[38;5;241m=\u001b[39m\u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mpadding_mode\n\u001b[0;32m (...)\u001b[0m\n\u001b[1;32m 547\u001b[0m \u001b[38;5;28mself\u001b[39m\u001b[38;5;241m.\u001b[39mgroups,\n\u001b[1;32m 548\u001b[0m )\n\u001b[0;32m--> 549\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m \u001b[43mF\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mconv2d\u001b[49m\u001b[43m(\u001b[49m\n\u001b[1;32m 550\u001b[0m \u001b[43m \u001b[49m\u001b[38;5;28;43minput\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mweight\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mbias\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mstride\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mpadding\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mdilation\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;28;43mself\u001b[39;49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mgroups\u001b[49m\n\u001b[1;32m 551\u001b[0m \u001b[43m\u001b[49m\u001b[43m)\u001b[49m\n", + "\u001b[0;31mRuntimeError\u001b[0m: slow_conv2d_forward_mps: input(device='cpu') and weight(device=mps:0') must be on the same device" + ] + } + ], + "source": [ + "summary(style_model,(3, 1080, 1920))" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "d250324c", + "metadata": {}, + "outputs": [], + "source": [] + } + ], + "metadata": { + "kernelspec": { + "display_name": ".venv", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.12.9" + } + }, + "nbformat": 4, + "nbformat_minor": 5 +} diff --git a/demos/video/chapel-webcam/net.cpp b/demos/video/chapel-webcam/net.cpp new file mode 100644 index 00000000..160ea498 --- /dev/null +++ b/demos/video/chapel-webcam/net.cpp @@ -0,0 +1,13 @@ +#include "transformer_net.hpp" + +int main() { + torch::manual_seed(0); + TransformerNet model; + model->eval(); + + // dummy input + auto input = torch::randn({1,3,256,256}); + auto output = model->forward(input); + + std::cout << "Output shape: " << output.sizes() << "\n"; +} \ No newline at end of file diff --git a/demos/video/chapel-webcam/smol.chpl b/demos/video/chapel-webcam/smol.chpl index f97453f8..be46852e 100644 --- a/demos/video/chapel-webcam/smol.chpl +++ b/demos/video/chapel-webcam/smol.chpl @@ -3,7 +3,7 @@ use Layer; import Utilities as util; config const cpuScale: real = 0.2; -config const accelScale: real = 0.45; +config const accelScale: real = 0.8; config const debugCPUOnly: bool = false; @@ -11,47 +11,27 @@ const cpuScaleFactor = cpuScale; const acceleratorScaleFactor = accelScale; -export proc acceleratorAvailable(): bool do - return Bridge.acceleratorAvailable(); - - export proc getScaledFrameWidth(width: int): int do - if acceleratorAvailable() then + if Env.acceleratorAvailable() then return (width:real * acceleratorScaleFactor):int; else return (width:real * cpuScaleFactor):int; export proc getScaledFrameHeight(height: int): int do - if acceleratorAvailable() then + if Env.acceleratorAvailable() then return (height:real * acceleratorScaleFactor):int; else return (height:real * cpuScaleFactor):int; -if debugCPUOnly then - writeln("Debugging CPU only!"); -Bridge.debugCpuOnlyMode(debugCPUOnly); +// if debugCPUOnly then +// writeln("Debugging CPU only!"); +// Env.debugCpuOnlyMode(debugCPUOnly); writeln("CPU Scale Factor: ", cpuScaleFactor); writeln("Accelerator Scale Factor: ", acceleratorScaleFactor); -writeln("Accelerator Available: ", acceleratorAvailable()); - +writeln("Accelerator Available: ", Env.acceleratorAvailable()); -export proc square(x: int): int { - writeln(x, " * ", x, " = ", x * x); - return x * x; -} - -// export proc sumArray(a: [] int): int { -// var sum: int = 0; -// for x in a do -// sum += x; -// return sum; -// } - -export proc printArray(a: [] int): void { - writeln(a); -} use Time; use Math; @@ -72,20 +52,13 @@ const startTime = getTime(); // ../style-transfer/models/exports/mps/starry_ep3_bt4_sw1e11_cw_1e5_float32.pt // This is the one // ../style-transfer/models/exports/cpu/mosaic_float16.pt config const modelPath: string = "../style-transfer/models/exports/cpu/mosaic_float16.pt"; -var model : Bridge.torchModuleHandle; +config var isStyleTransferModel: bool = modelPath != "sobel.pt"; var modelLayer : shared TorchModule(real(32))?; -use CTypes; - -export proc globalLoadModel() { - model = Bridge.loadModel(modelPath); - if modelPath == "sobel.pt" then - modelLayer = new shared LoadedTorchModel(modelPath); - else - modelLayer = new shared StyleTransfer(modelPath); -} +export proc globalLoadModel() do + modelLayer = new shared LoadedTorchModel(modelPath); var lastFrame = startTime; @@ -108,97 +81,22 @@ export proc getNewFrame(ref frame: [] real(32),height: int, width: int,channels: const dt = getTime() - lastFrame; const fps = 1.0 / dt; - /* - runningSum += fps; - frameCount += 1; - const overallAvgFPS = runningSum / frameCount; - const idx = (frameCount - 1) % windowSize; - if frameCount <= windowSize { - // still filling the buffer - windowSum += fps; - } else { - // subtract the old value at this slot, then add the new one - windowSum += fps - fpsBuffer[idx]; - } - fpsBuffer[idx] = fps; - const currentWindowSize = min(frameCount, windowSize); - const windowAvgFPS = windowSum / currentWindowSize; - writeln("FPS: ", fps, " avg FPS: ", windowAvgFPS, "max window FPS: ", max reduce fpsBuffer); - */ - // writeln("FPS: ", fps); lastFrame = getTime(); const shape = (height,width,channels); const frameDom = util.domainFromShape((...shape)); - // const frameArr = reshape(frame,frameDom); - - if chaiImpl { - - var ndframe = new ndarray(real(32),shape); - ref ndframeData = ndframe.data; - forall idx in frameDom do - ndframeData[idx] = frame[util.linearIdx(shape,idx)]; - const dtInput = new dynamicTensor(ndframe); // 20 fps - const dtOutput = modelLayer!.forward(dtInput); - const outputFrame = dtOutput.flatten().toArray(1); - return outputFrame; - - // 2 (no) - // if copyStrat == 1 { - // var ndframe = new ndarray(real(32),shape); - // ref ndframeData = ndframe.data; - // forall idx in frameDom do - // ndframeData[idx] = frame[util.linearIdx(shape,idx)]; - // const dtInput = new dynamicTensor(ndframe); // 20 fps - // const dtOutput = modelLayer!.forward(dtInput); - // const outputFrame = dtOutput.flatten().toArray(1); - // return outputFrame; - // } else { - // const dtInput = (new dynamicTensor(frame)).reshape((...shape)); - // const dtOutput = modelLayer!.forward(dtInput); - // const outputFrame = dtOutput.flatten().toArray(1); - // return outputFrame; - // } - - - // const dtInput = new dynamicTensor(reshape(frame,frameDom)); // Way slower - - // const dtOutput = modelLayer!.forward(dtInput); - // // const outputFrame = dtOutput.rankedData(1); - // const outputFrame = dtOutput.flatten().toArray(1); - // return outputFrame; - } else { - var btFrame: Bridge.bridge_tensor_t = Bridge.createBridgeTensorWithShape(frame,shape); - var bt: Bridge.bridge_tensor_t; - if modelPath == "sobel.pt" then - bt = Bridge.modelForward(model,btFrame); - else - bt = Bridge.modelForwardStyleTransfer(model,btFrame); - - const nextNDFrame = bt : ndarray(3, real(32)); - const flattenedNextFrame = nextNDFrame.flatten().data; - return flattenedNextFrame; - } - - // forall i in 0.. + +// +// --- ConvLayer ------------------------------------------------------------- +// +struct ConvLayerImpl : torch::nn::Module { + torch::nn::ReflectionPad2d reflection_pad{nullptr}; + torch::nn::Conv2d conv2d{nullptr}; + + ConvLayerImpl(int64_t in_channels, + int64_t out_channels, + int64_t kernel_size, + int64_t stride) + : reflection_pad(torch::nn::ReflectionPad2dOptions(kernel_size / 2)), + conv2d(torch::nn::Conv2dOptions(in_channels, out_channels, kernel_size) + .stride(stride)) + { + register_module("reflection_pad", reflection_pad); + register_module("conv2d", conv2d); + } + + torch::Tensor forward(torch::Tensor x) { + x = reflection_pad->forward(x); + x = conv2d->forward(x); + return x; + } +}; +TORCH_MODULE(ConvLayer); + +// +// --- ResidualBlock --------------------------------------------------------- +// +struct ResidualBlockImpl : torch::nn::Module { + ConvLayer conv1{nullptr}; + torch::nn::InstanceNorm2d in1{nullptr}; + ConvLayer conv2{nullptr}; + torch::nn::InstanceNorm2d in2{nullptr}; + torch::nn::ReLU relu{nullptr}; + + ResidualBlockImpl(int64_t channels) + : conv1(ConvLayer(channels, channels, 3, 1)), + in1(torch::nn::InstanceNorm2dOptions(channels).affine(true)), + conv2(ConvLayer(channels, channels, 3, 1)), + in2(torch::nn::InstanceNorm2dOptions(channels).affine(true)), + relu(torch::nn::ReLUOptions(true)) + { + register_module("conv1", conv1); + register_module("in1", in1); + register_module("conv2", conv2); + register_module("in2", in2); + register_module("relu", relu); + } + + torch::Tensor forward(torch::Tensor x) { + auto residual = x; + auto out = relu->forward(in1->forward(conv1->forward(x))); + out = in2->forward(conv2->forward(out)); + return out + residual; + } +}; +TORCH_MODULE(ResidualBlock); + +// +// --- UpsampleConvLayer ----------------------------------------------------- +// +struct UpsampleConvLayerImpl : torch::nn::Module { + torch::nn::Upsample upsample{nullptr}; + torch::nn::ReflectionPad2d reflection_pad{nullptr}; + torch::nn::Conv2d conv2d{nullptr}; + + UpsampleConvLayerImpl(int64_t in_channels, + int64_t out_channels, + int64_t kernel_size, + int64_t stride, + int64_t upsample_scale) + : upsample(torch::nn::UpsampleOptions() + .scale_factor(std::vector{(double)upsample_scale, (double)upsample_scale}) + .mode(torch::kNearest)), + reflection_pad(torch::nn::ReflectionPad2dOptions(kernel_size / 2)), + conv2d(torch::nn::Conv2dOptions(in_channels, out_channels, kernel_size) + .stride(stride)) + { + register_module("upsample", upsample); + register_module("reflection_pad", reflection_pad); + register_module("conv2d", conv2d); + } + + torch::Tensor forward(torch::Tensor x) { + x = upsample->forward(x); + x = reflection_pad->forward(x); + x = conv2d->forward(x); + return x; + } +}; +TORCH_MODULE(UpsampleConvLayer); + +// +// --- TransformerNet -------------------------------------------------------- +// +struct TransformerNetImpl : torch::nn::Module { + ConvLayer conv1{nullptr}; + torch::nn::InstanceNorm2d in1{nullptr}; + ConvLayer conv2{nullptr}; + torch::nn::InstanceNorm2d in2{nullptr}; + ConvLayer conv3{nullptr}; + torch::nn::InstanceNorm2d in3{nullptr}; + ResidualBlock res1{nullptr}, res2{nullptr}, res3{nullptr}, + res4{nullptr}, res5{nullptr}; + UpsampleConvLayer deconv1{nullptr}, deconv2{nullptr}; + torch::nn::InstanceNorm2d in4{nullptr}, in5{nullptr}; + ConvLayer deconv3{nullptr}; + torch::nn::ReLU relu{nullptr}; + + TransformerNetImpl() + : conv1(ConvLayer( 3, 32, 9, 1)), + in1 (torch::nn::InstanceNorm2dOptions(32).affine(true)), + conv2(ConvLayer( 32, 64, 3, 2)), + in2 (torch::nn::InstanceNorm2dOptions(64).affine(true)), + conv3(ConvLayer( 64, 128, 3, 2)), + in3 (torch::nn::InstanceNorm2dOptions(128).affine(true)), + res1 (ResidualBlock(128)), + res2 (ResidualBlock(128)), + res3 (ResidualBlock(128)), + res4 (ResidualBlock(128)), + res5 (ResidualBlock(128)), + deconv1(UpsampleConvLayer(128, 64, 3, 1, 2)), + in4 (torch::nn::InstanceNorm2dOptions(64).affine(true)), + deconv2(UpsampleConvLayer( 64, 32, 3, 1, 2)), + in5 (torch::nn::InstanceNorm2dOptions(32).affine(true)), + deconv3(ConvLayer( 32, 3, 9, 1)), + relu (torch::nn::ReLUOptions(true)) + { + register_module("conv1", conv1); + register_module("in1", in1); + register_module("conv2", conv2); + register_module("in2", in2); + register_module("conv3", conv3); + register_module("in3", in3); + register_module("res1", res1); + register_module("res2", res2); + register_module("res3", res3); + register_module("res4", res4); + register_module("res5", res5); + register_module("deconv1", deconv1); + register_module("in4", in4); + register_module("deconv2", deconv2); + register_module("in5", in5); + register_module("deconv3", deconv3); + register_module("relu", relu); + } + + torch::Tensor forward(torch::Tensor x) { + x = relu->forward(in1->forward(conv1->forward(x))); + x = relu->forward(in2->forward(conv2->forward(x))); + x = relu->forward(in3->forward(conv3->forward(x))); + x = res1->forward(x); + x = res2->forward(x); + x = res3->forward(x); + x = res4->forward(x); + x = res5->forward(x); + x = relu->forward(in4->forward(deconv1->forward(x))); + x = relu->forward(in5->forward(deconv2->forward(x))); + x = deconv3->forward(x); + return x; + } +}; +TORCH_MODULE(TransformerNet); diff --git a/demos/video/cpp-model-construction/CMakeLists.txt b/demos/video/cpp-model-construction/CMakeLists.txt new file mode 100644 index 00000000..ff3955ad --- /dev/null +++ b/demos/video/cpp-model-construction/CMakeLists.txt @@ -0,0 +1,34 @@ +include(CMakePrintHelpers) + +# project(MyProject) +# set(CMAKE_CXX_STANDARD 17) +# list(APPEND CMAKE_PREFIX_PATH "${CMAKE_CURRENT_SOURCE_DIR}/libtorch/share/cmake") +find_package(Torch REQUIRED) +find_package(OpenCV REQUIRED) +SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++23 -lm -ldl") + +cmake_print_variables(TORCH_LIBRARIES) +cmake_print_variables(TORCH_INCLUDE_DIRS) +cmake_print_variables(TORCH_INSTALL_PREFIX) +cmake_print_variables(TORCH_CXX_FLAGS) +cmake_print_variables(TORCH_LIBRARY) + +add_executable(CPPModelConstruction ${CMAKE_CURRENT_SOURCE_DIR}/net.cpp ${CMAKE_CURRENT_SOURCE_DIR}/transformer_net.hpp) + +target_link_libraries(CPPModelConstruction ${TORCH_LIBRARIES}) +set_property(TARGET CPPModelConstruction PROPERTY CXX_STANDARD 23) + +set_property(TARGET CPPModelConstruction PROPERTY + RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR} +) + +add_executable(CPPModelCam ${CMAKE_CURRENT_SOURCE_DIR}/cam.cpp ${CMAKE_CURRENT_SOURCE_DIR}/transformer_net.hpp) + +include_directories(${OpenCV_INCLUDE_DIRS}) + +target_link_libraries(CPPModelCam ${TORCH_LIBRARIES} ${OpenCV_LIBS}) +set_property(TARGET CPPModelCam PROPERTY CXX_STANDARD 23) + +set_property(TARGET CPPModelCam PROPERTY + RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR} +) \ No newline at end of file diff --git a/demos/video/cpp-model-construction/cam.cpp b/demos/video/cpp-model-construction/cam.cpp new file mode 100644 index 00000000..50aeafba --- /dev/null +++ b/demos/video/cpp-model-construction/cam.cpp @@ -0,0 +1,146 @@ +#include "transformer_net.hpp" + + +#include +#include + + +TransformerNet load_net() { + torch::manual_seed(0); + std::cout << "set seed\n"; + TransformerNet model; + std::cout << "TransformerNet model created.\n"; + + model->load_parameters("/Users/iainmoncrief/Documents/Github/ChAI/demos/video/cpp-model-construction/state_dict_raw.pt"); + // torch::serialize::InputArchive archive; + // archive.load_from("/Users/iainmoncrief/Documents/Github/ChAI/demos/video/cpp-model-construction/incomplete_sunday_afternoon.model"); // load the raw weights :contentReference[oaicite:2]{index=2} + // std::cout << "Loading model from archive...\n"; + // model->load(archive); + // std::cout << "Model loaded successfully!\n"; + // model->eval(); + std::cout << "Model is in evaluation mode.\n"; + + // dummy input + auto input = torch::randn({1,3,256,256}); + std::cout << "Input shape: " << input.sizes() << "\n"; + auto output = model->forward(input); + + std::cout << "Output shape: " << output.sizes() << "\n"; + + return model; +} + + +cv::Mat new_frame(cv::Mat &frame,TransformerNet &model) { + + cv::Mat rgb_float_frame; + cv::cvtColor(frame, rgb_float_frame, cv::COLOR_BGR2RGB); + rgb_float_frame.convertTo(rgb_float_frame, CV_32FC3, 1.0f/255.0f); + + // cv::MatSize size = rgb_frame.size; + // std::cout << "x " << size[0] << " y " << size[1] << " channels " << rgb_frame.dims << std::endl; + int64_t height = rgb_float_frame.rows; + int64_t width = rgb_float_frame.cols; + int64_t channels = rgb_float_frame.channels(); + int64_t pixels = rgb_float_frame.total(); + int64_t size = pixels * channels; + + // std::cout << "Width: " << width << ", Height: " << height << ", Channels: " << channels << ", Size: " << size << std::endl; + + torch::Tensor tensor = torch::from_blob(rgb_float_frame.data, + {height,width, channels}, + torch::kFloat32).to(torch::kCPU, torch::kFloat32, /*non_blocking=*/false, /*copy=*/true); + + tensor = tensor.permute({2, 0, 1}).unsqueeze(0).contiguous(); + + torch::Tensor output_tensor = model->forward(tensor); + output_tensor = output_tensor.squeeze(0).permute({1, 2, 0}).contiguous(); + output_tensor = output_tensor.to(torch::kCPU, torch::kFloat32, /*non_blocking=*/false, /*copy=*/true); + output_tensor.div_(255.0); + + // chpl_external_array + // rgb_float_frame_data_ptr = chpl_make_external_array_ptr(rgb_float_frame.data,size); + + // chpl_external_array + // rgb_float_output_frame_array = getNewFrame(&rgb_float_frame_data_ptr, height, width, channels); + + + // cv::Mat new_rgb_frame(height, width, CV_8UC3,new_frame_array.elts); + // cv::cvtColor(new_rgb_frame, new_rgb_frame, cv::COLOR_RGB2BGR); + + cv::Mat output_frame(height,width,CV_32FC3,output_tensor.data_ptr()); // frame to write to + output_frame.convertTo(output_frame, CV_8UC3, 255.0f); + cv::cvtColor(output_frame, output_frame, cv::COLOR_RGB2BGR); + + return output_frame; + + +} + + +int mirror() { + cv::VideoCapture cap(0); + if (!cap.isOpened()) { + std::cerr << "Error: Cannot open the webcam.\n"; + return -1; + } + + TransformerNet model = load_net(); + + cv::Mat frame; + const std::string windowName = "Webcam Feed"; + cv::namedWindow(windowName, cv::WINDOW_AUTOSIZE); + + cv::Size original_frame_size; + cv::Size processed_frame_size; + + while (true) { + + uint64_t start = cv::getTickCount(); + + // Capture a new frame from webcam + cap >> frame; + if (frame.empty()) { + std::cerr << "Error: Empty frame captured.\n"; + break; + } + + original_frame_size = frame.size(); + + const int width = (original_frame_size.width * 0.2); + const int height = (original_frame_size.height * 0.2); + processed_frame_size = cv::Size(width, height); + cv::resize(frame, frame, processed_frame_size); + + // std::cout << "Frame size: " << frame.size() << std::endl; + // std::cout << "New frame size: " << processed_frame_size << std::endl; + + cv::Mat next_frame = new_frame(frame,model); + + cv::resize(next_frame, next_frame, original_frame_size); + + // Display the captured frame + cv::imshow(windowName, next_frame); + + // Wait for 30ms or until 'q' key is pressed + char key = static_cast(cv::waitKey(1)); + if (key == 'q' || key == 27) { // 'q' or ESC to quit + break; + } + + double fps = cv::getTickFrequency() / (cv::getTickCount() - start); + std::cout << "\rcv::FPS : " << fps << "\t\r" << std::flush; + } + + // Release the camera and destroy all windows + cap.release(); + cv::destroyAllWindows(); + return 0; +} + + +int main(int argc, char* argv[]) { + return mirror(); +} + + diff --git a/demos/video/cpp-model-construction/incomplete_sunday_afternoon.model b/demos/video/cpp-model-construction/incomplete_sunday_afternoon.model new file mode 100644 index 00000000..2576c9aa Binary files /dev/null and b/demos/video/cpp-model-construction/incomplete_sunday_afternoon.model differ diff --git a/demos/video/cpp-model-construction/model.ipynb b/demos/video/cpp-model-construction/model.ipynb new file mode 100644 index 00000000..5700f333 --- /dev/null +++ b/demos/video/cpp-model-construction/model.ipynb @@ -0,0 +1,3595 @@ +{ + "cells": [ + { + "cell_type": "code", + "execution_count": 2, + "id": "f86646b5", + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/var/folders/jy/f3n1221n2pq22hjdxzygxggh0000gn/T/ipykernel_58755/3064374510.py:7: FutureWarning: You are using `torch.load` with `weights_only=False` (the current default value), which uses the default pickle module implicitly. It is possible to construct malicious pickle data which will execute arbitrary code during unpickling (See https://github.com/pytorch/pytorch/blob/main/SECURITY.md#untrusted-models for more details). In a future release, the default value for `weights_only` will be flipped to `True`. This limits the functions that could be executed during unpickling. Arbitrary objects will no longer be allowed to be loaded via this mode unless they are explicitly allowlisted by the user via `torch.serialization.add_safe_globals`. We recommend you start setting `weights_only=True` for any use case where you don't have full control of the loaded file. Please open an issue on GitHub for any issues related to this experimental feature.\n", + " state_dict = torch.load('incomplete_sunday_afternoon.model', map_location=device)\n" + ] + } + ], + "source": [ + "import torch\n", + "from transformer_net import *\n", + "import re\n", + "\n", + "device = torch.device('cpu')\n", + "style_model = TransformerNet().to(device)\n", + "state_dict = torch.load('incomplete_sunday_afternoon.model', map_location=device)\n", + "# # remove saved deprecated running_* keys in InstanceNorm from the checkpoint\n", + "# for k in list(state_dict.keys()):\n", + "# if re.search(r'in\\d+\\.running_(mean|var)$', k):\n", + "# del state_dict[k]\n", + "style_model.load_state_dict(state_dict)\n", + "style_model = style_model.to(device)# .to(torch.float16)\n", + "# style_model.eval()\n", + "\n", + "# in_img = img.to(device).to(torch.float16).unsqueeze(0).contiguous()\n", + "# style_image = style_model(in_img)\n", + "# x = style_image.squeeze(0).to(torch.float32) / 255.0\n", + "# print(x.shape)\n", + "# show_image(x)\n", + "\n", + "\n", + "\n", + "\n" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "id": "09e1b145", + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "{'conv1.conv2d.weight': tensor([[[[-3.9299e-03, 1.6051e-02, -7.3032e-02, ..., -1.0765e-02,\n", + " 4.8624e-02, -2.6804e-02],\n", + " [-8.9465e-03, -6.2304e-02, -2.3795e-02, ..., 4.0336e-02,\n", + " 8.3711e-02, 6.3736e-02],\n", + " [-7.4396e-02, -7.2950e-02, 4.3554e-02, ..., 2.5187e-02,\n", + " 5.4569e-02, 1.0156e-01],\n", + " ...,\n", + " [ 2.4696e-02, 8.0715e-03, 1.0228e-01, ..., 1.2915e-02,\n", + " 2.7674e-02, -3.5677e-03],\n", + " [-1.2024e-03, 5.8878e-02, 4.8657e-02, ..., -5.3146e-02,\n", + " -4.7457e-02, 2.3715e-02],\n", + " [ 2.7764e-02, 8.7959e-02, 9.0724e-02, ..., -1.2441e-02,\n", + " 1.2191e-02, -1.8631e-02]],\n", + " \n", + " [[ 1.6009e-01, 1.2714e-01, -2.3240e-02, ..., -2.2687e-02,\n", + " 6.8349e-02, 2.4296e-02],\n", + " [ 8.8018e-02, 5.6891e-02, 3.9254e-02, ..., -3.2902e-02,\n", + " 1.7921e-02, -6.3808e-03],\n", + " [-7.0373e-03, 2.4201e-02, -2.2446e-02, ..., -1.6539e-01,\n", + " -1.3416e-01, -3.9397e-02],\n", + " ...,\n", + " [ 1.4631e-02, -3.9055e-02, -1.3906e-01, ..., -4.3439e-02,\n", + " 1.3496e-02, -8.4453e-03],\n", + " [-6.7090e-03, -2.3043e-02, -2.4594e-02, ..., -2.2953e-01,\n", + " -9.2886e-02, 6.1065e-03],\n", + " [ 1.4606e-01, 1.3300e-01, -3.8568e-02, ..., -7.6237e-02,\n", + " 9.8798e-03, 1.6153e-02]],\n", + " \n", + " [[-8.2964e-02, -7.4784e-02, -2.4376e-02, ..., -6.0361e-02,\n", + " -1.7850e-02, 1.9769e-02],\n", + " [-5.9817e-02, 3.5059e-02, 3.1153e-02, ..., 2.0752e-02,\n", + " 8.2601e-02, -5.8434e-02],\n", + " [-1.8644e-02, -3.9996e-02, 6.7470e-02, ..., 7.8665e-02,\n", + " 3.3943e-02, -3.5367e-02],\n", + " ...,\n", + " [-5.2988e-02, -4.8773e-02, 4.6402e-02, ..., 3.3639e-01,\n", + " 2.1004e-01, 3.7482e-02],\n", + " [ 2.7696e-02, -7.5495e-02, -5.7956e-02, ..., 1.5444e-01,\n", + " 1.1335e-01, 9.3938e-02],\n", + " [ 6.2344e-02, -2.6333e-02, -2.0012e-05, ..., 6.8237e-02,\n", + " 7.0619e-02, -7.6059e-02]]],\n", + " \n", + " \n", + " [[[ 4.1151e-03, 5.1058e-02, -3.4246e-02, ..., -4.8661e-02,\n", + " -4.8574e-02, 2.9167e-02],\n", + " [ 5.8817e-02, -5.1065e-02, -1.0689e-01, ..., -9.8355e-02,\n", + " -7.6885e-02, 1.3258e-02],\n", + " [ 6.9126e-02, -1.1277e-02, -6.6503e-02, ..., -2.2912e-02,\n", + " 1.5037e-02, 6.0764e-02],\n", + " ...,\n", + " [-3.3341e-02, -1.0241e-02, -1.8529e-01, ..., 1.2651e-01,\n", + " 9.0661e-02, 2.9401e-02],\n", + " [ 3.3107e-02, 4.8843e-04, -6.9780e-02, ..., 1.2259e-01,\n", + " 1.6094e-02, 8.3104e-02],\n", + " [ 8.6126e-02, -1.7280e-02, -2.9286e-02, ..., 5.2551e-02,\n", + " -5.0136e-02, 3.2835e-02]],\n", + " \n", + " [[-1.5543e-02, 6.3408e-03, 3.0804e-02, ..., -6.1500e-02,\n", + " -9.3639e-02, 4.4828e-03],\n", + " [-5.9363e-03, -1.0242e-01, -5.7489e-05, ..., 1.0954e-01,\n", + " -1.5017e-02, -1.2241e-02],\n", + " [ 6.8831e-02, 4.2724e-02, 6.9807e-03, ..., 1.4413e-01,\n", + " 7.3682e-02, 1.3363e-02],\n", + " ...,\n", + " [ 1.0109e-01, 1.3335e-01, 8.9106e-02, ..., 2.4730e-01,\n", + " 1.4795e-01, 6.2162e-02],\n", + " [ 6.2083e-02, 2.0420e-03, 3.6361e-02, ..., 9.3002e-02,\n", + " -4.3166e-02, 4.0400e-02],\n", + " [ 5.4498e-02, -1.7541e-03, -5.5555e-02, ..., -3.1602e-03,\n", + " 1.6037e-02, -4.5101e-03]],\n", + " \n", + " [[-9.4704e-02, 5.2836e-03, 6.9087e-03, ..., -7.1602e-02,\n", + " 1.8066e-02, -1.7624e-03],\n", + " [-7.8684e-02, -8.7348e-02, -8.6503e-02, ..., -3.8982e-02,\n", + " -6.1242e-02, 5.6806e-02],\n", + " [ 4.9512e-02, 2.0009e-02, -7.3918e-02, ..., 2.7104e-02,\n", + " -6.1450e-02, 7.3878e-02],\n", + " ...,\n", + " [ 3.6283e-02, 3.7192e-03, -1.6979e-02, ..., -1.3190e-02,\n", + " -7.0449e-02, -4.1303e-02],\n", + " [-1.6897e-02, 6.3672e-03, 1.6297e-02, ..., -1.6848e-02,\n", + " -1.1756e-01, -2.5739e-02],\n", + " [ 3.7393e-02, -2.8365e-02, 1.1671e-02, ..., 8.3497e-03,\n", + " -2.7878e-02, 9.0953e-02]]],\n", + " \n", + " \n", + " [[[ 6.5927e-02, 4.1260e-02, 1.6523e-02, ..., 5.8615e-02,\n", + " -4.1055e-02, 8.0447e-03],\n", + " [ 1.5640e-02, -2.9976e-02, -8.8650e-03, ..., 5.0282e-02,\n", + " 3.4757e-02, 4.9386e-02],\n", + " [ 3.5966e-02, -5.1197e-02, 3.2304e-02, ..., 5.0377e-03,\n", + " 5.6764e-02, 7.2514e-03],\n", + " ...,\n", + " [-3.8467e-02, 3.8144e-02, 5.0881e-02, ..., -1.8606e-01,\n", + " 5.2689e-02, 5.1072e-02],\n", + " [ 4.9867e-02, -3.6796e-02, -8.9543e-03, ..., 3.1899e-02,\n", + " 8.3706e-02, 4.7856e-02],\n", + " [ 8.8847e-03, 6.4377e-02, -1.2926e-04, ..., -3.2778e-02,\n", + " 7.1868e-02, 4.2962e-02]],\n", + " \n", + " [[-4.3934e-02, 2.9331e-03, -1.2772e-02, ..., -6.1187e-02,\n", + " -7.0258e-02, -8.9908e-02],\n", + " [-3.4995e-02, -4.1705e-02, 7.4687e-02, ..., -5.9540e-02,\n", + " -3.4429e-02, -7.2259e-02],\n", + " [-7.0697e-02, 4.2653e-02, 1.3623e-01, ..., -1.7136e-02,\n", + " -9.3556e-02, -1.9235e-02],\n", + " ...,\n", + " [ 1.3722e-02, 5.4134e-02, 4.9093e-02, ..., -2.4604e-01,\n", + " -1.7036e-01, -1.7090e-01],\n", + " [-3.4938e-02, 1.4673e-02, 4.8408e-02, ..., -1.5356e-01,\n", + " -1.5618e-01, -2.6017e-01],\n", + " [-7.0845e-02, -4.2586e-02, 7.2623e-03, ..., -1.1416e-01,\n", + " -2.1105e-01, -1.4051e-01]],\n", + " \n", + " [[-5.3066e-02, -2.7455e-02, -3.8730e-04, ..., 2.7151e-02,\n", + " 3.4699e-02, 3.0108e-02],\n", + " [ 9.0818e-02, 9.6595e-03, 3.8691e-03, ..., -1.7960e-02,\n", + " 5.8105e-02, 1.2492e-01],\n", + " [ 6.1543e-04, 5.3223e-02, -8.4417e-02, ..., -1.7101e-02,\n", + " 7.7818e-02, 8.3146e-02],\n", + " ...,\n", + " [-5.0974e-02, 3.4401e-02, -1.3781e-02, ..., 5.0339e-02,\n", + " 2.9482e-01, 1.6374e-01],\n", + " [-3.5514e-02, -1.4360e-02, -2.2753e-02, ..., 1.7795e-01,\n", + " 2.4531e-01, 1.3863e-01],\n", + " [ 4.0031e-02, 3.7818e-02, 6.9576e-02, ..., 1.2454e-01,\n", + " 3.8821e-02, 1.1691e-01]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[ 5.9689e-03, -3.9445e-03, -6.4180e-02, ..., -1.9622e-02,\n", + " 9.6511e-03, -4.3269e-02],\n", + " [-3.8949e-02, -5.3979e-03, 6.6284e-02, ..., -5.1375e-02,\n", + " 6.5137e-02, -5.7254e-03],\n", + " [-3.9039e-02, -8.7159e-03, 2.0057e-02, ..., -1.4860e-02,\n", + " -1.9306e-02, 7.7894e-02],\n", + " ...,\n", + " [ 3.1782e-02, 1.7745e-02, 5.9499e-02, ..., 2.6831e-02,\n", + " 7.3018e-03, 6.0524e-02],\n", + " [-2.5640e-02, 9.4947e-04, 1.1179e-01, ..., 1.1144e-01,\n", + " -3.7578e-02, 7.8067e-02],\n", + " [-1.1412e-01, -2.8185e-02, 9.1961e-03, ..., -1.2786e-02,\n", + " 2.1312e-02, 1.3213e-01]],\n", + " \n", + " [[ 6.5299e-02, 5.8833e-02, 1.5936e-02, ..., -1.0432e-03,\n", + " 5.3750e-02, -1.5283e-02],\n", + " [ 5.3870e-02, 5.7011e-02, 8.9907e-02, ..., -1.2940e-02,\n", + " 2.7458e-02, -2.6597e-02],\n", + " [ 1.0445e-01, 5.9487e-02, 6.0681e-02, ..., -1.0439e-01,\n", + " -6.4704e-02, -2.0163e-02],\n", + " ...,\n", + " [ 4.4688e-02, 1.3435e-01, -1.1490e-02, ..., -1.1524e-01,\n", + " -5.2772e-02, -4.8109e-02],\n", + " [ 7.8147e-02, 1.4216e-01, 4.6629e-02, ..., 3.3138e-03,\n", + " -1.8068e-02, -2.2840e-02],\n", + " [ 1.3828e-01, 7.5218e-02, 1.0053e-01, ..., -4.2658e-02,\n", + " 7.3754e-03, 1.1799e-02]],\n", + " \n", + " [[-4.1182e-02, -4.6726e-02, 3.7973e-02, ..., 5.5960e-02,\n", + " -4.0192e-02, -1.5201e-02],\n", + " [ 2.3141e-02, -3.4249e-03, -7.6097e-02, ..., 8.2330e-02,\n", + " -1.0553e-02, 2.1423e-02],\n", + " [-1.0040e-01, -3.2713e-02, -7.4553e-02, ..., 1.6148e-01,\n", + " 1.5612e-01, 1.5796e-02],\n", + " ...,\n", + " [-6.5545e-02, -1.1267e-01, -1.4225e-01, ..., 1.5562e-01,\n", + " 5.9603e-02, -4.8523e-02],\n", + " [-9.4859e-02, -9.8251e-02, -1.5715e-01, ..., -1.7737e-03,\n", + " -4.0935e-02, -4.4255e-02],\n", + " [-6.8970e-02, -2.2807e-02, -3.7874e-02, ..., 2.2954e-02,\n", + " -4.1764e-02, 1.5797e-02]]],\n", + " \n", + " \n", + " [[[-5.2727e-02, 3.1790e-02, 1.4703e-03, ..., 5.8733e-02,\n", + " -1.1273e-02, 8.9067e-02],\n", + " [ 8.0143e-03, 2.9625e-03, 4.4554e-02, ..., 7.8509e-02,\n", + " 5.2079e-02, 5.7547e-02],\n", + " [-3.0533e-02, -7.8940e-02, 8.8668e-02, ..., 3.4596e-02,\n", + " 8.2150e-02, 3.2914e-02],\n", + " ...,\n", + " [ 3.0537e-02, -9.6512e-03, 4.1123e-02, ..., -6.3499e-02,\n", + " -4.8034e-02, 8.1432e-03],\n", + " [ 6.2784e-03, -2.6794e-02, 3.8131e-02, ..., -2.5192e-02,\n", + " 4.9604e-03, 3.1324e-02],\n", + " [ 3.6372e-02, 6.2163e-02, 4.3476e-02, ..., 2.1853e-02,\n", + " 1.1732e-02, 4.9426e-02]],\n", + " \n", + " [[ 7.7890e-02, 4.0147e-02, 1.1735e-01, ..., -2.7409e-02,\n", + " 4.5920e-02, 3.2203e-02],\n", + " [ 1.0550e-01, 1.6068e-02, 1.0526e-01, ..., 6.6541e-03,\n", + " -5.2564e-02, 1.0041e-01],\n", + " [ 1.3362e-01, 1.2555e-01, 6.1613e-02, ..., -1.5627e-02,\n", + " 5.6380e-02, 1.2230e-02],\n", + " ...,\n", + " [ 1.2086e-03, -6.0567e-03, -1.2562e-01, ..., -1.5578e-01,\n", + " -5.6307e-02, -3.2801e-02],\n", + " [-2.6044e-02, 7.5125e-02, -2.9542e-02, ..., -1.4322e-02,\n", + " 8.7715e-02, 7.0933e-02],\n", + " [ 8.8050e-02, 1.0286e-01, 1.0028e-03, ..., 1.8339e-02,\n", + " 6.3141e-02, 2.0684e-01]],\n", + " \n", + " [[-5.4082e-02, -8.9115e-02, 1.8238e-02, ..., 9.4396e-02,\n", + " 1.7950e-02, 8.0475e-02],\n", + " [-1.8161e-02, 6.4328e-02, -2.5183e-02, ..., 1.1347e-01,\n", + " 3.0215e-02, 8.9125e-03],\n", + " [-2.2818e-03, 5.7427e-02, 2.4098e-02, ..., 1.2350e-01,\n", + " 1.6133e-02, -2.6325e-02],\n", + " ...,\n", + " [-8.2397e-02, 3.0207e-02, 2.1304e-02, ..., 1.0406e-01,\n", + " 5.5909e-02, 1.2452e-02],\n", + " [-6.4091e-02, 5.3392e-02, -1.2806e-03, ..., 1.1005e-01,\n", + " 6.1090e-02, 2.2959e-02],\n", + " [-1.2733e-02, 2.7388e-02, 6.1486e-02, ..., 1.0875e-01,\n", + " 5.1922e-02, 1.1976e-01]]],\n", + " \n", + " \n", + " [[[ 7.4403e-02, 2.0773e-02, 1.8429e-02, ..., 3.6365e-02,\n", + " 1.2104e-02, -8.6513e-02],\n", + " [ 8.1371e-02, -2.7246e-02, 7.2651e-02, ..., 3.0960e-02,\n", + " -7.7529e-02, -6.8916e-02],\n", + " [-2.6744e-03, 8.7665e-02, 2.2528e-01, ..., 1.0880e-02,\n", + " -9.9100e-02, -2.8111e-02],\n", + " ...,\n", + " [ 1.5822e-01, 6.4454e-02, 1.6089e-02, ..., -2.2300e-01,\n", + " -9.5034e-02, -7.3942e-02],\n", + " [ 7.0927e-02, -7.4060e-03, 4.8322e-02, ..., -8.8901e-02,\n", + " -2.2420e-02, 2.4428e-02],\n", + " [ 7.6849e-02, -3.0120e-02, -5.9724e-02, ..., -2.9184e-02,\n", + " -5.4260e-02, 7.2335e-02]],\n", + " \n", + " [[-2.0276e-02, 3.9972e-02, -9.8132e-03, ..., 1.3052e-01,\n", + " 1.1879e-02, -5.1645e-02],\n", + " [-2.8924e-03, -4.7530e-02, 5.1663e-02, ..., 1.5644e-01,\n", + " 8.6122e-02, 9.8011e-02],\n", + " [-3.2250e-02, 5.3989e-02, 5.9623e-02, ..., 4.1628e-02,\n", + " 1.8125e-01, 7.2782e-02],\n", + " ...,\n", + " [-8.6995e-03, -9.8926e-03, -4.9757e-02, ..., -6.3228e-03,\n", + " -1.8954e-02, -2.2934e-02],\n", + " [-1.6437e-02, 3.3291e-02, -5.7570e-02, ..., -8.7426e-02,\n", + " 2.5421e-02, -2.2182e-02],\n", + " [ 1.0343e-01, -2.2553e-02, -1.1138e-02, ..., 1.6341e-02,\n", + " 2.9332e-02, -6.4060e-03]],\n", + " \n", + " [[-7.8285e-02, -3.8115e-02, -2.1521e-03, ..., -5.8638e-02,\n", + " -5.4498e-02, -1.7034e-02],\n", + " [-4.0067e-02, -8.7771e-02, -6.6574e-02, ..., -9.5918e-03,\n", + " -9.5635e-03, -2.3223e-02],\n", + " [-1.8750e-02, -8.6207e-02, -6.3324e-02, ..., 2.2328e-02,\n", + " -2.8383e-02, 2.3249e-02],\n", + " ...,\n", + " [-6.2350e-02, 3.3692e-02, 6.5531e-03, ..., 2.3007e-01,\n", + " 1.5306e-01, 3.7071e-02],\n", + " [ 2.4884e-02, -1.8626e-02, 3.4846e-02, ..., 8.4937e-02,\n", + " 1.7732e-02, 4.6905e-02],\n", + " [-7.3430e-02, -8.6837e-02, 3.3609e-02, ..., 1.0441e-01,\n", + " -7.7607e-03, 2.3069e-02]]]]),\n", + " 'conv1.conv2d.bias': tensor([ 0.0196, -0.0175, -0.0570, 0.1336, -0.1144, 0.0476, 0.0037, -0.0243,\n", + " -0.1192, -0.0432, -0.0359, 0.1437, 0.1711, -0.0013, -0.0158, -0.0281,\n", + " -0.1924, 0.0254, 0.1632, -0.0235, 0.0750, 0.0365, -0.0345, -0.0149,\n", + " -0.0080, -0.0132, -0.0426, -0.0692, 0.0494, 0.0902, -0.0475, -0.0722]),\n", + " 'in1.weight': tensor([0.4995, 0.3629, 0.6906, 1.2521, 0.3878, 0.8601, 0.5297, 0.5398, 0.8082,\n", + " 0.9980, 1.3335, 1.1498, 0.7596, 0.6604, 0.4047, 0.8026, 1.0651, 0.9420,\n", + " 1.2978, 0.5745, 0.3835, 0.8996, 0.3883, 0.5270, 0.8622, 0.3668, 0.5018,\n", + " 0.6149, 0.6673, 0.4340, 0.7557, 0.7622]),\n", + " 'in1.bias': tensor([ 0.4168, 0.3790, 0.9787, 0.8629, 0.2428, 0.6218, 0.0672, 0.2934,\n", + " -0.1040, -0.4192, 2.1031, 0.7321, 0.4158, 0.9371, 0.5454, 0.0216,\n", + " 1.0306, 0.4989, 0.6166, 0.9259, -0.1587, 1.6728, 0.5660, -0.1894,\n", + " 0.4066, -0.2124, 0.7067, 1.5929, 0.4018, -0.0203, -0.2061, 0.5875]),\n", + " 'conv2.conv2d.weight': tensor([[[[ 8.3289e-02, 1.9732e-01, 2.6146e-02],\n", + " [-3.2027e-02, 8.8667e-02, -8.7891e-03],\n", + " [-3.4168e-02, -5.9419e-02, -5.4620e-02]],\n", + " \n", + " [[-2.4977e-01, -1.4473e-01, -1.8072e-02],\n", + " [-2.2474e-01, -5.7386e-02, 1.2723e-02],\n", + " [-8.2460e-02, 6.9165e-03, 1.7105e-01]],\n", + " \n", + " [[ 3.0193e-01, 2.3532e-01, 1.2811e-01],\n", + " [ 2.8777e-01, 6.6046e-02, -5.3554e-02],\n", + " [ 6.9099e-02, -4.5837e-02, -2.6823e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[-1.1161e-01, -2.4662e-02, -3.0874e-02],\n", + " [ 2.1976e-02, 1.8717e-01, 4.2034e-02],\n", + " [ 2.1250e-01, 2.7026e-01, 1.7902e-01]],\n", + " \n", + " [[ 7.2882e-03, -5.2279e-03, -1.4506e-02],\n", + " [-1.0220e-01, -4.7996e-02, 8.7613e-04],\n", + " [-8.2120e-02, 1.7910e-02, 5.2476e-02]],\n", + " \n", + " [[ 1.8565e-01, 2.8774e-01, 1.8725e-01],\n", + " [ 3.8189e-01, 3.3456e-01, 2.4582e-01],\n", + " [ 3.8858e-01, 2.2061e-01, 1.0689e-01]]],\n", + " \n", + " \n", + " [[[ 3.6264e-02, -1.1055e-01, -1.2425e-02],\n", + " [ 5.6651e-02, -2.5825e-02, -1.2968e-02],\n", + " [ 4.6819e-02, 1.5495e-02, 9.7517e-02]],\n", + " \n", + " [[-4.7270e-02, 9.8718e-02, 1.4453e-02],\n", + " [-3.9839e-02, 1.9109e-01, 5.9219e-02],\n", + " [ 1.8752e-02, 1.5244e-01, -9.5023e-02]],\n", + " \n", + " [[ 3.5655e-02, 1.0817e-01, -7.4039e-02],\n", + " [ 6.4968e-02, 5.7866e-02, -6.5176e-02],\n", + " [-2.6372e-02, -1.2702e-01, 1.8188e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[-7.9624e-02, 3.8218e-02, 5.8803e-03],\n", + " [-1.0509e-03, 4.5832e-02, -6.0746e-02],\n", + " [ 5.1501e-02, 2.8528e-02, 8.6050e-02]],\n", + " \n", + " [[ 5.9862e-02, 1.6001e-02, 7.7810e-02],\n", + " [ 7.0854e-02, -8.6467e-02, -2.5765e-03],\n", + " [ 6.4285e-02, -4.0784e-03, -1.3630e-02]],\n", + " \n", + " [[-1.0515e-01, 7.5895e-04, 2.3324e-02],\n", + " [-2.9072e-01, -9.2733e-02, 3.4276e-02],\n", + " [-8.1397e-02, -1.0381e-01, -5.9956e-02]]],\n", + " \n", + " \n", + " [[[ 2.8225e-04, 1.2499e-02, 5.6419e-02],\n", + " [ 3.5360e-02, 1.4679e-01, 1.2511e-01],\n", + " [ 5.2449e-02, 8.5608e-02, -2.1646e-02]],\n", + " \n", + " [[ 8.5807e-02, 2.9972e-02, -4.1255e-02],\n", + " [ 5.6320e-03, -1.0754e-01, -1.3424e-01],\n", + " [-5.5074e-02, -1.6911e-01, -1.4768e-01]],\n", + " \n", + " [[ 1.1310e-01, 6.4495e-02, -1.0346e-01],\n", + " [ 9.9031e-02, -6.8858e-02, -1.6426e-01],\n", + " [ 2.4112e-01, -4.2699e-03, -1.6591e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[ 1.0724e-01, 6.8815e-02, 6.5681e-02],\n", + " [ 1.0503e-01, 1.7104e-01, 1.1451e-01],\n", + " [ 8.1849e-03, 1.0336e-01, 6.4519e-02]],\n", + " \n", + " [[-1.1182e-01, -6.8019e-02, -9.2363e-02],\n", + " [-1.1122e-01, -3.5180e-02, -1.0755e-02],\n", + " [-1.0368e-01, -1.2882e-01, -9.9582e-02]],\n", + " \n", + " [[-1.8789e-02, -5.1656e-02, 9.2468e-02],\n", + " [ 3.3635e-02, 3.0774e-02, 1.2078e-01],\n", + " [-5.0636e-02, -3.4268e-02, 2.1426e-02]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[-3.2440e-02, 9.0189e-02, 1.8910e-01],\n", + " [-2.2810e-01, -5.3943e-02, 8.8429e-02],\n", + " [-2.5049e-01, -1.6005e-01, 5.3223e-03]],\n", + " \n", + " [[ 6.6471e-02, 6.9893e-02, 2.3436e-02],\n", + " [-3.0420e-02, 8.9984e-04, -1.0588e-02],\n", + " [-3.5965e-02, -4.9263e-02, -9.5038e-02]],\n", + " \n", + " [[ 1.2748e-01, 8.0859e-02, 6.7137e-02],\n", + " [-8.2055e-02, 4.2380e-02, -8.8881e-02],\n", + " [-1.9628e-02, 4.3852e-02, -4.8321e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[ 4.3108e-02, 5.0314e-02, -1.9911e-02],\n", + " [ 4.8009e-02, 3.4009e-02, 1.3368e-01],\n", + " [-1.0504e-01, -6.4687e-02, 1.1427e-01]],\n", + " \n", + " [[ 1.0406e-01, -1.2046e-02, 2.8216e-02],\n", + " [ 1.8877e-02, 3.9146e-02, 7.7465e-02],\n", + " [-2.0310e-02, 9.4038e-02, 8.0003e-03]],\n", + " \n", + " [[ 3.3473e-02, -4.1820e-02, 2.9369e-02],\n", + " [-1.5304e-02, -1.8492e-01, -9.1424e-02],\n", + " [-8.5546e-02, -2.0485e-01, -5.7351e-02]]],\n", + " \n", + " \n", + " [[[-1.8843e-01, -2.1470e-01, -2.1451e-01],\n", + " [-1.0306e-01, -7.1697e-02, -1.6913e-01],\n", + " [ 3.9632e-02, -1.4938e-02, -1.7902e-01]],\n", + " \n", + " [[-6.4272e-02, -4.6200e-02, -2.7536e-02],\n", + " [-7.3205e-02, -1.6280e-01, -2.6380e-02],\n", + " [-1.2672e-02, -8.7997e-02, 2.1879e-02]],\n", + " \n", + " [[-8.0491e-02, -4.1850e-02, -6.8172e-03],\n", + " [ 4.8230e-02, -3.7830e-02, -2.6769e-03],\n", + " [-4.0758e-02, 4.8690e-02, -7.8193e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[-1.1202e-01, -1.4725e-01, -1.1601e-01],\n", + " [-1.3324e-01, -9.9313e-02, -1.8441e-01],\n", + " [ 4.4635e-03, -8.9834e-02, -1.6402e-01]],\n", + " \n", + " [[-1.2878e-01, -1.9783e-01, -1.7645e-01],\n", + " [-1.1427e-01, -1.0230e-01, -8.8727e-02],\n", + " [-1.0716e-01, -9.0837e-02, -1.6812e-01]],\n", + " \n", + " [[ 5.4898e-02, -1.4634e-03, -2.5286e-02],\n", + " [ 1.3752e-01, 1.0764e-01, 3.4601e-02],\n", + " [ 1.9796e-01, 1.6828e-01, 6.3711e-02]]],\n", + " \n", + " \n", + " [[[ 7.9107e-02, 1.3999e-01, 1.3158e-01],\n", + " [-1.5530e-01, -1.2620e-01, 2.6067e-02],\n", + " [ 5.0762e-02, -2.7054e-02, -3.7762e-02]],\n", + " \n", + " [[ 3.8061e-02, -7.9837e-02, -4.0674e-02],\n", + " [-1.3512e-02, 7.4261e-02, 1.4128e-01],\n", + " [-1.0774e-02, 4.0948e-03, 7.0290e-02]],\n", + " \n", + " [[ 6.0168e-03, 3.3030e-03, -1.8936e-01],\n", + " [-6.2498e-02, 1.6649e-01, 2.1017e-01],\n", + " [-1.1452e-01, -1.8696e-01, 6.3119e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[-1.2977e-01, -1.0702e-01, 1.0948e-01],\n", + " [ 1.0594e-01, -4.4424e-02, -1.1309e-01],\n", + " [-7.8589e-03, 1.6506e-01, 7.0235e-02]],\n", + " \n", + " [[ 1.5526e-01, 1.3346e-02, -5.5618e-02],\n", + " [ 3.4582e-02, -7.1804e-02, -8.5391e-02],\n", + " [ 4.7709e-02, 1.7933e-02, 1.4388e-02]],\n", + " \n", + " [[-2.3048e-02, 2.7419e-02, 1.0545e-01],\n", + " [-9.0032e-02, -5.3277e-02, 2.3637e-02],\n", + " [ 1.3886e-01, 3.8664e-02, -9.4362e-02]]]]),\n", + " 'conv2.conv2d.bias': tensor([-0.0261, 0.1675, -0.0293, 0.0488, 0.1914, 0.1653, 0.0554, -0.0247,\n", + " -0.0746, 0.0757, -0.0010, 0.1209, 0.0215, -0.0522, 0.0733, -0.0224,\n", + " 0.0218, -0.0147, 0.1302, -0.0924, -0.2051, -0.0686, -0.1731, 0.1976,\n", + " -0.0727, -0.1234, -0.0381, 0.1849, 0.0284, -0.0382, -0.1318, 0.1143,\n", + " 0.0739, 0.0817, -0.0185, -0.2380, -0.2643, -0.1213, 0.0762, -0.1960,\n", + " -0.0876, 0.0194, 0.1745, -0.0534, 0.2968, 0.1412, -0.2547, 0.0145,\n", + " -0.0278, 0.1744, 0.0057, -0.0309, 0.0549, -0.0022, 0.1055, 0.1007,\n", + " -0.1640, -0.0029, 0.0510, 0.1571, 0.2350, -0.0110, -0.1390, 0.1057]),\n", + " 'in2.weight': tensor([0.8069, 1.0242, 0.7587, 0.9197, 0.8042, 1.2121, 1.5010, 0.9680, 1.1996,\n", + " 0.5729, 0.8186, 0.9971, 0.6904, 0.5832, 1.3248, 0.7074, 0.7998, 0.8463,\n", + " 0.7104, 0.8536, 1.0227, 0.7366, 0.6800, 1.1136, 0.8912, 1.4169, 0.8671,\n", + " 1.2928, 0.8099, 1.2235, 0.7416, 0.6776, 1.0358, 0.9296, 1.1402, 1.2977,\n", + " 1.0288, 0.5420, 1.1251, 0.8605, 1.0423, 1.2185, 1.0500, 0.6871, 0.8506,\n", + " 0.8225, 0.9992, 1.1387, 1.5629, 1.0439, 1.2837, 0.9237, 0.8472, 1.0977,\n", + " 0.9193, 0.8416, 1.4587, 0.7148, 0.9773, 0.6568, 0.7063, 0.9354, 0.6313,\n", + " 1.4120]),\n", + " 'in2.bias': tensor([ 1.2312e-01, 6.8294e-01, 1.5469e-01, 3.0220e-02, -1.4820e-01,\n", + " 1.2256e-01, 2.8362e-02, 1.7918e-01, 2.7628e-01, 2.8611e-02,\n", + " 2.1497e-01, -3.2504e-01, -1.2501e-01, -2.6679e-02, 8.2333e-01,\n", + " -1.3596e-01, 2.3994e-01, 1.3781e-01, -1.5688e-01, 8.9394e-02,\n", + " 1.5670e-01, 3.1545e-01, -8.3328e-02, 3.7993e-01, 2.8741e-01,\n", + " 3.9421e-01, 1.3796e-01, 2.0143e-01, 7.6481e-02, 5.8158e-01,\n", + " 1.3844e-01, 1.1570e-01, 4.3291e-01, 1.3604e-01, 3.6399e-01,\n", + " 5.3711e-01, 7.3824e-02, -7.4304e-02, 3.6802e-01, -1.2490e-01,\n", + " 4.0140e-01, 7.1305e-02, 3.7830e-01, -8.2643e-02, 2.6869e-01,\n", + " 1.9472e-01, 1.9443e-01, 2.3822e-01, 1.5232e-04, 8.9157e-01,\n", + " 4.8493e-01, 1.0025e-01, -3.5345e-02, 2.5826e-01, 2.9003e-01,\n", + " 2.5140e-01, 1.4377e-01, -2.4080e-01, -2.1667e-01, 9.4698e-02,\n", + " 5.3572e-02, 3.0392e-01, -1.1564e-01, 3.9730e-01]),\n", + " 'conv3.conv2d.weight': tensor([[[[ 0.0065, 0.0295, 0.0895],\n", + " [ 0.0977, 0.1551, 0.1230],\n", + " [ 0.1503, 0.0979, 0.0351]],\n", + " \n", + " [[ 0.0116, -0.1181, 0.0717],\n", + " [-0.0188, -0.0368, 0.1717],\n", + " [ 0.0031, -0.1084, 0.1126]],\n", + " \n", + " [[ 0.0059, -0.1087, -0.0030],\n", + " [-0.0444, -0.0603, 0.0267],\n", + " [-0.0451, -0.0906, -0.0128]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0710, 0.1786, 0.0460],\n", + " [-0.0911, 0.2001, -0.0033],\n", + " [-0.0097, 0.0136, 0.0849]],\n", + " \n", + " [[ 0.0676, 0.0343, 0.1471],\n", + " [ 0.0848, -0.0050, 0.0159],\n", + " [ 0.0594, -0.0495, -0.0190]],\n", + " \n", + " [[ 0.0366, 0.0040, -0.0519],\n", + " [ 0.0826, -0.1041, -0.0191],\n", + " [ 0.0963, -0.0296, -0.0390]]],\n", + " \n", + " \n", + " [[[ 0.0669, -0.0231, 0.0562],\n", + " [-0.1791, -0.1161, 0.0244],\n", + " [ 0.0739, 0.2753, 0.1249]],\n", + " \n", + " [[ 0.0168, 0.0123, -0.0063],\n", + " [-0.0592, -0.0695, -0.0266],\n", + " [-0.0515, 0.1139, 0.0798]],\n", + " \n", + " [[ 0.0019, -0.0376, 0.0348],\n", + " [-0.0696, 0.0117, 0.0090],\n", + " [-0.0799, -0.0724, -0.0705]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0741, 0.0128, -0.0436],\n", + " [-0.1224, -0.0920, 0.0584],\n", + " [ 0.0769, 0.1088, 0.0454]],\n", + " \n", + " [[-0.0687, -0.0190, -0.0307],\n", + " [-0.0202, 0.0981, 0.1410],\n", + " [-0.0521, 0.0380, 0.0238]],\n", + " \n", + " [[ 0.0326, 0.0581, -0.1009],\n", + " [-0.2725, -0.3432, -0.1945],\n", + " [ 0.0753, 0.2340, -0.0188]]],\n", + " \n", + " \n", + " [[[ 0.0889, 0.0085, -0.0322],\n", + " [ 0.0129, -0.0580, -0.0080],\n", + " [ 0.0123, -0.1090, -0.0094]],\n", + " \n", + " [[-0.0537, -0.0432, 0.0353],\n", + " [-0.1245, -0.1929, 0.1103],\n", + " [-0.0211, -0.0700, -0.0452]],\n", + " \n", + " [[ 0.0909, -0.0084, -0.1603],\n", + " [ 0.2445, 0.1186, -0.1600],\n", + " [ 0.1940, -0.0675, -0.3662]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.1398, 0.0800, 0.0398],\n", + " [ 0.0048, -0.0360, 0.0165],\n", + " [ 0.0763, -0.0120, 0.0711]],\n", + " \n", + " [[-0.0968, 0.0203, 0.0006],\n", + " [-0.0935, -0.0401, -0.0402],\n", + " [-0.1245, 0.0511, 0.0263]],\n", + " \n", + " [[-0.1833, -0.0760, 0.0357],\n", + " [-0.1165, -0.0960, 0.0565],\n", + " [ 0.0268, 0.0961, 0.1707]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[ 0.0669, 0.0131, -0.0975],\n", + " [ 0.1700, -0.0849, 0.0892],\n", + " [-0.0418, 0.0244, 0.0576]],\n", + " \n", + " [[ 0.0005, 0.0705, -0.1489],\n", + " [ 0.0784, 0.1286, 0.1062],\n", + " [-0.0766, -0.0636, 0.0959]],\n", + " \n", + " [[ 0.1344, -0.0146, -0.0417],\n", + " [-0.0109, -0.0013, 0.0064],\n", + " [-0.0048, 0.0017, 0.0327]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0027, -0.0737, 0.0267],\n", + " [-0.0987, -0.0079, 0.0049],\n", + " [ 0.0357, 0.2128, -0.0332]],\n", + " \n", + " [[-0.1447, -0.1033, -0.0106],\n", + " [-0.0520, 0.0414, 0.0072],\n", + " [-0.0872, -0.0568, 0.0104]],\n", + " \n", + " [[ 0.0535, -0.1089, -0.0724],\n", + " [-0.1485, -0.1843, -0.0455],\n", + " [-0.0754, 0.1179, -0.0299]]],\n", + " \n", + " \n", + " [[[-0.0233, -0.1136, -0.0832],\n", + " [-0.0962, -0.1792, -0.1441],\n", + " [-0.0860, -0.1861, -0.1270]],\n", + " \n", + " [[-0.0637, -0.0127, -0.1039],\n", + " [-0.0312, -0.1677, -0.1083],\n", + " [ 0.0298, 0.0469, -0.1258]],\n", + " \n", + " [[ 0.1464, 0.0690, 0.0281],\n", + " [ 0.0888, -0.0870, -0.0657],\n", + " [ 0.0290, -0.0443, -0.0341]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0230, -0.1083, -0.0695],\n", + " [-0.0772, -0.1686, 0.0713],\n", + " [-0.0166, -0.0537, 0.0727]],\n", + " \n", + " [[-0.0343, -0.0502, 0.0112],\n", + " [-0.0262, 0.0472, 0.0591],\n", + " [-0.0281, 0.0601, 0.0461]],\n", + " \n", + " [[ 0.0751, 0.0210, -0.0570],\n", + " [ 0.0783, -0.0623, -0.0383],\n", + " [-0.1386, 0.0648, 0.1995]]],\n", + " \n", + " \n", + " [[[ 0.0404, -0.0544, -0.0014],\n", + " [ 0.0713, -0.0235, -0.0466],\n", + " [ 0.0517, 0.0379, 0.0096]],\n", + " \n", + " [[-0.0240, -0.0074, 0.0085],\n", + " [ 0.0846, 0.0197, -0.0013],\n", + " [ 0.0437, -0.0303, -0.0422]],\n", + " \n", + " [[ 0.0260, -0.0274, -0.0339],\n", + " [ 0.0645, 0.0380, 0.0707],\n", + " [-0.0187, -0.0226, -0.0336]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0615, -0.0981, -0.0409],\n", + " [ 0.2038, 0.2010, 0.1175],\n", + " [-0.0496, -0.1081, -0.0626]],\n", + " \n", + " [[ 0.0608, 0.0595, 0.0382],\n", + " [-0.0281, -0.0669, -0.0116],\n", + " [-0.1454, -0.0308, 0.0384]],\n", + " \n", + " [[-0.2299, -0.1853, -0.2017],\n", + " [ 0.3882, 0.3230, -0.0039],\n", + " [-0.0900, -0.1035, -0.0563]]]]),\n", + " 'conv3.conv2d.bias': tensor([ 1.0194e-01, 2.4678e-02, 2.8636e-02, -1.8480e-03, -1.2408e-01,\n", + " 2.1479e-04, 6.3129e-02, 4.4626e-02, -6.1239e-02, 4.3553e-02,\n", + " 2.3288e-02, -6.4785e-02, -3.0239e-02, -1.7113e-01, -2.4260e-01,\n", + " 1.2283e-01, 2.4282e-02, 4.9491e-02, 1.7601e-01, 7.9365e-02,\n", + " -1.0822e-01, -8.9898e-02, 6.0185e-02, 1.2195e-01, -1.2970e-01,\n", + " -1.2266e-01, 1.6213e-01, -8.2205e-02, 1.0288e-01, -1.3014e-01,\n", + " -2.5489e-02, -7.3829e-02, -2.2353e-02, -4.5089e-02, 1.6561e-01,\n", + " -7.2363e-02, -2.3814e-02, 2.4523e-01, -8.3162e-02, 8.0363e-02,\n", + " 1.0618e-01, 1.2792e-01, 1.6342e-01, -1.7300e-01, -5.1166e-02,\n", + " -4.2894e-02, 8.7484e-02, -1.1153e-01, -9.6206e-02, 1.3092e-02,\n", + " -3.6404e-02, -6.2190e-03, -9.7031e-02, 6.4421e-02, -5.8877e-02,\n", + " 2.6036e-02, 1.4222e-01, 3.3598e-02, -6.1637e-02, -3.1119e-02,\n", + " 3.5748e-02, -4.5988e-02, -1.2942e-02, 1.3717e-01, 1.6123e-02,\n", + " -1.7269e-02, -4.3437e-03, -5.1885e-02, 4.9598e-02, 1.1033e-01,\n", + " 6.6223e-02, -6.8821e-02, -1.6470e-01, 1.1226e-02, -4.6222e-02,\n", + " 2.1903e-01, 1.6593e-01, -2.3289e-02, 1.6111e-02, -2.1037e-01,\n", + " -1.8199e-02, -3.1360e-02, -6.6394e-02, -1.7776e-01, -4.8337e-02,\n", + " -1.9718e-01, -1.0013e-02, -4.7527e-03, -6.0905e-03, 1.0659e-01,\n", + " 5.6494e-02, -2.3239e-02, 8.1521e-02, 5.2152e-02, 1.2628e-01,\n", + " -2.5421e-03, 1.6061e-01, -1.0730e-01, 1.4904e-01, 9.6609e-03,\n", + " -1.2296e-01, -1.7489e-02, -2.2044e-01, 7.7403e-02, 5.3610e-02,\n", + " -8.2700e-02, 4.5502e-02, 1.3002e-01, -7.3065e-02, -2.2319e-01,\n", + " 2.5133e-02, 3.5435e-02, -1.9936e-01, 1.9520e-02, -7.4913e-02,\n", + " -1.6374e-02, 1.5933e-02, 1.0952e-01, 1.0475e-02, -1.0687e-01,\n", + " -3.5883e-01, -1.1257e-01, 1.2576e-01, 9.7992e-02, 4.9123e-02,\n", + " -1.9408e-02, 1.0979e-01, 7.8486e-02]),\n", + " 'in3.weight': tensor([1.0638, 1.2093, 1.0276, 0.6075, 0.9626, 0.7194, 0.8470, 0.8174, 1.0952,\n", + " 0.7243, 1.2864, 0.6957, 0.8937, 1.0512, 1.0178, 1.0778, 0.9223, 0.9274,\n", + " 0.7749, 1.1933, 0.9609, 0.8226, 0.9711, 0.8441, 1.2350, 1.0388, 1.1288,\n", + " 1.1114, 0.8308, 0.9122, 1.0150, 1.1359, 1.2098, 1.1096, 1.2799, 1.0696,\n", + " 0.9589, 0.9819, 0.9552, 0.9021, 1.2858, 0.9997, 1.4447, 1.0806, 0.9283,\n", + " 1.0101, 0.8484, 1.0967, 0.9334, 0.9426, 0.8658, 1.3449, 0.9794, 1.0873,\n", + " 0.7406, 1.0302, 1.0916, 0.8231, 0.8961, 1.0580, 0.8950, 1.3716, 0.9151,\n", + " 1.0519, 1.4394, 1.4507, 1.0435, 0.7959, 1.0081, 0.8717, 0.8903, 1.1371,\n", + " 0.8810, 0.7283, 0.7897, 0.9319, 1.2388, 1.4531, 0.8062, 0.7954, 1.3265,\n", + " 1.0002, 0.9771, 0.8423, 1.3119, 1.2154, 1.0728, 0.8937, 0.8408, 1.0255,\n", + " 1.0621, 0.8516, 1.5538, 0.7643, 1.3574, 0.9643, 0.7985, 1.0198, 1.1696,\n", + " 0.8279, 0.8969, 1.3378, 0.9452, 0.8968, 1.1282, 1.0439, 0.8715, 1.1519,\n", + " 0.9837, 1.4297, 0.9461, 1.3048, 1.5128, 0.9250, 1.0811, 1.1009, 1.0091,\n", + " 1.1366, 0.9738, 1.4749, 1.3707, 0.9805, 0.8876, 0.8156, 0.8087, 1.0348,\n", + " 0.9029, 1.4788]),\n", + " 'in3.bias': tensor([ 1.1070e-01, 3.2944e-01, 1.9625e-01, 4.7169e-03, 5.7622e-02,\n", + " 2.2786e-01, 3.3572e-01, 2.3915e-01, -9.6800e-02, 7.1040e-01,\n", + " 1.7710e-01, 2.1755e-01, 2.6478e-01, 2.6071e-01, 1.4037e-01,\n", + " 4.6066e-01, 3.2928e-01, 2.3550e-01, 6.5005e-01, 1.6770e-01,\n", + " 2.5070e-01, 6.6386e-02, 2.0195e-01, 3.2442e-01, -1.7003e-02,\n", + " 3.3309e-01, 2.5729e-01, 3.8876e-01, 5.5629e-02, 2.4434e-01,\n", + " 6.5466e-01, 5.7230e-01, 1.6685e-01, 3.5245e-01, 2.9543e-01,\n", + " 2.9605e-01, 7.3126e-01, 1.5857e-01, 5.8489e-01, 2.0405e-01,\n", + " 4.4950e-01, 3.9461e-01, 3.6418e-01, 2.8988e-02, 2.2680e-01,\n", + " 7.6742e-01, 3.1984e-01, 1.1298e-01, 3.0882e-01, -5.4089e-02,\n", + " -1.7424e-01, 4.0441e-01, 2.9965e-01, 2.4304e-01, 1.8582e-01,\n", + " 1.0870e-02, 4.0504e-01, 7.7639e-02, -1.2010e-01, 1.2394e-01,\n", + " 2.4855e-01, 2.0278e-01, 8.3190e-02, -6.0753e-04, 1.3995e-01,\n", + " -1.0276e-01, 2.2686e-01, -1.8294e-01, 2.6676e-01, 1.8317e-01,\n", + " 5.8263e-01, -1.8501e-02, 1.9452e-01, 1.1396e-01, 2.1548e-01,\n", + " 1.5246e-01, 3.4535e-01, 2.4284e-01, 2.1748e-01, -1.7830e-01,\n", + " 1.7922e-01, 1.4609e-01, 2.0395e-01, 4.3753e-02, 4.0286e-01,\n", + " 4.7855e-01, -9.4514e-02, 4.1779e-02, 3.9728e-01, 5.6558e-01,\n", + " 1.3138e-01, 6.5901e-02, 1.4245e-01, -6.8797e-03, 1.1948e-01,\n", + " 1.8214e-01, -1.5845e-02, 2.3576e-01, 2.1236e-01, 3.2610e-01,\n", + " 3.6187e-01, 2.2094e-01, -9.1965e-02, 2.2497e-01, 3.3252e-01,\n", + " 3.2433e-01, 4.9709e-01, 3.6117e-01, 1.2742e-01, 1.7441e-01,\n", + " 8.6633e-02, 1.7696e-01, 1.5004e-01, 1.5824e-01, 5.6848e-01,\n", + " 4.9741e-01, 2.8188e-01, 1.0514e-01, 2.6674e-01, 1.2337e-01,\n", + " 2.1926e-02, 1.0644e-01, 7.9277e-02, 1.9199e-02, 3.7754e-01,\n", + " 3.8980e-01, 5.5823e-02, 1.6802e-01]),\n", + " 'res1.conv1.conv2d.weight': tensor([[[[ 0.1120, -0.0984, 0.0847],\n", + " [ 0.0395, -0.1056, 0.0697],\n", + " [-0.0061, 0.1049, -0.0531]],\n", + " \n", + " [[-0.0175, -0.0877, 0.0099],\n", + " [-0.0987, -0.1614, -0.0234],\n", + " [ 0.0964, -0.0622, 0.0026]],\n", + " \n", + " [[-0.0113, 0.1100, 0.0302],\n", + " [-0.0913, 0.0040, 0.0245],\n", + " [-0.0829, -0.0567, 0.0558]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0201, 0.1599, -0.0793],\n", + " [-0.1523, 0.1529, -0.1180],\n", + " [ 0.0782, -0.0352, -0.0981]],\n", + " \n", + " [[-0.1783, 0.3062, -0.1296],\n", + " [-0.3554, 0.4626, -0.0275],\n", + " [-0.2440, 0.4414, 0.1361]],\n", + " \n", + " [[ 0.0564, -0.0221, 0.1083],\n", + " [ 0.0614, -0.1423, -0.0181],\n", + " [ 0.0080, -0.0103, 0.0689]]],\n", + " \n", + " \n", + " [[[ 0.0703, 0.0929, 0.1898],\n", + " [ 0.1196, 0.1768, 0.0967],\n", + " [ 0.1895, 0.1836, 0.1558]],\n", + " \n", + " [[ 0.0007, 0.0186, 0.0305],\n", + " [-0.0657, 0.0166, -0.2009],\n", + " [-0.0201, -0.0529, -0.0290]],\n", + " \n", + " [[-0.0275, 0.0857, -0.0762],\n", + " [-0.0242, 0.0400, -0.0256],\n", + " [-0.0643, -0.1127, -0.0842]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0564, -0.0724, -0.0041],\n", + " [-0.1102, -0.0305, -0.1256],\n", + " [-0.0731, -0.0011, -0.0069]],\n", + " \n", + " [[ 0.1364, 0.0393, 0.1587],\n", + " [ 0.0901, 0.0535, 0.0945],\n", + " [ 0.0994, 0.0884, 0.0770]],\n", + " \n", + " [[-0.2427, -0.0482, -0.1219],\n", + " [-0.0195, -0.2100, -0.0440],\n", + " [-0.0045, 0.0827, -0.1124]]],\n", + " \n", + " \n", + " [[[ 0.0054, 0.1761, 0.0979],\n", + " [ 0.0358, -0.0073, 0.0386],\n", + " [ 0.0802, 0.0211, 0.0067]],\n", + " \n", + " [[ 0.0676, -0.0314, -0.2192],\n", + " [-0.2038, -0.0727, 0.1859],\n", + " [-0.0823, -0.0603, -0.0220]],\n", + " \n", + " [[ 0.0700, 0.2304, 0.0584],\n", + " [ 0.1480, -0.0234, -0.1014],\n", + " [-0.0480, -0.1668, 0.0040]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0391, -0.1311, 0.0124],\n", + " [-0.0076, 0.0498, -0.0266],\n", + " [-0.0158, -0.0271, 0.0237]],\n", + " \n", + " [[ 0.0620, 0.1415, 0.0456],\n", + " [ 0.2232, 0.0121, -0.3605],\n", + " [-0.1281, -0.2056, 0.1354]],\n", + " \n", + " [[ 0.0346, 0.0820, 0.0814],\n", + " [-0.0109, -0.0891, -0.0300],\n", + " [ 0.0521, -0.0999, -0.0975]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[ 0.1041, -0.1487, -0.2806],\n", + " [-0.0914, -0.3610, -0.2262],\n", + " [ 0.0957, 0.0235, -0.0301]],\n", + " \n", + " [[-0.0439, -0.0302, 0.0205],\n", + " [ 0.0277, 0.0442, -0.0412],\n", + " [ 0.0412, 0.0243, 0.0420]],\n", + " \n", + " [[ 0.1123, -0.1125, -0.0961],\n", + " [-0.0202, -0.0302, 0.0222],\n", + " [ 0.1103, 0.0045, -0.0435]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.2750, -0.0281, -0.1756],\n", + " [-0.3419, -0.3859, -0.1775],\n", + " [-0.3091, -0.3048, 0.0569]],\n", + " \n", + " [[ 0.0611, 0.0691, 0.1565],\n", + " [ 0.1749, 0.2114, 0.1577],\n", + " [ 0.0423, 0.2063, 0.0843]],\n", + " \n", + " [[-0.0157, 0.0765, 0.1368],\n", + " [ 0.0496, 0.0413, 0.1492],\n", + " [-0.0113, 0.1057, 0.1156]]],\n", + " \n", + " \n", + " [[[ 0.0395, -0.1733, 0.0706],\n", + " [-0.1393, -0.0780, 0.0152],\n", + " [ 0.0995, 0.0589, -0.0831]],\n", + " \n", + " [[ 0.0387, 0.0036, -0.0657],\n", + " [-0.0046, -0.1426, -0.1132],\n", + " [-0.0970, -0.1085, 0.0104]],\n", + " \n", + " [[-0.1585, 0.2919, 0.0764],\n", + " [ 0.2004, 0.2146, 0.0367],\n", + " [ 0.3098, -0.0262, -0.0767]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0617, 0.1598, -0.1706],\n", + " [ 0.2184, 0.0182, -0.0951],\n", + " [ 0.0050, -0.1419, -0.1659]],\n", + " \n", + " [[ 0.1117, 0.0080, 0.0703],\n", + " [ 0.0118, 0.0113, 0.0524],\n", + " [ 0.0508, -0.0284, 0.0106]],\n", + " \n", + " [[ 0.0327, -0.0381, -0.0587],\n", + " [-0.0290, -0.2076, -0.0071],\n", + " [-0.0348, 0.0253, -0.0154]]],\n", + " \n", + " \n", + " [[[ 0.0386, -0.0610, -0.0066],\n", + " [ 0.0733, -0.1009, 0.0155],\n", + " [-0.0115, 0.0064, -0.1212]],\n", + " \n", + " [[-0.1362, -0.0458, -0.0480],\n", + " [ 0.0013, -0.3571, 0.1420],\n", + " [ 0.2611, 0.0110, -0.5974]],\n", + " \n", + " [[ 0.0750, -0.0479, -0.0138],\n", + " [ 0.0143, 0.0338, 0.1623],\n", + " [-0.0623, -0.0314, -0.1275]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0512, 0.1233, -0.0056],\n", + " [-0.0717, -0.2044, -0.1419],\n", + " [ 0.0146, 0.2483, -0.1089]],\n", + " \n", + " [[ 0.0056, -0.0345, 0.0671],\n", + " [ 0.1126, -0.0017, -0.1425],\n", + " [-0.1701, -0.0968, -0.1099]],\n", + " \n", + " [[ 0.0936, 0.0416, -0.0364],\n", + " [ 0.1436, 0.0505, -0.3638],\n", + " [-0.2301, 0.4236, -0.0659]]]]),\n", + " 'res1.conv1.conv2d.bias': tensor([-1.1022e-01, 3.9905e-02, 1.9646e-02, 9.6041e-03, 7.4591e-02,\n", + " 1.5580e-03, -2.8510e-02, 1.5249e-02, 1.4569e-01, 2.6159e-02,\n", + " 1.5171e-01, 2.4091e-01, 1.5140e-01, 6.1415e-02, -8.8541e-02,\n", + " -6.1408e-02, 2.4682e-01, -1.6808e-02, 1.5542e-01, 1.0887e-01,\n", + " -1.4346e-01, 1.4230e-01, -1.3213e-01, 1.0459e-01, -1.5981e-01,\n", + " -7.6402e-02, 3.7662e-02, 1.3492e-01, 1.6912e-01, -2.5681e-01,\n", + " -1.6195e-02, 9.6887e-02, 1.2173e-01, -1.2395e-01, 4.1509e-02,\n", + " -2.7307e-03, -4.7257e-02, -9.8080e-02, 7.1222e-02, 1.4283e-01,\n", + " 1.8834e-02, 1.7753e-01, -1.3004e-01, -5.7268e-02, 4.5435e-03,\n", + " -1.3765e-01, -2.0367e-02, -7.2077e-02, -1.2572e-02, 1.9010e-01,\n", + " -1.2987e-01, 1.8363e-01, -7.1875e-02, -5.9557e-02, 7.2809e-02,\n", + " -1.4723e-02, 9.7119e-03, 6.2526e-02, 1.5395e-01, 1.2130e-01,\n", + " -1.0846e-01, -1.7782e-01, 2.9210e-02, -1.2528e-01, 2.3288e-02,\n", + " 5.6239e-02, 4.5956e-02, -4.2603e-02, -5.7570e-02, 2.9494e-02,\n", + " -1.1981e-01, 2.8704e-02, 1.7110e-01, -4.7822e-02, 6.3252e-02,\n", + " 8.0041e-02, -8.5468e-02, -2.1819e-01, -7.6494e-03, -3.4969e-02,\n", + " -3.4313e-02, 1.5787e-01, -1.5935e-01, -7.5892e-02, -1.7306e-01,\n", + " -3.6545e-02, 1.7451e-02, 3.1512e-02, 7.3312e-02, -7.1627e-03,\n", + " -1.4129e-01, 6.8708e-02, 1.2526e-01, 1.0508e-01, -9.6351e-02,\n", + " -1.7497e-01, -3.3624e-02, 1.2878e-02, -3.7867e-02, -1.3377e-01,\n", + " -5.5113e-02, -2.0429e-01, 2.0450e-02, 1.4793e-01, -5.1145e-02,\n", + " 4.1119e-03, 5.8200e-02, -4.8449e-02, -7.7571e-02, -1.8303e-01,\n", + " -6.2679e-02, -1.1783e-01, -7.1455e-02, -8.7714e-02, 5.1434e-02,\n", + " -2.8280e-01, -2.3587e-01, 3.8612e-02, -2.2062e-01, 1.2656e-01,\n", + " 6.6284e-02, -2.3327e-01, 1.4856e-01, 1.4475e-02, 2.7585e-01,\n", + " -1.0569e-01, -1.3348e-04, -1.6936e-01]),\n", + " 'res1.in1.weight': tensor([1.0561, 0.7881, 1.1374, 0.9538, 1.0175, 0.9538, 0.7949, 1.1095, 1.2229,\n", + " 1.1235, 0.9759, 1.1261, 0.8108, 0.9313, 0.7781, 0.9358, 1.1856, 0.9977,\n", + " 1.2003, 0.6331, 0.9553, 0.7246, 0.7369, 0.9656, 0.9409, 0.9469, 1.0002,\n", + " 1.0332, 0.8228, 1.1498, 1.0717, 0.8926, 0.9173, 0.8038, 1.0633, 0.7198,\n", + " 1.0547, 0.8543, 1.0321, 0.9001, 1.0226, 0.7245, 0.9532, 1.0448, 0.9431,\n", + " 0.7035, 0.9534, 1.1520, 0.9706, 0.9458, 1.1118, 0.9669, 0.9284, 0.9796,\n", + " 1.1692, 0.9596, 0.9503, 1.2410, 0.7813, 1.1764, 0.9822, 0.8084, 0.8796,\n", + " 0.8814, 1.1118, 0.9342, 0.8043, 0.9742, 0.9707, 1.0005, 1.0413, 1.0152,\n", + " 1.0736, 0.9139, 0.6112, 0.9289, 0.9357, 0.7656, 0.9988, 0.9522, 0.9991,\n", + " 0.7112, 1.3297, 1.1827, 1.1581, 0.9904, 0.9388, 1.0434, 0.8123, 1.0315,\n", + " 1.0311, 0.9759, 0.8541, 0.9737, 0.9719, 0.9814, 1.2327, 0.8521, 0.9720,\n", + " 0.8582, 1.0341, 0.7253, 1.0328, 0.8057, 0.6920, 0.6738, 0.9104, 0.9260,\n", + " 0.9714, 0.7912, 0.9056, 0.8037, 1.2186, 1.1166, 0.7973, 1.2299, 0.7662,\n", + " 0.7806, 1.1180, 1.0190, 0.7497, 1.0448, 0.8009, 1.1551, 1.0183, 1.0733,\n", + " 0.9745, 1.1386]),\n", + " 'res1.in1.bias': tensor([ 0.2431, -0.1877, 0.1873, -0.3951, -0.0309, 0.0685, -0.3976, 0.0852,\n", + " -0.2581, 0.1720, -0.6146, -0.2124, -0.4155, 0.1279, -0.2474, -0.0261,\n", + " -0.2588, 0.0419, -0.0901, -0.1367, -0.0826, -0.2342, -0.0959, 0.0906,\n", + " 0.1294, -0.1577, 0.0186, -0.1891, -0.3339, -0.2242, 0.2604, -0.3534,\n", + " 0.0216, -0.2803, 0.2391, -0.1012, 0.0197, -0.2841, -0.4157, 0.0408,\n", + " 0.0040, -0.1910, 0.0278, -0.3898, 0.0990, -0.0785, -0.1970, -0.3084,\n", + " -0.2259, 0.0069, -0.0109, 0.3114, 0.0890, 0.0761, -0.1211, -0.2096,\n", + " -0.1407, -0.2188, -0.1023, 0.1775, -0.0103, -0.0725, -0.1234, -0.4286,\n", + " 0.4395, -0.1784, 0.3911, -0.0937, 0.0141, 0.0393, -0.1475, 0.1308,\n", + " -0.0899, -0.6771, -0.3104, -0.1077, -0.3089, -0.1907, 0.2652, 0.3980,\n", + " -0.2152, -0.1792, 0.0882, -0.0072, -0.4003, 0.0627, -0.0922, -0.1443,\n", + " -0.2630, -0.0874, -0.0261, 0.3473, -0.4211, 0.1422, 0.0690, 0.0542,\n", + " -0.1390, -0.3417, 0.0146, -0.3249, 0.0220, -0.1826, -0.0184, -0.2948,\n", + " -0.2126, -0.2331, -0.2078, -0.2089, 0.0729, 0.1254, -0.0425, -0.4911,\n", + " 0.3935, -0.1087, -0.0874, -0.1579, 0.0859, -0.1292, -0.2858, -0.0659,\n", + " -0.2326, -0.1343, 0.1033, -0.1074, -0.0415, 0.0441, -0.3123, 0.4528]),\n", + " 'res1.conv2.conv2d.weight': tensor([[[[ 0.0373, 0.1138, 0.0170],\n", + " [ 0.1553, 0.1189, -0.0701],\n", + " [ 0.1935, 0.2194, -0.0528]],\n", + " \n", + " [[-0.1034, -0.1270, -0.2139],\n", + " [-0.0070, -0.1532, -0.3481],\n", + " [ 0.0767, 0.0063, 0.0043]],\n", + " \n", + " [[-0.0775, 0.0483, -0.0259],\n", + " [ 0.0516, 0.0134, 0.2092],\n", + " [-0.0529, 0.0098, 0.0300]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0612, -0.1238, 0.1401],\n", + " [-0.0941, 0.0143, -0.0680],\n", + " [ 0.1504, 0.1117, 0.0210]],\n", + " \n", + " [[-0.0389, -0.2248, 0.1036],\n", + " [-0.1675, -0.0390, -0.0238],\n", + " [-0.1530, -0.0760, -0.0999]],\n", + " \n", + " [[ 0.1675, -0.0743, -0.0983],\n", + " [-0.0921, 0.3702, 0.1004],\n", + " [-0.0505, -0.0814, -0.0464]]],\n", + " \n", + " \n", + " [[[-0.0354, 0.1882, 0.0741],\n", + " [ 0.0182, -0.0922, 0.0794],\n", + " [-0.0823, -0.0161, 0.1401]],\n", + " \n", + " [[ 0.1353, 0.0764, 0.1218],\n", + " [ 0.0682, 0.0916, 0.1373],\n", + " [ 0.0699, 0.1323, 0.0575]],\n", + " \n", + " [[-0.0740, 0.1302, -0.0284],\n", + " [ 0.0946, -0.1458, 0.0299],\n", + " [-0.0084, 0.3100, 0.2348]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0588, 0.0208, -0.0234],\n", + " [ 0.1276, 0.0326, -0.0866],\n", + " [-0.0359, 0.1948, 0.0114]],\n", + " \n", + " [[-0.0201, -0.1044, -0.1841],\n", + " [-0.0698, -0.1683, 0.1310],\n", + " [-0.0236, 0.0851, 0.1520]],\n", + " \n", + " [[-0.2872, 0.0691, 0.1984],\n", + " [ 0.3271, -0.2450, 0.2467],\n", + " [ 0.0103, 0.1642, 0.2307]]],\n", + " \n", + " \n", + " [[[ 0.0672, -0.0465, 0.0247],\n", + " [ 0.0218, 0.0240, -0.0622],\n", + " [-0.0308, -0.0801, -0.0206]],\n", + " \n", + " [[-0.0288, 0.0847, 0.0941],\n", + " [ 0.0249, 0.0709, 0.1154],\n", + " [-0.0088, 0.1338, 0.0883]],\n", + " \n", + " [[-0.0886, -0.0145, -0.1658],\n", + " [-0.0290, 0.0530, -0.0845],\n", + " [ 0.0264, -0.0419, -0.0225]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.1078, -0.0739, 0.2778],\n", + " [-0.1079, -0.0062, 0.0366],\n", + " [ 0.0416, 0.0196, 0.1248]],\n", + " \n", + " [[-0.1951, -0.0515, -0.0335],\n", + " [-0.0789, -0.0626, -0.2516],\n", + " [-0.1382, -0.1432, -0.2495]],\n", + " \n", + " [[-0.0538, 0.0414, -0.2195],\n", + " [-0.1534, -0.0128, -0.0779],\n", + " [-0.0862, 0.0172, 0.0827]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[-0.1002, -0.0119, -0.0188],\n", + " [-0.0648, -0.1232, -0.0857],\n", + " [-0.0918, -0.0807, -0.1050]],\n", + " \n", + " [[-0.0720, -0.1394, -0.1030],\n", + " [-0.2684, -0.2801, -0.1298],\n", + " [-0.2025, -0.3432, -0.1093]],\n", + " \n", + " [[-0.1037, 0.0311, 0.0171],\n", + " [ 0.0071, 0.0084, 0.0897],\n", + " [ 0.0727, 0.2265, 0.1088]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0673, 0.0432, -0.0507],\n", + " [-0.0201, -0.0507, -0.1438],\n", + " [-0.1547, -0.1464, 0.0873]],\n", + " \n", + " [[ 0.0595, 0.0807, 0.0928],\n", + " [ 0.0763, 0.1298, 0.2425],\n", + " [ 0.0585, 0.2827, 0.0775]],\n", + " \n", + " [[-0.0839, -0.0134, -0.0389],\n", + " [-0.0496, -0.0006, 0.1301],\n", + " [ 0.0441, -0.0486, 0.0232]]],\n", + " \n", + " \n", + " [[[-0.1737, -0.1219, 0.0326],\n", + " [-0.0750, -0.0267, -0.0228],\n", + " [ 0.0491, -0.1367, -0.0514]],\n", + " \n", + " [[ 0.0136, 0.1246, 0.1323],\n", + " [ 0.0112, -0.1015, 0.0476],\n", + " [ 0.0431, 0.0190, 0.1291]],\n", + " \n", + " [[ 0.0497, 0.0684, 0.0290],\n", + " [ 0.0689, 0.1283, -0.0615],\n", + " [ 0.0324, 0.0284, -0.0694]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0076, -0.2188, 0.0857],\n", + " [ 0.0163, -0.1977, -0.1611],\n", + " [ 0.0179, -0.0168, -0.0247]],\n", + " \n", + " [[-0.0721, 0.1794, 0.0929],\n", + " [-0.1060, -0.0073, -0.0454],\n", + " [-0.1913, -0.1406, -0.0750]],\n", + " \n", + " [[-0.0471, 0.2459, -0.1004],\n", + " [ 0.2487, -0.0285, -0.0766],\n", + " [ 0.0434, 0.0816, -0.0676]]],\n", + " \n", + " \n", + " [[[ 0.0166, 0.0388, 0.0043],\n", + " [-0.2078, 0.1775, 0.0195],\n", + " [-0.0948, -0.0045, 0.0198]],\n", + " \n", + " [[-0.0603, 0.0789, -0.0236],\n", + " [-0.0698, -0.1798, 0.0686],\n", + " [-0.1340, -0.0007, 0.1219]],\n", + " \n", + " [[ 0.1678, 0.0808, 0.1010],\n", + " [ 0.1279, 0.1048, -0.0244],\n", + " [-0.0699, -0.1120, -0.1513]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0569, 0.1164, 0.0369],\n", + " [-0.0645, 0.1251, 0.1737],\n", + " [-0.1048, 0.0259, -0.1260]],\n", + " \n", + " [[-0.0132, 0.0113, -0.1149],\n", + " [-0.0537, -0.1806, -0.1119],\n", + " [-0.2694, -0.0556, 0.0344]],\n", + " \n", + " [[-0.1134, 0.2136, -0.0380],\n", + " [-0.0492, -0.2865, 0.0285],\n", + " [ 0.1754, -0.0220, 0.0146]]]]),\n", + " 'res1.conv2.conv2d.bias': tensor([ 0.0386, -0.1940, -0.0714, -0.0592, 0.1086, -0.0026, 0.0709, 0.2885,\n", + " -0.0556, 0.3015, 0.1033, 0.2170, -0.0856, -0.0612, -0.1550, -0.1855,\n", + " 0.0412, -0.1295, 0.1302, 0.0430, -0.0547, -0.0907, -0.1458, -0.0465,\n", + " -0.0970, 0.1747, -0.1582, -0.0515, -0.0891, -0.0695, 0.0449, 0.0844,\n", + " 0.1629, 0.1932, -0.1043, 0.2476, 0.1696, 0.0936, -0.1866, 0.0080,\n", + " 0.0472, 0.0161, 0.1984, -0.0539, -0.0115, -0.0442, 0.0324, 0.0745,\n", + " -0.0017, 0.0131, 0.1116, -0.1759, 0.0207, -0.0407, 0.0712, -0.0323,\n", + " -0.1189, 0.0951, 0.0890, 0.0539, -0.0176, -0.1185, 0.0528, 0.1540,\n", + " 0.0160, 0.1856, 0.0095, 0.1386, 0.0559, 0.0178, -0.1843, 0.0048,\n", + " -0.0624, 0.0719, 0.0746, -0.1620, 0.1153, -0.2999, -0.1770, -0.1064,\n", + " -0.0538, -0.0188, -0.1334, 0.1673, 0.1394, -0.0715, -0.0474, -0.1252,\n", + " 0.1255, 0.0967, 0.0819, -0.1097, 0.0077, 0.0862, -0.0908, -0.2212,\n", + " 0.1313, 0.1631, 0.0910, -0.0181, -0.0511, -0.2744, 0.0556, -0.0682,\n", + " 0.0512, 0.0632, -0.2078, -0.0724, -0.1075, -0.2103, 0.0112, 0.0643,\n", + " -0.0599, 0.2203, 0.1830, 0.0563, -0.1384, 0.1838, 0.1736, -0.1409,\n", + " 0.0100, 0.2769, -0.0041, -0.0377, 0.0595, 0.0736, -0.0153, -0.0323]),\n", + " 'res1.in2.weight': tensor([0.8126, 0.8904, 0.8808, 0.4945, 0.9212, 0.9114, 1.0023, 0.8585, 0.9281,\n", + " 0.9175, 0.8369, 0.9104, 0.9056, 0.9935, 1.0445, 0.8266, 0.7800, 0.9061,\n", + " 0.7468, 0.7898, 0.8276, 0.8135, 0.9451, 0.8540, 0.7828, 0.9528, 0.7693,\n", + " 0.8253, 1.0056, 0.7976, 1.0205, 0.8087, 0.7033, 0.9079, 0.6436, 0.7293,\n", + " 0.8358, 0.9071, 0.9829, 0.7950, 0.6460, 1.1217, 0.9361, 0.7470, 0.9678,\n", + " 0.9226, 0.6795, 0.6859, 0.8737, 0.9604, 0.8574, 0.7694, 0.9325, 0.8350,\n", + " 0.9294, 0.9899, 0.9922, 0.9176, 0.8130, 0.8377, 0.9009, 0.8360, 0.8631,\n", + " 0.7176, 0.8313, 0.8855, 0.8331, 0.9426, 1.1242, 1.0314, 0.8276, 0.8502,\n", + " 0.9909, 1.2316, 0.7600, 1.1269, 1.0396, 0.8352, 1.0468, 0.7186, 0.8852,\n", + " 0.8866, 0.7449, 0.8195, 0.7804, 0.8930, 0.7691, 1.3903, 0.9619, 0.8094,\n", + " 0.8923, 1.0844, 0.9585, 0.9477, 0.8004, 0.8446, 0.7509, 0.7834, 0.9218,\n", + " 0.7838, 0.7950, 0.9362, 0.8860, 0.7259, 0.7596, 0.8580, 0.9393, 1.0505,\n", + " 0.8559, 0.6719, 1.0402, 1.0201, 0.7183, 0.6809, 0.9771, 0.7743, 0.8862,\n", + " 0.7144, 0.8715, 0.7842, 0.7209, 0.7053, 1.0821, 0.8535, 0.8079, 0.7587,\n", + " 1.1839, 0.7743]),\n", + " 'res1.in2.bias': tensor([-0.1979, 0.0532, -0.2482, 0.0437, 0.0375, -0.0730, 0.0922, 0.1362,\n", + " 0.2180, -0.1460, 0.1703, -0.0505, 0.0493, 0.0287, 0.3272, 0.0120,\n", + " -0.2833, 0.1173, 0.0009, 0.1265, 0.0791, 0.0624, -0.2618, 0.3602,\n", + " 0.0326, 0.0869, 0.1114, -0.1794, -0.1118, 0.0944, 0.0996, 0.0574,\n", + " -0.1372, 0.1664, 0.0715, 0.0816, 0.0263, -0.0295, -0.0112, 0.0562,\n", + " -0.1075, -0.0439, -0.1017, -0.2687, -0.0747, 0.0830, 0.1411, -0.1951,\n", + " 0.1770, 0.2004, -0.0231, 0.1336, -0.0333, -0.1685, -0.1892, -0.0606,\n", + " -0.0935, -0.0918, -0.0661, -0.1038, -0.1091, 0.1899, 0.1976, 0.2102,\n", + " 0.1640, 0.1080, 0.1473, 0.1007, -0.1476, 0.1616, 0.0728, 0.1338,\n", + " -0.1855, -0.0297, 0.0540, -0.2904, -0.0517, 0.1741, -0.0420, 0.3041,\n", + " 0.3847, 0.0842, 0.1851, 0.3414, 0.0222, -0.1350, 0.0046, 0.1332,\n", + " 0.1630, -0.0436, 0.2873, -0.3657, 0.1451, -0.1844, -0.0176, 0.2510,\n", + " 0.0938, -0.1369, -0.1596, 0.2342, -0.0515, -0.1080, 0.1355, 0.2345,\n", + " 0.0675, -0.0417, -0.1794, -0.0299, 0.2514, -0.0580, 0.2600, 0.1556,\n", + " 0.0266, 0.1503, -0.0103, 0.2910, -0.1993, -0.0096, -0.0035, 0.1479,\n", + " 0.1164, 0.0386, 0.1574, 0.1825, 0.1326, -0.1137, 0.0642, -0.0252]),\n", + " 'res2.conv1.conv2d.weight': tensor([[[[-2.8479e-01, -1.1376e-01, 5.0692e-02],\n", + " [-8.5171e-02, -4.3133e-02, -7.1392e-02],\n", + " [-5.7773e-02, 3.6198e-02, -8.6921e-02]],\n", + " \n", + " [[-2.8316e-02, -6.9627e-02, -4.9544e-02],\n", + " [-1.3503e-01, 7.9855e-02, -2.3667e-01],\n", + " [ 1.5325e-02, -1.0592e-01, 7.9682e-02]],\n", + " \n", + " [[ 4.6703e-02, 8.2506e-02, 7.3950e-02],\n", + " [ 1.5729e-02, -2.7815e-03, -3.2276e-02],\n", + " [ 2.1214e-01, 2.8458e-02, -9.3028e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[-1.2584e-01, 7.6694e-02, 1.2798e-01],\n", + " [-2.2262e-01, 1.0346e-01, -6.3810e-02],\n", + " [ 9.1254e-02, -1.9460e-01, -2.4800e-02]],\n", + " \n", + " [[-2.5773e-02, -1.3711e-01, -1.2150e-01],\n", + " [-4.6972e-02, 2.0716e-02, -9.6030e-02],\n", + " [-2.1936e-02, -1.0281e-01, 8.4468e-02]],\n", + " \n", + " [[-2.0135e-02, -2.0767e-02, -1.0153e-01],\n", + " [-9.6833e-02, 1.5206e-02, -1.3013e-02],\n", + " [-2.7634e-01, -1.9917e-01, 1.1305e-01]]],\n", + " \n", + " \n", + " [[[-4.4738e-02, -2.4138e-02, 2.0883e-01],\n", + " [-1.1371e-01, 3.2768e-02, 5.0831e-02],\n", + " [ 5.9276e-02, 1.4819e-01, -1.4779e-01]],\n", + " \n", + " [[-9.5869e-02, -6.3196e-02, -1.4029e-01],\n", + " [ 1.4222e-01, -5.1501e-02, -3.2304e-03],\n", + " [-4.3834e-02, -7.2932e-02, 1.1378e-02]],\n", + " \n", + " [[-7.9616e-03, -2.2925e-03, 3.3935e-02],\n", + " [ 1.8366e-01, -1.7261e-02, -1.6606e-02],\n", + " [ 1.3559e-01, 4.2260e-02, 7.1392e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[-1.7992e-02, -1.7567e-02, 6.8419e-02],\n", + " [ 7.7323e-02, -9.8840e-02, 7.9072e-03],\n", + " [ 2.4141e-01, 1.1578e-01, 1.1615e-01]],\n", + " \n", + " [[-5.1211e-01, -5.1572e-02, 1.2112e-01],\n", + " [ 6.8758e-02, -6.0716e-02, -6.8529e-02],\n", + " [ 3.0947e-01, 2.1492e-04, -4.7278e-01]],\n", + " \n", + " [[-3.5098e-02, -1.3191e-01, 9.6339e-02],\n", + " [-8.4178e-02, -2.1918e-01, -1.6247e-01],\n", + " [-5.6405e-02, -1.2303e-01, -8.1314e-02]]],\n", + " \n", + " \n", + " [[[ 1.3421e-02, 7.3120e-02, 2.2100e-01],\n", + " [ 3.8003e-02, 3.1248e-02, 2.9217e-01],\n", + " [ 4.2738e-02, 1.3531e-01, 2.9013e-01]],\n", + " \n", + " [[ 8.8988e-02, 5.1762e-02, 1.3509e-01],\n", + " [-3.4609e-02, 4.6003e-02, 6.2882e-02],\n", + " [-6.5363e-02, 6.3685e-02, 1.1935e-01]],\n", + " \n", + " [[-1.5669e-01, -1.2196e-01, -1.7771e-01],\n", + " [-1.5287e-01, -1.1223e-01, -1.7345e-01],\n", + " [-1.7849e-01, -1.2040e-01, -1.0419e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[ 6.4569e-04, 4.0181e-02, 2.4985e-02],\n", + " [-2.0145e-02, -6.9468e-03, -1.0124e-02],\n", + " [-5.5323e-02, -5.2330e-02, 1.8089e-02]],\n", + " \n", + " [[ 7.8233e-02, 8.9882e-02, 5.6995e-02],\n", + " [ 1.6940e-01, 1.2506e-02, 6.0991e-02],\n", + " [ 2.5320e-01, 2.2339e-01, 1.4000e-01]],\n", + " \n", + " [[ 5.7363e-03, 3.4286e-02, -1.4833e-03],\n", + " [ 6.3811e-02, 3.2866e-02, 3.9516e-02],\n", + " [ 1.9363e-01, 1.7717e-01, 5.1986e-02]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[ 6.6383e-02, 2.9339e-02, 1.9726e-02],\n", + " [ 1.6299e-01, -6.8368e-02, -6.8581e-02],\n", + " [ 2.1983e-02, -7.4191e-02, -1.4236e-01]],\n", + " \n", + " [[ 1.8606e-01, 1.7197e-01, 1.2165e-01],\n", + " [ 1.9064e-01, -1.6553e-02, -3.0372e-02],\n", + " [ 1.5302e-01, 4.5091e-02, 1.7909e-01]],\n", + " \n", + " [[-8.4109e-02, -1.0504e-01, -1.1702e-01],\n", + " [ 3.3992e-02, -3.0636e-02, -8.9367e-02],\n", + " [-1.5562e-02, 4.1291e-02, -1.2215e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[-3.3011e-02, -1.9025e-02, -2.7377e-02],\n", + " [ 3.8928e-02, -5.5531e-03, 9.5138e-02],\n", + " [ 9.8072e-02, 6.9311e-02, 2.2578e-01]],\n", + " \n", + " [[ 1.3526e-01, 1.7052e-01, 1.9049e-01],\n", + " [-4.1896e-02, 4.3765e-02, -3.8684e-02],\n", + " [-7.5982e-02, 4.0699e-02, -6.9123e-02]],\n", + " \n", + " [[-2.5591e-02, 5.1988e-02, 8.5148e-02],\n", + " [ 8.5042e-02, 1.4489e-01, -8.0668e-02],\n", + " [ 4.0126e-02, 2.5494e-03, -1.5435e-01]]],\n", + " \n", + " \n", + " [[[-8.4371e-02, -1.6862e-01, -3.1526e-02],\n", + " [ 3.3333e-02, -3.8504e-04, -2.8103e-02],\n", + " [-1.9198e-02, 5.2741e-02, 8.0938e-02]],\n", + " \n", + " [[-7.6486e-02, -5.3100e-02, 4.1229e-02],\n", + " [ 1.8591e-02, 1.6539e-01, 1.7931e-02],\n", + " [ 6.6389e-02, 1.7699e-01, 1.0483e-03]],\n", + " \n", + " [[ 7.0223e-02, 7.3977e-03, -9.3232e-02],\n", + " [ 4.5527e-02, -9.7121e-02, -5.6488e-02],\n", + " [-9.1450e-02, 1.8219e-02, 2.2820e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[-7.2914e-03, -8.5358e-04, 7.7435e-02],\n", + " [ 2.0099e-02, 1.4631e-02, 5.9743e-02],\n", + " [ 2.4020e-02, -3.8166e-02, -9.6868e-02]],\n", + " \n", + " [[-2.0597e-01, -7.4379e-02, 1.1554e-01],\n", + " [ 1.0736e-01, 2.4631e-01, -8.0362e-02],\n", + " [ 1.4911e-01, -6.4002e-03, -3.2974e-01]],\n", + " \n", + " [[ 1.2490e-01, 3.2242e-02, -9.6472e-02],\n", + " [-3.5961e-03, -9.9591e-04, 3.3279e-03],\n", + " [-1.3029e-01, -6.2220e-02, 4.5827e-02]]],\n", + " \n", + " \n", + " [[[-8.7540e-02, -9.1347e-02, 1.9992e-01],\n", + " [ 6.1110e-02, -6.4786e-02, -4.9955e-03],\n", + " [ 9.7108e-02, 2.3121e-01, 1.8071e-01]],\n", + " \n", + " [[-5.8416e-02, -5.8389e-02, -2.7218e-02],\n", + " [ 3.3095e-02, 8.1417e-02, 1.1825e-01],\n", + " [-1.5703e-02, -7.8033e-02, -7.1982e-02]],\n", + " \n", + " [[-7.3219e-04, -7.2365e-02, 9.3478e-02],\n", + " [ 1.0628e-02, -3.5603e-02, 1.4938e-01],\n", + " [ 6.4779e-02, -8.1958e-02, 1.1816e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[-7.7516e-03, 2.8566e-02, 1.4159e-01],\n", + " [ 3.9773e-02, 1.3069e-02, -3.7199e-02],\n", + " [ 2.8886e-02, 2.4836e-01, 1.6801e-01]],\n", + " \n", + " [[-1.1075e-01, -2.8951e-02, -1.6041e-02],\n", + " [ 8.1426e-02, -1.1449e-01, -2.8688e-01],\n", + " [ 8.2087e-02, -2.5962e-02, -2.4265e-01]],\n", + " \n", + " [[ 1.1961e-01, 6.5363e-02, -1.7581e-02],\n", + " [-1.0804e-01, 1.5037e-01, -1.9746e-02],\n", + " [-3.6212e-02, -9.1343e-02, 1.1711e-01]]]]),\n", + " 'res2.conv1.conv2d.bias': tensor([-8.7241e-02, 1.4030e-01, -1.6523e-01, 1.7316e-01, -4.0162e-02,\n", + " -3.8368e-02, 7.6883e-02, -2.7687e-02, 3.7742e-03, 5.9076e-02,\n", + " -3.0139e-03, -4.0350e-03, 9.5900e-02, 1.3813e-01, 2.4517e-01,\n", + " -1.4761e-01, 7.6438e-02, 4.9377e-02, -3.1375e-02, 2.1554e-01,\n", + " -2.1280e-01, -1.3131e-01, 2.4181e-02, -1.7489e-01, 1.8054e-02,\n", + " -1.0493e-01, -1.2309e-01, 6.3604e-02, 9.6463e-03, -8.2891e-02,\n", + " 6.9706e-02, -9.4247e-02, 1.2283e-01, 3.1911e-01, -2.0299e-02,\n", + " -1.0123e-01, -5.9033e-02, 1.0086e-01, 1.6430e-01, 7.5684e-02,\n", + " -5.7442e-02, -2.0380e-01, -3.0294e-02, 2.1450e-03, -7.5572e-02,\n", + " 3.1395e-02, -6.7353e-02, -7.4787e-02, 1.2933e-01, 6.1168e-02,\n", + " -2.7459e-01, -2.4250e-01, -1.1291e-02, -2.2323e-01, 1.0627e-01,\n", + " 7.5568e-02, 4.5574e-03, -1.9905e-01, 1.2436e-01, 8.7904e-02,\n", + " -1.0607e-01, 1.4078e-01, 2.7969e-02, 6.9021e-02, -4.7584e-02,\n", + " -7.4815e-02, 1.1936e-02, 1.2834e-01, -5.1362e-02, -3.5112e-02,\n", + " 6.8270e-02, -2.5286e-01, 7.1597e-02, -2.4004e-02, -1.2387e-01,\n", + " -1.8619e-02, -7.1291e-02, 1.5444e-01, -5.5454e-02, 8.9045e-03,\n", + " 1.2018e-01, -1.4099e-02, 4.8979e-03, 7.0708e-02, 6.8364e-02,\n", + " -5.1996e-02, -7.9809e-02, 1.0058e-01, -3.0849e-02, 1.9275e-01,\n", + " 2.2545e-02, 1.9840e-02, 4.3712e-03, 1.1858e-01, 3.9916e-02,\n", + " -5.5756e-02, 3.9744e-02, 6.6787e-02, -5.7589e-02, 6.3466e-02,\n", + " 5.2455e-02, -6.4927e-02, 2.2643e-01, -1.5167e-02, 2.7593e-05,\n", + " -1.9279e-01, 7.6345e-02, -3.3356e-02, 6.8572e-02, -4.2105e-02,\n", + " -1.1802e-01, -2.5146e-01, 3.2298e-02, -1.4123e-01, 1.1376e-02,\n", + " 1.7273e-01, -1.5453e-02, 9.4080e-02, 2.6096e-01, 1.3264e-01,\n", + " -1.7024e-02, 2.0531e-01, -9.7501e-02, 1.1224e-01, 1.5144e-01,\n", + " 5.2359e-02, 9.9881e-02, -1.1766e-02]),\n", + " 'res2.in1.weight': tensor([1.0072, 1.0078, 1.0149, 0.7818, 1.0094, 1.0287, 0.8168, 0.9444, 0.7376,\n", + " 0.9626, 1.1959, 0.8480, 0.9693, 1.1482, 0.9366, 0.9535, 0.9091, 0.8317,\n", + " 0.8985, 0.9290, 1.0744, 0.9662, 0.9018, 1.1184, 1.0081, 1.0772, 0.8015,\n", + " 0.8354, 1.0327, 1.0239, 1.0530, 0.9499, 1.1564, 0.7803, 0.8339, 1.3326,\n", + " 0.7801, 1.0157, 1.0163, 0.9133, 0.9050, 1.0374, 0.9585, 0.9980, 0.9240,\n", + " 0.8584, 0.9186, 0.8444, 0.7040, 0.9748, 0.8526, 1.0364, 1.0675, 0.9681,\n", + " 0.8656, 0.7672, 0.8914, 1.0584, 0.8611, 0.8486, 1.0292, 0.8776, 0.9472,\n", + " 0.9588, 0.8185, 0.8572, 1.0474, 0.8487, 0.7257, 1.1224, 1.1310, 0.8882,\n", + " 1.2475, 0.7928, 0.7516, 1.0727, 1.0480, 0.8590, 0.9195, 0.9096, 0.9052,\n", + " 0.8069, 0.7983, 0.9137, 0.8598, 0.9398, 1.3406, 1.0342, 0.9466, 0.9326,\n", + " 0.8349, 0.8435, 0.9075, 0.9179, 1.0912, 0.8564, 1.0442, 0.7915, 0.8096,\n", + " 1.0051, 1.1002, 0.9855, 0.7888, 1.0989, 0.9061, 0.9482, 0.8764, 1.0260,\n", + " 1.0157, 1.1017, 0.7780, 0.7204, 0.9485, 1.0137, 0.9067, 1.2583, 0.7878,\n", + " 0.8200, 1.0648, 0.8652, 1.0363, 1.1386, 0.7597, 0.9016, 0.8352, 0.9996,\n", + " 1.1942, 1.0047]),\n", + " 'res2.in1.bias': tensor([-0.3947, -0.3362, 0.1934, -0.3369, -0.2297, -0.2770, -0.2097, 0.2689,\n", + " -0.3146, -0.2141, -0.2449, -0.5072, -0.0276, -0.1467, -0.1463, -0.6926,\n", + " 0.0488, -0.2221, -0.4008, -0.1415, -0.3599, -0.0603, -0.2532, -0.1574,\n", + " -0.4552, 0.2918, -0.1668, -0.0824, -0.2971, 0.0385, -0.0976, -0.1434,\n", + " 0.0340, -0.3161, -0.2127, -0.2905, -0.1159, -0.1823, -0.2435, -0.1475,\n", + " -0.5257, -0.4812, -0.4977, -0.3090, -0.1631, -0.0241, 0.0919, -0.3967,\n", + " -0.1718, -0.1646, -0.2262, 0.0795, 0.1104, 0.0404, -0.2230, -0.2946,\n", + " -0.0800, -0.2553, -0.2351, -0.1101, -0.2705, -0.2680, -0.1688, -0.2328,\n", + " -0.1141, -0.4127, -0.3807, -0.3362, -0.2284, -0.1614, -0.3729, -0.0898,\n", + " -0.0501, -0.2705, -0.3137, -0.1809, -0.1370, -0.2614, -0.1283, -0.4282,\n", + " -0.0669, -0.0911, -0.5140, -0.5998, -0.3082, -0.1071, -0.6764, -0.0373,\n", + " -0.1224, -0.3057, -0.1327, -0.2387, -0.1405, -0.1121, -0.1113, -0.2006,\n", + " -0.1983, -0.1813, -0.2956, -0.0055, 0.0980, -0.4703, -0.1767, -0.1254,\n", + " 0.0836, -0.1449, -0.1688, -0.1847, -0.2162, -0.3506, -0.1478, -0.3334,\n", + " -0.1574, -0.4001, -0.4888, -0.3332, -0.3287, -0.0410, -0.0015, 0.1051,\n", + " 0.0864, 0.0144, -0.3388, -0.2910, 0.1865, -0.1983, -0.0579, -0.4068]),\n", + " 'res2.conv2.conv2d.weight': tensor([[[[-4.9499e-02, -1.6921e-01, 6.0861e-02],\n", + " [-6.6181e-04, 8.5626e-03, 5.8769e-02],\n", + " [-7.9904e-02, 2.5523e-01, 7.4656e-02]],\n", + " \n", + " [[-1.9706e-01, 1.1437e-01, 3.9871e-02],\n", + " [ 4.6853e-02, -2.3521e-01, 2.1049e-01],\n", + " [ 3.6400e-01, 2.0239e-02, -3.5251e-01]],\n", + " \n", + " [[-1.2394e-01, -1.4345e-01, -2.0679e-01],\n", + " [ 5.6679e-02, 3.2603e-02, -8.6562e-02],\n", + " [ 1.2240e-01, 6.8327e-02, 3.8458e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[-1.1311e-01, 6.3331e-02, 1.5955e-01],\n", + " [-1.2890e-01, -5.7980e-02, -2.7332e-02],\n", + " [ 1.9999e-03, 2.1436e-02, -1.3781e-01]],\n", + " \n", + " [[-5.6987e-02, -9.9898e-02, 2.3030e-02],\n", + " [-8.1516e-02, 1.0774e-01, 1.0121e-01],\n", + " [ 4.3780e-02, 1.3737e-01, 4.2032e-02]],\n", + " \n", + " [[-1.4286e-01, 2.2515e-01, -2.5396e-01],\n", + " [-1.7712e-01, -9.2411e-02, 8.4502e-02],\n", + " [ 2.6528e-01, -6.2999e-02, -1.1645e-01]]],\n", + " \n", + " \n", + " [[[ 1.1430e-01, -1.1313e-02, 5.9837e-02],\n", + " [ 9.3805e-02, 1.1073e-01, -1.7295e-01],\n", + " [-3.6359e-02, -1.1769e-01, -1.1349e-01]],\n", + " \n", + " [[-2.8318e-01, -1.7381e-02, -2.1786e-02],\n", + " [-5.2052e-02, -3.6936e-02, 1.0051e-01],\n", + " [-1.8004e-01, 9.8540e-02, 2.6147e-02]],\n", + " \n", + " [[-3.0985e-01, -1.3821e-01, -6.1678e-02],\n", + " [-1.5760e-01, 4.0244e-03, 1.8992e-02],\n", + " [-1.9815e-01, -1.9200e-02, 1.8535e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[ 3.9484e-02, 4.3804e-02, -3.6764e-02],\n", + " [-1.2528e-01, -2.3319e-01, -2.8604e-01],\n", + " [-2.2678e-01, -2.9253e-01, -9.6451e-02]],\n", + " \n", + " [[-8.4570e-03, 7.5773e-02, -5.2023e-02],\n", + " [ 2.1059e-01, 1.9977e-01, 3.7698e-02],\n", + " [ 3.2332e-02, 1.3139e-01, 1.2700e-01]],\n", + " \n", + " [[-8.3059e-02, -1.1336e-01, -1.2366e-01],\n", + " [-8.9097e-02, -3.1433e-02, -3.5860e-02],\n", + " [-1.1567e-01, -1.1160e-01, -1.1035e-01]]],\n", + " \n", + " \n", + " [[[-6.5513e-02, 1.0726e-01, 1.9726e-01],\n", + " [ 4.4812e-02, -9.9547e-02, 1.3372e-01],\n", + " [-6.5367e-02, -5.0966e-03, -1.1904e-01]],\n", + " \n", + " [[-2.6945e-03, -3.8052e-02, -1.1531e-01],\n", + " [-3.9906e-02, 2.2703e-03, 6.6922e-03],\n", + " [-5.0333e-02, -1.4022e-01, -5.9635e-03]],\n", + " \n", + " [[ 2.0781e-02, 4.6462e-02, 8.7550e-02],\n", + " [ 1.0478e-02, 2.3837e-02, 1.1676e-01],\n", + " [-4.6284e-02, -2.5770e-02, 2.3781e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[ 1.5299e-01, 9.3693e-02, 5.9692e-02],\n", + " [ 4.6835e-02, -3.0243e-02, 1.0438e-02],\n", + " [ 9.5483e-02, 9.5892e-02, 1.2023e-01]],\n", + " \n", + " [[-2.3618e-01, -1.1173e-01, -8.3494e-02],\n", + " [ 5.0051e-03, 1.0608e-01, -3.4035e-02],\n", + " [ 4.6339e-02, 6.4561e-02, -4.9160e-02]],\n", + " \n", + " [[ 7.2047e-03, -2.7344e-02, -1.4849e-01],\n", + " [ 2.6306e-02, 2.0425e-01, 2.3514e-02],\n", + " [-5.4583e-02, 1.8510e-01, 1.7382e-01]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[-3.0920e-02, -1.8528e-01, 4.4792e-02],\n", + " [-2.2710e-02, 4.7410e-02, 6.6992e-02],\n", + " [-2.2404e-02, 7.2010e-02, 1.0823e-01]],\n", + " \n", + " [[ 9.4218e-02, 1.3266e-02, -8.3503e-02],\n", + " [-5.9164e-02, -2.9558e-02, -2.9574e-02],\n", + " [ 3.3591e-03, 7.5116e-02, -3.1897e-02]],\n", + " \n", + " [[-4.7244e-02, 5.6799e-02, 8.6695e-02],\n", + " [-3.2492e-02, 5.2959e-02, -1.3672e-02],\n", + " [-3.2704e-02, 8.1105e-03, -1.1960e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[-2.3260e-01, -1.5324e-01, -1.4286e-01],\n", + " [-8.3255e-02, 3.3887e-03, 5.0203e-02],\n", + " [ 5.7969e-02, 4.1244e-02, 2.6512e-01]],\n", + " \n", + " [[ 1.6965e-01, 8.8070e-02, 1.2276e-01],\n", + " [ 5.4114e-02, 8.5052e-03, 9.0397e-02],\n", + " [ 3.4805e-02, 1.4262e-01, 2.0103e-01]],\n", + " \n", + " [[-1.8788e-01, -6.0789e-02, 6.2761e-02],\n", + " [-1.1616e-01, -1.0704e-01, 3.2513e-02],\n", + " [ 6.4125e-03, 4.8944e-02, 1.1826e-01]]],\n", + " \n", + " \n", + " [[[-5.7765e-02, 1.3480e-01, 1.9648e-02],\n", + " [ 6.8762e-02, 8.2725e-02, -9.4882e-02],\n", + " [ 7.6113e-02, 1.1860e-01, -1.1551e-03]],\n", + " \n", + " [[ 7.8994e-02, 2.2005e-01, -3.8513e-02],\n", + " [ 1.1985e-01, -3.4199e-03, 1.8276e-01],\n", + " [ 1.2585e-01, 4.7585e-02, -1.7434e-01]],\n", + " \n", + " [[ 8.8875e-03, -1.2302e-01, -6.8870e-02],\n", + " [-3.1787e-02, -1.9748e-02, -3.7442e-03],\n", + " [-1.9490e-02, -6.6684e-02, 7.3693e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[ 1.7585e-02, 3.7644e-02, 1.8113e-01],\n", + " [ 1.1804e-02, -5.4994e-03, -9.2490e-02],\n", + " [-3.2816e-02, -3.5521e-02, -3.8899e-02]],\n", + " \n", + " [[ 2.2757e-02, 1.1734e-01, -1.9051e-01],\n", + " [ 2.4648e-01, 3.1440e-01, -1.1106e-01],\n", + " [ 1.7772e-01, 1.1779e-01, -8.0991e-02]],\n", + " \n", + " [[ 9.9837e-02, 2.5919e-01, -1.0856e-01],\n", + " [-2.4754e-02, -5.9647e-02, 7.1373e-02],\n", + " [ 2.4026e-01, -8.1840e-02, -1.2106e-01]]],\n", + " \n", + " \n", + " [[[ 6.7444e-02, -2.5566e-02, -1.5283e-02],\n", + " [-5.8239e-02, -5.5279e-02, -1.0442e-01],\n", + " [-7.3907e-02, 9.3639e-02, 1.7078e-01]],\n", + " \n", + " [[-1.9951e-05, 1.1322e-01, -7.1210e-02],\n", + " [ 2.3559e-02, 9.0756e-02, 5.9844e-02],\n", + " [ 5.8806e-02, 4.0438e-02, 2.9969e-02]],\n", + " \n", + " [[-3.7815e-01, -1.9857e-01, -8.4507e-02],\n", + " [-2.6360e-01, -7.3541e-02, 4.3112e-02],\n", + " [-8.4886e-02, 2.6406e-02, 1.7010e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[-1.2021e-01, -8.7870e-02, -1.7276e-01],\n", + " [-6.7242e-03, 2.4623e-02, -8.3595e-02],\n", + " [ 1.5567e-01, 1.7177e-01, 5.4462e-02]],\n", + " \n", + " [[ 3.7190e-02, -9.0624e-03, 3.4346e-02],\n", + " [-1.7025e-02, -1.3693e-01, -7.1966e-02],\n", + " [-1.6731e-01, -1.3859e-01, -9.8624e-02]],\n", + " \n", + " [[ 3.1070e-01, 8.3704e-02, -1.9167e-02],\n", + " [ 1.2796e-01, 1.8873e-01, -2.7085e-02],\n", + " [ 4.4296e-02, -1.9290e-02, -1.5059e-02]]]]),\n", + " 'res2.conv2.conv2d.bias': tensor([-0.1360, 0.0084, 0.1275, 0.4759, 0.0767, -0.0970, 0.2368, 0.0860,\n", + " -0.0933, 0.2198, -0.1143, 0.0149, 0.1942, 0.0607, -0.0872, -0.2284,\n", + " 0.1061, 0.3948, -0.1840, -0.2097, -0.0216, -0.0067, 0.0211, 0.1957,\n", + " -0.2044, 0.1199, 0.1710, -0.1315, 0.0913, 0.1982, 0.0430, -0.0526,\n", + " 0.0375, -0.0835, -0.1044, -0.0991, -0.0388, -0.0889, 0.0030, -0.1404,\n", + " -0.0163, -0.2504, 0.1412, -0.0657, -0.0169, -0.0577, -0.0297, -0.2277,\n", + " 0.0572, -0.0838, -0.0587, -0.0501, -0.1128, 0.0628, 0.1479, -0.4309,\n", + " 0.1308, 0.1003, -0.0255, -0.0919, 0.1406, 0.0418, 0.0126, 0.1555,\n", + " -0.0963, 0.0596, -0.2893, -0.2128, 0.2051, 0.0643, 0.1285, -0.1031,\n", + " 0.1513, 0.1562, -0.3233, 0.0006, -0.0026, 0.2529, -0.2412, -0.0428,\n", + " -0.0595, 0.0701, -0.0062, -0.0551, -0.0299, 0.0919, 0.1187, 0.0843,\n", + " 0.1983, -0.0800, 0.0540, 0.0152, -0.2908, -0.0231, 0.0225, -0.0465,\n", + " 0.1954, 0.0374, -0.0107, -0.1831, 0.1309, -0.2177, 0.0271, 0.1824,\n", + " 0.0842, -0.0188, 0.1180, -0.2002, -0.1223, 0.0902, 0.1254, 0.0255,\n", + " 0.1562, -0.0332, 0.0713, 0.1172, -0.0439, -0.2072, -0.1939, 0.0035,\n", + " -0.0652, 0.1511, 0.0896, 0.0250, 0.1143, -0.0710, -0.1212, -0.1463]),\n", + " 'res2.in2.weight': tensor([0.9658, 0.7586, 0.7982, 0.9283, 0.9076, 0.7419, 0.6606, 0.8581, 0.7752,\n", + " 0.8015, 0.8155, 0.7972, 0.8387, 0.9579, 0.8744, 0.8076, 0.7514, 0.8723,\n", + " 0.7357, 0.8738, 0.7829, 0.8417, 0.7404, 0.8753, 0.8395, 0.7669, 0.6660,\n", + " 0.5886, 0.8903, 0.8112, 0.8267, 0.8655, 0.7343, 0.8644, 0.8538, 0.8100,\n", + " 0.8490, 0.7579, 0.9706, 0.6839, 0.8654, 0.9326, 0.7986, 0.8965, 0.8792,\n", + " 0.8464, 0.7922, 0.8295, 0.7785, 0.7529, 0.5797, 0.6456, 0.6899, 0.7809,\n", + " 0.8135, 0.7214, 0.7746, 0.9887, 0.7864, 0.8531, 0.7679, 0.9652, 0.8558,\n", + " 0.7437, 0.8987, 0.6745, 0.7349, 0.7141, 0.6936, 1.0570, 0.8255, 1.0258,\n", + " 0.8919, 0.7238, 0.9900, 0.8674, 0.8482, 0.7874, 0.9007, 0.6715, 0.7515,\n", + " 0.6576, 0.6183, 0.7475, 0.7938, 0.6511, 0.5682, 1.0286, 0.8850, 0.8145,\n", + " 0.7607, 0.7151, 0.9610, 0.8398, 0.8568, 0.7784, 0.9738, 0.9767, 0.7894,\n", + " 0.7290, 0.8513, 0.8563, 0.8827, 0.8059, 0.9764, 0.7418, 0.8971, 0.7019,\n", + " 1.0056, 0.8537, 1.2733, 0.6755, 0.8619, 0.7207, 0.8373, 0.6675, 0.8518,\n", + " 0.6108, 0.8038, 0.7502, 0.8623, 0.7407, 0.8235, 1.0165, 0.8151, 0.7468,\n", + " 0.9754, 0.7427]),\n", + " 'res2.in2.bias': tensor([-1.8965e-01, -5.2305e-02, -1.8839e-01, 8.3806e-02, -1.0245e-01,\n", + " -1.1940e-01, 1.0310e-01, 1.9733e-02, 1.0057e-01, -7.4085e-02,\n", + " 1.9997e-01, -1.4206e-02, 1.7688e-01, 3.6577e-02, 2.2658e-01,\n", + " -7.3383e-03, -4.5483e-01, 1.2103e-01, 2.7434e-02, -2.7690e-02,\n", + " 2.6573e-01, 1.2365e-01, -2.7135e-01, 1.6157e-01, -5.1876e-02,\n", + " 4.9882e-02, 1.5032e-01, -7.4297e-02, -7.7606e-02, 1.0260e-01,\n", + " -4.6720e-02, -5.9714e-02, -1.9428e-01, 1.3112e-01, 6.4156e-02,\n", + " 4.3017e-02, 6.2715e-02, -8.7879e-02, -7.4889e-02, -2.6815e-02,\n", + " -1.0118e-01, -1.2942e-01, -4.7342e-04, -1.8591e-01, -1.4096e-01,\n", + " 9.0691e-02, 6.2615e-02, -6.2229e-02, 1.0558e-01, 2.9326e-01,\n", + " 5.9525e-02, 6.5302e-02, 6.5374e-02, -4.7276e-02, -1.0552e-01,\n", + " -2.4129e-01, -1.6290e-01, -6.7363e-02, -6.5475e-02, -1.1750e-01,\n", + " 9.2670e-03, 1.5529e-01, 1.2393e-01, 1.8439e-01, 8.6457e-02,\n", + " -3.7866e-02, 1.2306e-01, -2.2550e-02, -1.2976e-01, 1.9973e-01,\n", + " 2.3922e-01, 4.1310e-02, -1.7378e-01, 9.3484e-02, -8.7396e-02,\n", + " -3.4168e-01, -1.4361e-01, 2.8536e-01, -6.2676e-02, 4.1869e-01,\n", + " 3.3745e-01, 5.7893e-02, 2.1105e-01, 3.3082e-01, 1.2066e-03,\n", + " -1.2763e-01, 6.3111e-02, 1.7331e-01, 1.7355e-01, -9.5349e-02,\n", + " 1.8616e-01, -5.1015e-01, 1.2550e-01, -1.3148e-01, -9.9083e-02,\n", + " 8.5861e-02, 2.3314e-01, -1.7176e-01, -1.9761e-01, 1.2500e-01,\n", + " 2.2488e-02, -4.8003e-02, 4.9331e-02, 2.7510e-01, -1.0740e-02,\n", + " -1.0087e-01, -3.2137e-01, -7.4408e-02, 1.0600e-01, -1.1437e-01,\n", + " 1.9349e-01, 4.0413e-02, 5.2862e-02, 1.4657e-01, -8.9429e-02,\n", + " 4.3375e-01, -2.3903e-01, -8.0893e-02, 6.0108e-02, 1.9601e-01,\n", + " 1.1543e-01, 2.9684e-02, 1.5473e-01, 6.7264e-02, 1.4373e-01,\n", + " -6.6453e-02, 2.0669e-02, -1.5275e-02]),\n", + " 'res3.conv1.conv2d.weight': tensor([[[[-1.1609e-01, -1.7069e-01, -2.5837e-01],\n", + " [-1.0469e-02, -3.0430e-02, 2.9421e-02],\n", + " [ 2.0498e-01, 1.9280e-01, 1.0123e-01]],\n", + " \n", + " [[-5.3386e-01, -1.7067e-01, -2.3497e-01],\n", + " [-2.1265e-01, 8.1629e-03, -8.8374e-02],\n", + " [-1.8601e-01, 3.9074e-02, -5.2061e-02]],\n", + " \n", + " [[-2.2077e-01, -7.6561e-02, -1.6181e-01],\n", + " [-4.8389e-02, 3.8926e-02, -2.8892e-02],\n", + " [-1.0759e-01, -1.3117e-02, -9.7990e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[-3.9469e-03, 2.4033e-02, 1.0112e-01],\n", + " [ 5.4674e-02, -1.3728e-03, 8.6878e-02],\n", + " [ 8.3609e-02, 2.3179e-02, 1.7826e-01]],\n", + " \n", + " [[ 1.3498e-01, 1.2010e-01, -2.3411e-01],\n", + " [ 2.7302e-01, 1.5974e-01, -1.3512e-01],\n", + " [ 4.2025e-01, 3.0661e-01, -3.9469e-02]],\n", + " \n", + " [[-1.9019e-01, -5.3550e-02, 1.4613e-01],\n", + " [-1.4127e-01, -3.3497e-02, 6.4636e-02],\n", + " [-1.6675e-01, 1.3767e-03, 1.3199e-01]]],\n", + " \n", + " \n", + " [[[ 3.8339e-03, 4.2126e-02, -2.0730e-02],\n", + " [ 3.7282e-02, 1.4567e-01, -1.2878e-01],\n", + " [-3.4215e-02, 1.6833e-02, -6.8646e-02]],\n", + " \n", + " [[-1.0571e-01, -6.2654e-02, 6.5977e-02],\n", + " [ 5.1375e-04, 1.1260e-01, 1.6235e-01],\n", + " [ 1.9897e-02, 5.9935e-02, 2.5737e-01]],\n", + " \n", + " [[-1.1808e-01, -7.9385e-02, 1.0705e-01],\n", + " [-3.4120e-02, -1.2226e-03, 1.3681e-01],\n", + " [-5.3098e-02, -7.8288e-03, -8.9244e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[ 1.9072e-02, -1.1240e-02, -2.3656e-01],\n", + " [-1.6754e-02, 1.8569e-02, -1.4753e-01],\n", + " [-8.7849e-02, -2.2057e-02, -1.5469e-01]],\n", + " \n", + " [[-9.5852e-02, -6.8315e-02, 1.0081e-02],\n", + " [-3.5441e-04, -1.0303e-02, 2.9195e-02],\n", + " [-1.4856e-01, -1.2354e-01, 8.8742e-02]],\n", + " \n", + " [[-8.0049e-02, -1.0436e-01, -2.5480e-01],\n", + " [-1.1132e-01, -5.7981e-02, -4.0847e-01],\n", + " [-1.3531e-01, -1.0147e-01, -2.0211e-01]]],\n", + " \n", + " \n", + " [[[ 8.1888e-02, -8.5160e-02, -1.4952e-02],\n", + " [-4.7041e-02, -1.5701e-02, 1.2149e-01],\n", + " [-1.4303e-01, -1.6713e-01, -6.4522e-03]],\n", + " \n", + " [[-1.1280e-01, 8.4119e-02, 1.0815e-01],\n", + " [-8.2066e-02, -6.6016e-02, 2.2916e-02],\n", + " [-4.0736e-02, 4.7370e-02, 1.5833e-01]],\n", + " \n", + " [[-1.2837e-01, 1.6771e-02, 5.2128e-02],\n", + " [-3.3256e-02, 1.5227e-01, 5.4502e-02],\n", + " [ 6.5546e-02, 1.3288e-01, 2.0091e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[ 1.2449e-01, 7.7291e-02, -2.3609e-01],\n", + " [ 2.7022e-02, 8.1956e-02, -2.9050e-01],\n", + " [-2.1514e-01, 1.5280e-02, -2.7384e-01]],\n", + " \n", + " [[-6.0487e-02, -1.5560e-01, -9.1618e-03],\n", + " [ 2.4009e-02, -1.4126e-01, 2.0171e-01],\n", + " [-1.1292e-02, -1.2330e-01, 1.4027e-01]],\n", + " \n", + " [[ 2.6771e-01, 3.2275e-02, 9.6203e-02],\n", + " [ 8.2198e-02, -1.1988e-01, 7.1176e-04],\n", + " [-1.3736e-01, 4.4135e-02, -1.4750e-01]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[-1.1563e-02, 3.9454e-02, -1.1141e-01],\n", + " [ 1.8863e-01, -9.7773e-02, -2.8746e-01],\n", + " [-6.1049e-02, -1.2775e-02, 1.6897e-01]],\n", + " \n", + " [[ 7.3220e-02, -2.5745e-02, -4.7972e-02],\n", + " [ 1.1278e-01, 4.4096e-02, 5.1552e-02],\n", + " [-1.0096e-01, -1.1795e-01, 6.3297e-02]],\n", + " \n", + " [[ 2.9113e-01, 8.8301e-02, -8.3575e-02],\n", + " [-3.6150e-02, -6.7838e-02, 1.3970e-02],\n", + " [-1.3112e-01, 3.5019e-02, 6.0396e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[ 3.0627e-02, -1.1056e-01, -4.9309e-02],\n", + " [-1.0242e-01, 1.5759e-03, -1.8348e-01],\n", + " [ 8.2292e-02, -1.4049e-01, -9.0351e-02]],\n", + " \n", + " [[ 3.2791e-02, 3.3366e-02, -1.7203e-02],\n", + " [-3.9484e-02, -1.8502e-01, -2.2880e-01],\n", + " [-1.4092e-01, -2.0215e-01, -8.4469e-02]],\n", + " \n", + " [[-6.9072e-02, -4.9129e-03, -1.5211e-02],\n", + " [-3.6900e-03, 8.8558e-02, -1.3210e-02],\n", + " [ 8.1512e-02, -5.0019e-02, -1.1155e-01]]],\n", + " \n", + " \n", + " [[[ 1.2913e-01, -2.6147e-01, 1.3539e-02],\n", + " [-1.5405e-01, 4.0035e-01, -6.4897e-02],\n", + " [ 1.5826e-01, -2.4377e-02, 2.4680e-01]],\n", + " \n", + " [[ 8.3347e-02, 7.9119e-02, 5.2384e-02],\n", + " [ 7.1092e-02, -7.0102e-02, 8.6303e-02],\n", + " [-7.3022e-02, 2.2444e-01, 1.2770e-01]],\n", + " \n", + " [[-3.7014e-03, 5.9168e-02, -1.5715e-02],\n", + " [-1.0885e-01, 1.1232e-01, 1.2586e-02],\n", + " [-4.0675e-03, 1.4467e-01, -1.3884e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[-1.1082e-02, -3.5962e-02, -9.9846e-02],\n", + " [ 8.9019e-02, -1.5646e-01, 9.6009e-02],\n", + " [ 1.0487e-03, 2.2669e-01, 4.2763e-02]],\n", + " \n", + " [[ 1.4505e-02, -4.7036e-02, -8.7927e-02],\n", + " [-3.0930e-01, 2.1180e-01, -4.8425e-02],\n", + " [ 1.2404e-01, -7.0750e-02, 3.2839e-02]],\n", + " \n", + " [[ 1.6709e-01, -2.5157e-01, 1.3788e-01],\n", + " [ 7.2827e-03, 7.0279e-02, 1.4321e-01],\n", + " [-3.3459e-02, 6.9324e-02, -5.4028e-02]]],\n", + " \n", + " \n", + " [[[-9.6677e-02, -1.6946e-02, 1.2855e-01],\n", + " [-8.4707e-02, 1.0252e-01, 1.8556e-01],\n", + " [ 1.3415e-01, -1.7065e-01, 3.5239e-03]],\n", + " \n", + " [[-1.1608e-01, -1.4518e-02, -1.0537e-01],\n", + " [ 3.0889e-02, 3.8211e-02, -6.6929e-02],\n", + " [ 1.7197e-02, -1.4086e-01, 1.6077e-01]],\n", + " \n", + " [[ 6.7687e-02, 4.3321e-02, -5.1000e-02],\n", + " [ 1.8070e-01, 1.2294e-01, 4.4548e-02],\n", + " [ 9.1826e-02, 3.4974e-02, -1.0485e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[-5.9668e-03, 3.4441e-02, -1.4585e-02],\n", + " [ 3.7610e-02, -1.4849e-02, -7.8803e-03],\n", + " [ 8.3645e-02, 6.9245e-03, -1.2612e-02]],\n", + " \n", + " [[-6.0025e-02, -1.1669e-01, -1.1945e-01],\n", + " [-1.5887e-01, -6.7179e-02, -6.6477e-02],\n", + " [ 9.9711e-02, -1.4505e-01, -1.1687e-01]],\n", + " \n", + " [[-2.5156e-01, -9.3575e-02, 2.0201e-01],\n", + " [-8.2606e-02, -6.7653e-02, -7.0182e-02],\n", + " [ 3.8416e-02, 1.9871e-01, -2.5996e-02]]]]),\n", + " 'res3.conv1.conv2d.bias': tensor([ 0.1117, 0.0981, 0.0745, -0.1668, -0.1008, -0.1090, -0.0561, -0.0058,\n", + " -0.2345, -0.0555, -0.2739, 0.1045, 0.1778, 0.1857, 0.0148, 0.0558,\n", + " 0.0781, 0.1115, -0.0216, -0.0258, 0.0939, 0.0191, 0.0402, -0.2678,\n", + " 0.0048, -0.1281, 0.1640, 0.0785, 0.1198, -0.1777, 0.1418, -0.0379,\n", + " 0.2064, 0.0399, -0.0820, -0.1173, -0.0767, 0.0221, 0.0676, -0.1843,\n", + " -0.2098, 0.1709, -0.1934, 0.0441, 0.0297, 0.2466, 0.0524, -0.0463,\n", + " -0.0931, -0.0145, -0.0478, 0.2393, 0.1210, 0.0681, -0.0352, 0.0839,\n", + " -0.1295, -0.1381, -0.0683, 0.0545, -0.0639, 0.0059, 0.1267, 0.0447,\n", + " -0.0906, 0.0514, 0.2153, 0.1402, 0.2274, 0.1241, 0.1548, -0.0277,\n", + " 0.0277, 0.2791, 0.0886, -0.0655, 0.1576, 0.0350, 0.0581, -0.0694,\n", + " 0.0968, 0.1483, 0.0007, -0.0316, 0.0458, 0.1097, -0.0318, 0.0151,\n", + " 0.0966, -0.0066, 0.2940, 0.0487, 0.1881, 0.0205, -0.1905, -0.1053,\n", + " 0.1425, -0.1231, 0.1212, 0.0850, -0.0051, -0.1852, -0.0638, 0.0178,\n", + " 0.1673, 0.1092, 0.2684, 0.1033, 0.0383, -0.0337, 0.0185, 0.1027,\n", + " -0.0743, -0.1898, 0.0054, 0.0139, 0.2922, 0.1832, -0.0958, 0.0789,\n", + " 0.2155, 0.1160, 0.0992, -0.1991, 0.0762, 0.3442, 0.0028, 0.2370]),\n", + " 'res3.in1.weight': tensor([1.1768, 0.7900, 0.8260, 0.8714, 0.9367, 0.9654, 0.8747, 0.8610, 1.0593,\n", + " 1.0637, 1.1329, 0.8668, 0.8398, 1.0084, 0.8791, 0.9168, 0.7735, 0.8846,\n", + " 1.0017, 0.7313, 0.9520, 0.8610, 1.0807, 0.9313, 1.1252, 0.8891, 0.7741,\n", + " 0.8631, 0.9151, 0.9777, 0.8401, 1.0335, 0.8485, 0.8802, 0.8677, 1.0341,\n", + " 0.9392, 0.8342, 0.9435, 0.9701, 0.9437, 0.9334, 0.8599, 0.7962, 1.0932,\n", + " 0.8915, 0.8901, 0.8877, 0.9639, 0.8958, 1.0257, 0.9140, 0.9202, 0.8887,\n", + " 0.8971, 1.0156, 0.9423, 0.8967, 0.8515, 1.0347, 0.9704, 0.9675, 0.9444,\n", + " 0.9662, 0.9732, 1.0764, 0.9235, 0.9542, 0.8517, 0.8176, 0.8342, 0.8296,\n", + " 0.8876, 0.8987, 0.9457, 0.8399, 0.8521, 0.9141, 0.9725, 0.8562, 0.7293,\n", + " 0.9484, 0.8929, 1.0700, 0.7808, 0.8655, 0.7903, 0.6969, 0.9035, 0.9769,\n", + " 0.8739, 1.0362, 0.9120, 0.7562, 0.9282, 1.0979, 0.8386, 0.8965, 0.9999,\n", + " 0.8265, 1.0670, 0.8306, 0.8938, 0.8680, 0.8388, 0.9360, 0.8329, 0.9227,\n", + " 0.9175, 0.8611, 0.8286, 1.0674, 0.9387, 0.8699, 0.8641, 1.1296, 1.0638,\n", + " 0.8989, 0.9244, 0.8980, 0.9605, 0.7617, 0.9529, 1.0780, 0.7941, 0.8626,\n", + " 1.0856, 0.8684]),\n", + " 'res3.in1.bias': tensor([-0.1420, -0.1657, -0.1682, -0.1522, -0.1005, -0.2354, -0.3610, -0.2282,\n", + " -0.1064, -1.2700, -0.6240, -0.2690, -0.0929, -0.7240, -0.4012, -0.5499,\n", + " -0.4673, -0.2932, -0.2753, -0.0554, 0.0838, -0.7346, 0.1085, -0.1935,\n", + " -0.3231, -0.1260, -0.4741, -0.1803, -0.2349, 0.1756, -0.5484, -0.2838,\n", + " -0.0966, -0.3858, -0.0456, -0.2816, -0.2176, -0.2220, -0.5023, -0.0963,\n", + " -0.0023, -0.4224, -0.2180, 0.0833, -0.3304, -0.6287, -0.4498, -0.2859,\n", + " -0.5113, -0.7781, 0.0283, -0.3522, -0.1712, -0.3047, -0.2939, -0.1014,\n", + " -0.5645, -0.6912, -0.3866, -0.8890, -0.1139, -0.1624, -0.3062, -0.2612,\n", + " -0.1680, -0.9880, -0.1313, -0.2380, -0.7599, -0.4702, -0.4803, -0.3610,\n", + " 0.2720, -0.4311, -0.5672, 0.3365, -0.2808, 0.2081, -0.2686, -0.4154,\n", + " -0.3759, -0.2772, 0.0494, -0.2324, -0.4389, -0.0149, -0.3091, -0.4503,\n", + " -0.2168, -0.0143, -0.1682, -0.2698, -0.6359, -0.5734, -0.0986, -0.0874,\n", + " -0.3126, -0.3302, 0.0219, -0.1685, -0.6010, -0.3508, -0.2390, -0.5489,\n", + " -0.5646, -0.1540, -0.1738, 0.0328, -0.5210, -0.0088, -0.1393, -0.5733,\n", + " -0.0600, -0.1498, -0.0923, -0.3488, 0.2807, -0.2389, -0.7532, -0.2529,\n", + " -0.4365, -0.1894, 0.2036, -0.4901, -0.0051, -0.3982, -0.1396, -0.4696]),\n", + " 'res3.conv2.conv2d.weight': tensor([[[[-3.6048e-02, -1.9029e-01, -2.1392e-01],\n", + " [ 3.7167e-02, -8.0805e-02, -5.6552e-02],\n", + " [ 2.7027e-01, 1.1765e-01, 1.8593e-01]],\n", + " \n", + " [[-2.8534e-02, -1.1566e-01, 1.2959e-01],\n", + " [ 2.0133e-02, -6.5767e-02, 6.3957e-02],\n", + " [ 5.4273e-02, 1.9444e-01, 9.0579e-03]],\n", + " \n", + " [[-5.7275e-02, 1.4187e-02, 1.3815e-02],\n", + " [-1.2080e-01, 7.5764e-02, -3.9675e-03],\n", + " [ 3.9009e-02, 3.5433e-02, 5.0508e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[ 2.9252e-02, 5.3217e-02, -7.4351e-02],\n", + " [-3.0836e-04, -7.5401e-02, 1.4454e-01],\n", + " [-5.9745e-02, 1.3515e-01, -2.2275e-01]],\n", + " \n", + " [[-1.4812e-01, -1.0975e-01, 9.3697e-03],\n", + " [-3.2081e-01, 2.5620e-01, -2.2091e-01],\n", + " [ 1.1866e-01, -1.4676e-01, -1.1365e-01]],\n", + " \n", + " [[-6.8539e-02, 1.4799e-02, -2.1411e-02],\n", + " [-5.8087e-02, -3.2960e-02, 4.5942e-03],\n", + " [ 1.9802e-01, -1.5336e-02, 4.9920e-03]]],\n", + " \n", + " \n", + " [[[ 2.6668e-02, 3.9871e-02, 2.6900e-02],\n", + " [-4.2741e-02, -2.5045e-02, -3.9515e-02],\n", + " [-1.5705e-02, 6.7783e-02, 6.1343e-02]],\n", + " \n", + " [[-4.2318e-02, -3.9389e-02, -3.5795e-02],\n", + " [-8.2648e-02, 1.6521e-01, -9.2203e-03],\n", + " [-1.0766e-01, -2.0539e-01, -4.0151e-02]],\n", + " \n", + " [[ 2.7368e-02, 7.4645e-02, -3.4156e-02],\n", + " [-1.3676e-01, -3.1905e-02, 5.8259e-02],\n", + " [-8.2770e-02, -8.0066e-03, 1.5846e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[ 3.7125e-02, 3.4975e-02, -5.1996e-02],\n", + " [ 1.6226e-01, 5.2388e-02, 1.5883e-03],\n", + " [ 2.7598e-02, 2.8792e-03, -3.0051e-02]],\n", + " \n", + " [[ 2.5410e-02, -9.7856e-02, -1.7766e-02],\n", + " [ 1.5873e-01, -2.2654e-01, 3.2504e-02],\n", + " [-1.1606e-01, -3.1836e-02, 5.3390e-02]],\n", + " \n", + " [[ 6.8531e-02, 7.3807e-02, -2.5438e-02],\n", + " [ 5.7421e-02, 1.1168e-02, 9.9935e-02],\n", + " [-8.8385e-03, -6.5296e-02, 1.0394e-03]]],\n", + " \n", + " \n", + " [[[ 6.1019e-02, 5.2481e-02, 4.4960e-02],\n", + " [ 1.1867e-01, 1.0318e-01, 5.0429e-02],\n", + " [ 8.1501e-04, 2.2838e-03, 4.9273e-02]],\n", + " \n", + " [[ 9.2184e-02, 1.3309e-01, 9.0749e-02],\n", + " [ 1.4979e-01, 1.0594e-01, 8.0226e-02],\n", + " [ 5.8763e-02, 1.1959e-01, 1.3015e-01]],\n", + " \n", + " [[ 9.7154e-02, -1.0363e-01, -1.7009e-01],\n", + " [ 2.1247e-03, -2.2782e-01, -1.2755e-01],\n", + " [-9.4223e-02, -2.0301e-01, 5.6641e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[-1.3435e-01, -5.9684e-02, -1.0467e-01],\n", + " [ 2.0369e-02, 2.9608e-02, 7.4630e-02],\n", + " [ 6.5095e-02, -1.6396e-01, 8.2674e-02]],\n", + " \n", + " [[-2.8955e-02, -1.5645e-02, -9.0651e-02],\n", + " [-4.9454e-02, -1.0599e-01, -1.1138e-01],\n", + " [-9.8444e-02, -1.1073e-01, -9.4295e-02]],\n", + " \n", + " [[-7.8994e-02, -5.5578e-02, -3.3525e-02],\n", + " [ 6.9237e-04, 8.1981e-03, -3.3499e-02],\n", + " [-1.2967e-01, -9.4675e-02, 1.0604e-01]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[-7.2462e-02, 8.8525e-02, 7.4402e-02],\n", + " [-1.2797e-01, 7.5592e-02, 2.1384e-02],\n", + " [-2.9420e-01, -1.0395e-01, -2.3508e-01]],\n", + " \n", + " [[-3.7998e-02, 1.1202e-02, -9.7367e-02],\n", + " [ 4.2506e-02, -8.4688e-05, 7.5174e-03],\n", + " [ 2.0839e-02, -1.0240e-01, -4.9254e-02]],\n", + " \n", + " [[ 2.1464e-01, -1.1460e-01, 3.3649e-02],\n", + " [ 2.1697e-01, 3.8149e-05, -5.4135e-02],\n", + " [ 2.3380e-01, -7.7468e-03, 1.1990e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[ 1.5210e-01, 5.4148e-02, -8.0549e-02],\n", + " [ 7.5881e-02, 5.1542e-02, -1.1855e-01],\n", + " [ 2.6889e-02, -7.7237e-02, 7.9418e-02]],\n", + " \n", + " [[-4.0914e-02, 4.3307e-02, 5.2727e-03],\n", + " [-7.8892e-02, -1.5784e-01, 8.2513e-02],\n", + " [-3.0463e-02, -1.9523e-02, -7.6031e-02]],\n", + " \n", + " [[ 2.2535e-01, 9.2765e-02, -1.0853e-01],\n", + " [-4.8466e-02, -5.2539e-02, 2.7693e-02],\n", + " [-1.0780e-01, -2.6245e-01, -1.7536e-01]]],\n", + " \n", + " \n", + " [[[ 6.8265e-02, -6.4267e-02, -7.8462e-02],\n", + " [ 7.1109e-02, -1.0031e-01, -8.2332e-02],\n", + " [ 5.3313e-02, -5.0521e-02, -7.3301e-02]],\n", + " \n", + " [[ 5.6225e-02, -7.3316e-02, -1.0511e-01],\n", + " [ 3.9097e-02, -1.1211e-01, -1.1470e-01],\n", + " [ 1.1688e-02, -4.1653e-02, -2.2511e-01]],\n", + " \n", + " [[-5.6840e-02, -3.7220e-02, -8.5166e-02],\n", + " [-8.3427e-02, 4.6673e-02, 1.0894e-01],\n", + " [-5.6008e-02, 3.3300e-02, 1.3595e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[-2.6878e-02, 1.4680e-01, 1.9964e-01],\n", + " [-1.9783e-03, 2.5644e-01, 1.8074e-02],\n", + " [ 1.0214e-01, 1.0043e-01, -1.8961e-02]],\n", + " \n", + " [[-6.2664e-02, 1.5064e-01, -1.5178e-01],\n", + " [-6.9762e-03, 3.6903e-02, -7.1640e-03],\n", + " [-1.0700e-01, 9.0514e-02, -7.8123e-02]],\n", + " \n", + " [[-1.0691e-01, 1.4074e-01, 9.2375e-02],\n", + " [ 1.0719e-01, 2.8308e-03, -6.8270e-02],\n", + " [ 4.9269e-02, -4.3296e-02, 1.5884e-02]]],\n", + " \n", + " \n", + " [[[-3.7902e-01, -1.5038e-01, -9.9616e-02],\n", + " [-2.1708e-01, -2.4423e-02, 1.2358e-01],\n", + " [-1.0269e-01, 9.7897e-02, 2.2735e-01]],\n", + " \n", + " [[ 1.4145e-01, 1.1004e-01, 5.0616e-04],\n", + " [ 1.6339e-01, 1.9676e-01, -5.9122e-02],\n", + " [ 1.1629e-01, 7.7067e-02, -1.4693e-02]],\n", + " \n", + " [[ 1.1432e-01, 1.5856e-01, 1.2212e-01],\n", + " [-9.9686e-02, -1.0319e-01, 1.1228e-01],\n", + " [-1.7612e-01, -1.4091e-01, -2.7761e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[-1.0939e-01, -1.7909e-01, 5.8055e-02],\n", + " [ 2.5802e-02, -4.4849e-02, 5.8219e-02],\n", + " [ 1.1105e-01, -1.2531e-01, -1.0372e-01]],\n", + " \n", + " [[ 9.4454e-02, -1.3725e-01, 7.0944e-02],\n", + " [ 2.0309e-01, 7.6540e-02, -6.1643e-02],\n", + " [-7.1881e-04, -2.2224e-01, 3.1247e-02]],\n", + " \n", + " [[-1.4531e-01, -2.0471e-01, -1.0648e-01],\n", + " [-1.2659e-03, -4.4630e-02, 4.6348e-02],\n", + " [ 8.4137e-02, 1.4212e-01, 1.1812e-01]]]]),\n", + " 'res3.conv2.conv2d.bias': tensor([ 0.1083, 0.0344, 0.1124, 0.0651, 0.0463, 0.0606, -0.0411, -0.0338,\n", + " -0.2189, -0.0618, 0.0031, -0.0664, -0.0336, 0.0414, -0.1968, 0.0239,\n", + " -0.0792, 0.2168, -0.1063, -0.0899, -0.0837, -0.0863, -0.1380, 0.0496,\n", + " -0.1129, 0.0265, -0.0795, 0.0579, 0.1069, 0.0957, 0.0155, 0.0970,\n", + " -0.2426, -0.2038, -0.2579, -0.1537, -0.1244, -0.1590, -0.0100, -0.1291,\n", + " 0.0794, 0.0261, 0.1904, -0.1918, -0.2471, -0.0183, 0.0297, 0.2396,\n", + " 0.0491, -0.2513, -0.1152, 0.2937, 0.0163, 0.0691, 0.0066, 0.2431,\n", + " -0.0758, 0.0101, -0.0658, 0.0331, 0.1042, 0.1169, 0.0154, -0.0981,\n", + " -0.2206, 0.0376, -0.2221, -0.1563, 0.1101, -0.3427, 0.2874, -0.1109,\n", + " -0.1594, 0.1718, 0.2667, -0.0010, 0.0397, -0.0511, 0.2653, -0.0954,\n", + " -0.0608, 0.0966, -0.1145, -0.0425, -0.0629, 0.1730, -0.0544, -0.0692,\n", + " -0.0309, 0.2505, 0.0864, -0.2531, 0.0261, -0.0034, -0.2340, 0.0631,\n", + " -0.0722, -0.0294, 0.2084, 0.2131, -0.0933, -0.0440, 0.2726, 0.0352,\n", + " -0.0375, 0.1254, -0.2115, -0.0207, 0.0457, -0.0720, -0.0167, -0.0224,\n", + " -0.1039, 0.1656, 0.1985, -0.0706, -0.1466, 0.1185, 0.0120, 0.3350,\n", + " -0.0502, 0.0342, -0.0576, 0.0013, 0.0902, 0.1960, -0.0073, -0.2768]),\n", + " 'res3.in2.weight': tensor([0.9807, 0.7718, 0.9824, 0.6939, 1.0438, 0.9285, 0.7412, 0.8949, 0.7294,\n", + " 1.0008, 0.8819, 0.8683, 0.9224, 0.9415, 0.8662, 0.9585, 0.8329, 0.9165,\n", + " 0.8397, 0.7927, 0.8467, 0.7153, 0.9315, 0.7909, 0.9717, 0.9200, 1.1091,\n", + " 0.6709, 0.9609, 1.0677, 1.0562, 0.9013, 0.9159, 0.9994, 0.7941, 0.9339,\n", + " 1.0419, 0.8851, 0.9984, 0.9432, 0.8464, 0.9677, 0.8863, 0.8560, 1.0000,\n", + " 0.9336, 1.0158, 0.9736, 0.8266, 0.9778, 0.7113, 0.7633, 1.0723, 0.8934,\n", + " 0.8084, 0.8164, 0.8406, 0.7884, 1.1219, 0.8969, 0.8177, 1.0945, 1.2450,\n", + " 0.9841, 0.9882, 0.9218, 0.8553, 0.8561, 0.6422, 0.9167, 0.7831, 0.9545,\n", + " 0.9545, 0.8020, 0.9178, 0.9069, 0.8976, 0.9115, 0.9430, 0.7387, 0.7867,\n", + " 0.8669, 0.8122, 0.8517, 0.7683, 0.9179, 1.0088, 0.7941, 0.8956, 0.9759,\n", + " 0.8482, 0.6895, 0.8308, 0.9393, 0.9674, 1.1246, 1.0103, 1.0601, 1.0080,\n", + " 0.9021, 1.0560, 0.8441, 1.0116, 1.0002, 0.9058, 0.9071, 1.0067, 0.8757,\n", + " 1.1266, 0.9651, 0.9499, 0.8828, 0.7530, 0.8820, 1.0854, 0.8902, 0.8680,\n", + " 0.7681, 0.9635, 0.8631, 0.9099, 0.8956, 0.9141, 0.9548, 0.9174, 0.7558,\n", + " 1.1098, 0.7119]),\n", + " 'res3.in2.bias': tensor([-0.3008, -0.1828, -0.1703, -0.0546, 0.1036, -0.0849, 0.1021, 0.0235,\n", + " 0.2254, -0.1168, 0.2451, 0.1257, 0.1215, 0.1946, 0.2353, -0.1158,\n", + " -0.3846, 0.1698, 0.0811, -0.0604, 0.2619, -0.0842, -0.2078, 0.1207,\n", + " 0.0011, 0.0949, 0.0735, -0.0706, 0.0195, 0.1698, 0.0197, 0.0315,\n", + " -0.0890, 0.1082, 0.0864, 0.1610, 0.0024, -0.0476, 0.0258, -0.0664,\n", + " -0.1308, -0.0330, 0.0287, -0.1459, -0.0379, 0.1079, 0.2103, -0.0237,\n", + " 0.0944, 0.1719, 0.1375, 0.1609, 0.0625, -0.0619, -0.0233, -0.2595,\n", + " -0.1126, 0.0182, -0.2356, -0.1038, 0.1241, 0.2597, 0.1134, 0.2317,\n", + " 0.1199, -0.0056, 0.1338, 0.0669, -0.1081, 0.1947, 0.3074, -0.0018,\n", + " -0.1970, 0.0042, -0.1253, -0.3827, -0.1349, 0.2408, -0.0508, 0.2744,\n", + " 0.3108, 0.0420, 0.1593, 0.4218, 0.0016, -0.0748, -0.0810, -0.0094,\n", + " 0.1282, 0.0076, 0.1364, -0.4118, 0.0462, -0.1251, -0.1061, 0.2273,\n", + " 0.2417, -0.2555, -0.2201, 0.2053, 0.2192, -0.0076, 0.0241, 0.3502,\n", + " -0.0522, 0.0444, -0.2188, -0.0649, 0.0822, -0.0246, 0.1808, 0.1575,\n", + " -0.0925, 0.1278, -0.1623, 0.4142, -0.1671, -0.1497, -0.0359, 0.1005,\n", + " 0.0399, 0.0351, 0.2640, 0.0371, 0.2290, -0.0961, 0.0007, -0.0028]),\n", + " 'res4.conv1.conv2d.weight': tensor([[[[-0.0603, 0.0410, 0.0606],\n", + " [-0.0985, -0.0810, -0.1935],\n", + " [ 0.0689, -0.0425, -0.0270]],\n", + " \n", + " [[ 0.0345, -0.0869, -0.1188],\n", + " [-0.1832, -0.0522, -0.1819],\n", + " [-0.0543, -0.0595, -0.1016]],\n", + " \n", + " [[ 0.0497, 0.0803, 0.1173],\n", + " [-0.0726, 0.0540, 0.0451],\n", + " [ 0.0967, 0.1277, 0.0809]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0436, -0.0016, 0.2045],\n", + " [-0.1578, -0.0884, 0.1520],\n", + " [-0.0895, -0.0048, 0.0414]],\n", + " \n", + " [[-0.0417, -0.0777, -0.1954],\n", + " [-0.0730, -0.1610, -0.0430],\n", + " [-0.0551, -0.1574, -0.1366]],\n", + " \n", + " [[-0.0494, -0.0716, -0.1490],\n", + " [-0.0195, 0.0482, -0.0012],\n", + " [-0.0142, -0.0201, -0.1229]]],\n", + " \n", + " \n", + " [[[-0.0676, -0.0720, -0.0661],\n", + " [ 0.0966, 0.0321, -0.0196],\n", + " [ 0.0712, -0.0017, -0.0490]],\n", + " \n", + " [[ 0.1450, 0.0400, -0.1905],\n", + " [ 0.0464, 0.0504, -0.1075],\n", + " [ 0.0207, 0.0263, -0.1013]],\n", + " \n", + " [[-0.0163, -0.1752, -0.1882],\n", + " [ 0.0731, -0.1521, -0.2315],\n", + " [ 0.0295, -0.2751, -0.3476]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0171, -0.0494, -0.0674],\n", + " [ 0.0264, 0.0033, 0.0565],\n", + " [ 0.0649, 0.1243, 0.0993]],\n", + " \n", + " [[ 0.1564, 0.1504, -0.0265],\n", + " [ 0.1450, 0.0954, -0.0306],\n", + " [ 0.1256, 0.0688, 0.0760]],\n", + " \n", + " [[ 0.0850, -0.1285, -0.1622],\n", + " [ 0.0486, -0.0849, -0.1125],\n", + " [ 0.0104, -0.1215, -0.1603]]],\n", + " \n", + " \n", + " [[[-0.0695, -0.0537, -0.0193],\n", + " [ 0.0137, 0.1030, 0.0170],\n", + " [ 0.0656, -0.0564, -0.0416]],\n", + " \n", + " [[ 0.0314, 0.0073, -0.0862],\n", + " [ 0.0546, -0.0407, -0.0578],\n", + " [-0.0656, 0.1365, 0.0883]],\n", + " \n", + " [[-0.0332, 0.0971, 0.0770],\n", + " [-0.0250, -0.0088, 0.0357],\n", + " [-0.0201, 0.0242, -0.0044]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0057, 0.1056, 0.0965],\n", + " [ 0.0271, 0.0829, 0.2095],\n", + " [-0.0876, 0.0092, 0.1845]],\n", + " \n", + " [[-0.1207, 0.0494, -0.0082],\n", + " [-0.1105, 0.0016, -0.0533],\n", + " [-0.1312, 0.0611, -0.0813]],\n", + " \n", + " [[-0.0539, 0.0508, 0.2272],\n", + " [ 0.0235, -0.0776, 0.1325],\n", + " [ 0.0517, 0.1640, -0.1039]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[ 0.1308, 0.1676, -0.0038],\n", + " [ 0.0867, 0.2636, 0.0574],\n", + " [ 0.0550, 0.1733, 0.0707]],\n", + " \n", + " [[-0.0010, -0.1490, 0.0458],\n", + " [-0.0811, -0.0230, -0.0277],\n", + " [-0.0600, 0.0089, 0.0145]],\n", + " \n", + " [[ 0.0287, 0.2343, 0.2767],\n", + " [-0.0403, 0.0976, 0.2380],\n", + " [-0.0036, 0.0780, 0.2587]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0308, -0.1870, -0.0552],\n", + " [ 0.1067, -0.1740, -0.0117],\n", + " [ 0.0257, -0.1496, 0.0411]],\n", + " \n", + " [[ 0.0676, 0.1171, 0.0798],\n", + " [-0.0395, -0.0136, 0.1185],\n", + " [-0.0122, -0.0889, 0.1299]],\n", + " \n", + " [[ 0.0414, -0.0608, 0.2135],\n", + " [-0.0293, -0.0931, 0.0312],\n", + " [-0.0409, -0.0088, 0.1265]]],\n", + " \n", + " \n", + " [[[-0.1684, -0.2467, -0.1234],\n", + " [ 0.0252, -0.0275, 0.0041],\n", + " [-0.0038, -0.0038, 0.0197]],\n", + " \n", + " [[ 0.0804, 0.0535, 0.0609],\n", + " [ 0.0479, 0.1024, 0.0900],\n", + " [ 0.0396, -0.1196, -0.0883]],\n", + " \n", + " [[ 0.2117, 0.1537, 0.0765],\n", + " [ 0.2081, 0.2021, 0.2159],\n", + " [ 0.2172, 0.1269, 0.1309]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.1996, 0.0629, 0.1322],\n", + " [ 0.0017, -0.0808, 0.1293],\n", + " [ 0.0188, -0.0112, 0.0949]],\n", + " \n", + " [[ 0.0205, -0.1078, -0.2327],\n", + " [-0.0424, -0.0835, -0.3947],\n", + " [-0.1487, 0.0083, -0.2265]],\n", + " \n", + " [[-0.0523, -0.0435, 0.0132],\n", + " [ 0.0960, 0.2249, 0.2056],\n", + " [ 0.1354, 0.0422, 0.1538]]],\n", + " \n", + " \n", + " [[[ 0.0890, 0.1936, -0.0821],\n", + " [-0.0982, -0.2045, 0.0284],\n", + " [ 0.0289, 0.1427, -0.0790]],\n", + " \n", + " [[-0.0021, 0.0467, 0.1328],\n", + " [ 0.2035, -0.0358, -0.1028],\n", + " [-0.0632, 0.0474, -0.0368]],\n", + " \n", + " [[-0.1401, -0.0215, 0.0561],\n", + " [ 0.0990, -0.2038, -0.1091],\n", + " [-0.1921, 0.0146, 0.0699]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.1815, 0.1052, 0.0323],\n", + " [ 0.1929, -0.2670, -0.1529],\n", + " [ 0.0622, 0.1235, 0.0119]],\n", + " \n", + " [[ 0.1233, 0.0327, 0.0404],\n", + " [ 0.0796, 0.0928, 0.0240],\n", + " [-0.0439, -0.1701, 0.0330]],\n", + " \n", + " [[-0.0625, -0.1402, 0.1248],\n", + " [-0.0904, -0.0620, 0.0527],\n", + " [-0.1937, -0.1066, 0.0810]]]]),\n", + " 'res4.conv1.conv2d.bias': tensor([ 0.2138, -0.1006, 0.1038, -0.1289, -0.0948, 0.0307, 0.1075, -0.1677,\n", + " -0.0278, 0.0742, -0.0570, 0.0429, 0.0302, -0.0442, 0.1025, -0.0500,\n", + " 0.2590, -0.0787, -0.2587, 0.1333, -0.0143, -0.0924, 0.0203, -0.0231,\n", + " 0.1430, 0.0749, 0.1745, -0.1262, 0.1699, -0.1105, 0.0032, -0.1924,\n", + " 0.2543, 0.0722, -0.1921, -0.0191, -0.1069, 0.0556, 0.0140, 0.1411,\n", + " 0.0679, 0.1280, 0.1984, -0.0713, -0.0007, -0.0497, -0.1540, 0.0738,\n", + " -0.1342, -0.0799, 0.1401, 0.0286, -0.0394, -0.1085, -0.2425, 0.2104,\n", + " 0.0674, -0.3254, 0.0644, 0.0816, -0.1362, 0.3302, -0.1231, 0.0873,\n", + " 0.0322, 0.0268, -0.1541, 0.1827, 0.0379, -0.0891, -0.1098, 0.0472,\n", + " -0.0295, 0.0657, -0.0279, 0.1626, -0.2516, 0.0587, -0.0328, 0.1208,\n", + " 0.0330, -0.0674, -0.0368, 0.1406, -0.0587, -0.0197, 0.0622, 0.0206,\n", + " -0.0138, -0.0833, 0.0868, 0.0025, 0.1601, -0.0488, 0.2272, 0.0698,\n", + " 0.0232, 0.0723, 0.1322, 0.0699, -0.0212, -0.1798, -0.0950, 0.1633,\n", + " 0.1977, -0.1566, -0.0843, -0.0163, 0.1351, -0.1767, -0.0817, -0.1156,\n", + " -0.0630, -0.1131, 0.0765, 0.1959, -0.1658, -0.1318, -0.1201, -0.0315,\n", + " 0.0883, 0.3614, -0.0721, 0.0460, -0.2240, -0.0870, -0.0089, 0.0250]),\n", + " 'res4.in1.weight': tensor([0.9132, 1.0350, 0.7930, 0.8961, 0.8261, 0.8464, 1.0305, 1.0595, 0.9391,\n", + " 1.0110, 1.0273, 0.8053, 0.9963, 0.9423, 0.8438, 0.8723, 0.8769, 0.8698,\n", + " 0.9363, 0.9122, 0.8767, 0.8870, 0.8573, 1.0670, 0.9079, 0.9662, 1.0815,\n", + " 1.0080, 0.7941, 0.7525, 0.8976, 0.8454, 0.9465, 0.9542, 0.8740, 0.7878,\n", + " 0.8729, 0.7725, 0.9578, 0.9357, 0.7902, 0.9844, 1.1080, 0.7830, 1.0253,\n", + " 0.9480, 0.9397, 0.8240, 0.8572, 1.0213, 0.8445, 0.8673, 0.9910, 0.9342,\n", + " 0.8601, 0.9816, 0.8084, 0.8434, 0.9946, 0.8277, 0.9046, 1.0221, 1.0298,\n", + " 0.9575, 0.8515, 0.9277, 0.9009, 0.8785, 1.0548, 0.8138, 0.8337, 0.9629,\n", + " 0.8755, 0.7712, 1.0583, 0.9847, 0.9467, 0.8955, 0.9340, 0.7516, 0.9835,\n", + " 0.9159, 1.1298, 0.7822, 0.8550, 0.9142, 0.8616, 0.8478, 0.8541, 0.8314,\n", + " 0.9424, 0.8407, 1.0386, 1.0202, 0.8303, 0.8824, 0.9208, 0.9486, 0.7172,\n", + " 0.8320, 1.0575, 1.1135, 0.8430, 0.8097, 1.0247, 1.0397, 1.0270, 0.9915,\n", + " 0.7841, 0.8343, 0.6920, 1.1096, 0.8020, 0.7358, 0.8544, 0.8611, 0.9175,\n", + " 0.9590, 1.0198, 0.7310, 0.7504, 0.8075, 0.7581, 0.9213, 0.9861, 0.7876,\n", + " 0.8835, 0.7837]),\n", + " 'res4.in1.bias': tensor([-7.8974e-01, -3.1465e-01, -6.6415e-01, -2.2593e-01, -6.4283e-02,\n", + " -5.6051e-01, -3.0837e-01, 2.4151e-01, -2.0346e-01, -4.7003e-01,\n", + " 2.5752e-01, 1.1556e-02, -3.5436e-01, -4.7069e-01, -4.8784e-01,\n", + " -6.3390e-01, -1.9159e-01, -2.0451e-01, -2.5763e-01, -1.6291e-01,\n", + " -1.9539e-01, -5.4510e-01, -4.7685e-01, -2.7659e-01, -2.9806e-01,\n", + " -2.2370e-01, -9.8331e-01, -1.3149e+00, 1.5474e-01, -1.5335e-01,\n", + " 3.6163e-02, 1.5741e-01, -5.7168e-01, -2.9197e-01, -1.3130e-01,\n", + " -6.1395e-02, -4.6124e-01, -2.7369e-01, -4.1121e-01, -3.9836e-01,\n", + " 9.6457e-03, -1.6384e-01, 1.7674e-01, -2.8483e-01, -2.6791e-01,\n", + " -4.6442e-01, -3.5655e-01, -1.4920e-01, -4.7727e-01, 3.1902e-02,\n", + " -1.2241e-01, -2.3627e-01, -2.7463e-01, -4.2591e-02, -4.7511e-01,\n", + " 1.4427e-02, -2.6116e-01, -7.5077e-03, 2.3208e-02, -4.9251e-02,\n", + " -1.9570e-01, -6.4339e-03, 2.0591e-01, -1.1789e-01, -4.4809e-01,\n", + " -1.5113e-01, -7.8852e-01, -4.8581e-01, -4.8038e-01, -2.2840e-01,\n", + " -7.3317e-01, -1.1288e-01, 5.5669e-03, -2.8745e-01, -1.5465e-02,\n", + " -6.6100e-01, -6.6001e-01, -4.4658e-01, -4.7619e-01, -6.0907e-02,\n", + " -3.0579e-01, -1.6212e-01, -1.2992e+00, -9.8765e-02, -3.8199e-01,\n", + " -4.8398e-01, -3.0344e-01, -4.1333e-01, -4.0840e-01, -4.5493e-01,\n", + " -5.0969e-01, -3.6235e-01, 4.1009e-02, -1.8625e-01, -3.0728e-01,\n", + " -6.1801e-01, -9.4602e-01, -4.8536e-01, -2.0858e-01, -4.9464e-01,\n", + " -1.3093e-01, -1.2903e-01, -1.8681e-01, -4.9900e-04, -3.5824e-01,\n", + " -1.3561e-01, -5.8696e-01, -1.0492e-01, 3.0367e-03, 7.2190e-03,\n", + " 9.0804e-02, 1.5155e-01, -3.9864e-01, -2.8239e-01, -4.6369e-01,\n", + " -3.0778e-01, -4.8966e-01, -4.7701e-01, 1.4033e-02, -3.6713e-01,\n", + " -1.9340e-01, -3.0025e-01, 1.0050e-01, -2.2447e-01, -3.0750e-01,\n", + " -3.5729e-01, -6.8048e-01, -2.8797e-02]),\n", + " 'res4.conv2.conv2d.weight': tensor([[[[ 0.1144, 0.1711, 0.1239],\n", + " [ 0.1137, 0.1048, 0.1272],\n", + " [ 0.1706, 0.1534, 0.0253]],\n", + " \n", + " [[-0.0682, -0.0521, -0.0947],\n", + " [-0.0701, -0.0252, -0.0628],\n", + " [-0.0980, -0.0465, -0.0548]],\n", + " \n", + " [[-0.0905, -0.0559, 0.0674],\n", + " [ 0.0710, -0.0117, -0.0040],\n", + " [ 0.1569, 0.0699, -0.0942]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0259, -0.0732, -0.0881],\n", + " [ 0.0555, 0.0915, -0.1851],\n", + " [ 0.0717, 0.0604, -0.0987]],\n", + " \n", + " [[ 0.1829, 0.0754, 0.0940],\n", + " [-0.0003, 0.0635, 0.1046],\n", + " [-0.0436, -0.1485, -0.1302]],\n", + " \n", + " [[-0.0331, 0.0570, 0.1484],\n", + " [-0.0893, 0.0892, -0.0350],\n", + " [ 0.1109, -0.1038, -0.0582]]],\n", + " \n", + " \n", + " [[[ 0.1094, 0.1325, 0.0665],\n", + " [-0.0053, 0.1640, 0.0384],\n", + " [ 0.1265, 0.1869, 0.0068]],\n", + " \n", + " [[-0.0096, -0.0418, 0.0240],\n", + " [ 0.0641, -0.0265, 0.0177],\n", + " [ 0.1255, 0.0574, 0.0113]],\n", + " \n", + " [[ 0.0585, 0.0469, -0.0925],\n", + " [ 0.2147, 0.0930, 0.1347],\n", + " [ 0.0485, 0.0866, 0.0608]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0325, -0.1454, 0.0046],\n", + " [ 0.0729, -0.0727, 0.1526],\n", + " [ 0.0722, 0.0369, 0.1909]],\n", + " \n", + " [[-0.1050, 0.0129, 0.1473],\n", + " [-0.1332, 0.0806, 0.0814],\n", + " [-0.0153, -0.0773, -0.0407]],\n", + " \n", + " [[ 0.0345, 0.0066, -0.0511],\n", + " [ 0.0203, -0.1577, 0.0081],\n", + " [-0.1733, -0.0548, -0.0563]]],\n", + " \n", + " \n", + " [[[-0.1126, -0.0836, -0.0907],\n", + " [-0.0061, -0.0491, -0.0937],\n", + " [-0.0557, -0.0338, -0.0322]],\n", + " \n", + " [[ 0.0639, -0.0919, -0.0964],\n", + " [ 0.1468, -0.0436, -0.1484],\n", + " [ 0.2032, 0.0587, -0.1406]],\n", + " \n", + " [[ 0.1888, 0.0345, -0.0233],\n", + " [ 0.0310, 0.0927, -0.0253],\n", + " [ 0.0758, -0.0760, 0.0151]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0464, -0.0314, 0.0698],\n", + " [-0.1443, 0.0133, 0.0643],\n", + " [-0.0925, -0.1203, 0.0881]],\n", + " \n", + " [[-0.0289, -0.0537, -0.0339],\n", + " [-0.0159, -0.0704, -0.0112],\n", + " [ 0.0684, 0.0230, 0.0585]],\n", + " \n", + " [[-0.0058, 0.2661, 0.1171],\n", + " [ 0.1158, 0.0372, -0.0716],\n", + " [-0.0561, -0.0105, 0.0173]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[-0.1248, -0.0346, -0.0863],\n", + " [-0.1086, 0.0077, 0.1325],\n", + " [-0.0512, 0.1506, 0.0731]],\n", + " \n", + " [[-0.0442, -0.0014, -0.0380],\n", + " [-0.0727, 0.0231, -0.0234],\n", + " [-0.0422, 0.0323, 0.0429]],\n", + " \n", + " [[-0.1557, -0.1497, 0.1669],\n", + " [-0.3003, 0.0170, -0.1822],\n", + " [ 0.1379, -0.1667, -0.0585]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0168, 0.1324, 0.1426],\n", + " [-0.2597, 0.0341, 0.0390],\n", + " [-0.0409, -0.1569, 0.0859]],\n", + " \n", + " [[-0.0954, -0.1377, -0.0590],\n", + " [ 0.0223, -0.1228, 0.1419],\n", + " [-0.0765, 0.1502, 0.0805]],\n", + " \n", + " [[ 0.0315, 0.0147, -0.0833],\n", + " [ 0.0979, -0.0726, -0.1016],\n", + " [ 0.0834, 0.0472, -0.0623]]],\n", + " \n", + " \n", + " [[[ 0.0385, 0.0525, -0.0273],\n", + " [ 0.0624, -0.0118, -0.0888],\n", + " [-0.0082, -0.0752, -0.1344]],\n", + " \n", + " [[-0.0455, -0.1372, -0.2229],\n", + " [-0.0227, -0.1126, -0.2249],\n", + " [-0.0227, -0.1027, -0.1945]],\n", + " \n", + " [[ 0.0917, 0.0927, -0.0348],\n", + " [ 0.0522, 0.0877, 0.0674],\n", + " [ 0.0724, 0.0583, 0.0329]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0048, -0.0747, -0.0756],\n", + " [ 0.0655, -0.0821, 0.0555],\n", + " [ 0.1136, -0.0688, 0.0871]],\n", + " \n", + " [[ 0.0275, -0.0446, 0.0342],\n", + " [-0.0406, 0.1320, 0.0331],\n", + " [-0.1243, 0.0095, 0.0383]],\n", + " \n", + " [[-0.0375, 0.0785, -0.0845],\n", + " [-0.0472, -0.0190, -0.0044],\n", + " [ 0.0139, 0.0130, 0.0493]]],\n", + " \n", + " \n", + " [[[-0.0119, -0.0635, 0.0941],\n", + " [ 0.0578, 0.0196, 0.0051],\n", + " [-0.0712, 0.1801, -0.0301]],\n", + " \n", + " [[-0.0244, 0.0302, 0.0969],\n", + " [-0.0826, -0.0003, 0.1097],\n", + " [-0.1986, -0.1098, 0.0690]],\n", + " \n", + " [[-0.0036, -0.1297, 0.0166],\n", + " [-0.0424, -0.0707, 0.0100],\n", + " [ 0.0744, -0.0967, 0.0646]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.1265, -0.0941, -0.1730],\n", + " [-0.0012, -0.1663, -0.1541],\n", + " [-0.0482, -0.0530, -0.1251]],\n", + " \n", + " [[-0.0093, 0.1337, 0.1138],\n", + " [ 0.0157, -0.0548, -0.0227],\n", + " [-0.0369, -0.0019, 0.0602]],\n", + " \n", + " [[ 0.1285, 0.1386, 0.0726],\n", + " [ 0.0617, 0.0336, 0.0724],\n", + " [ 0.0414, 0.0963, 0.1229]]]]),\n", + " 'res4.conv2.conv2d.bias': tensor([ 2.1868e-01, -2.3774e-01, -2.5003e-01, -1.0501e-01, -1.6379e-01,\n", + " 1.4751e-01, -1.5693e-01, 7.0750e-02, -9.2839e-02, -6.3253e-02,\n", + " -1.4480e-01, 8.3915e-02, 5.0666e-02, -1.5166e-01, 1.0290e-01,\n", + " -1.0335e-01, -9.8686e-06, 1.4720e-01, 3.3836e-02, -2.1180e-01,\n", + " 1.2515e-01, -3.5596e-01, 8.9397e-02, 1.8768e-01, 9.6856e-02,\n", + " -8.9703e-03, -4.3982e-02, 1.2791e-01, 4.3850e-02, -5.5051e-02,\n", + " 2.0758e-01, -6.1616e-02, -9.3818e-02, -2.6872e-02, 6.2629e-02,\n", + " 1.6888e-01, 1.2509e-01, -2.5762e-02, -7.7452e-02, 2.2232e-01,\n", + " 1.5303e-01, -2.6998e-02, 4.1397e-02, 5.8782e-02, -5.6055e-02,\n", + " 9.3836e-02, 4.1524e-02, 1.4422e-01, 2.0259e-01, -4.9366e-02,\n", + " 4.4448e-02, 2.1634e-01, -4.4901e-02, 3.0780e-02, 1.4056e-01,\n", + " -2.2267e-01, 1.0212e-01, -1.3899e-02, -2.0356e-01, 7.4653e-02,\n", + " -5.6952e-02, 3.2324e-02, -1.1943e-01, 2.7814e-02, 7.7670e-02,\n", + " -9.3135e-02, 6.3520e-03, -1.6425e-01, 1.7618e-01, 2.4734e-01,\n", + " -1.2148e-02, -5.0429e-02, 3.0284e-01, -1.0841e-01, 2.0505e-02,\n", + " 6.1372e-02, -5.0814e-02, -1.6893e-02, 1.1255e-01, 5.6455e-02,\n", + " 5.8222e-02, -3.9471e-02, 9.4910e-02, 1.9040e-01, -1.6633e-01,\n", + " 2.1785e-02, 4.4189e-02, -8.7576e-02, 2.1436e-01, 1.5465e-01,\n", + " -1.9321e-01, 2.0013e-01, -9.7036e-02, 2.1588e-01, -5.6524e-02,\n", + " -1.2121e-01, 1.6933e-01, 8.3568e-02, -2.2663e-01, 4.1645e-02,\n", + " -6.0279e-02, 3.0151e-02, 2.8905e-01, -7.1027e-02, 1.7761e-02,\n", + " 1.8356e-01, 2.2203e-01, 1.4250e-01, -1.4153e-01, -1.7136e-01,\n", + " -1.3191e-01, -4.6973e-02, -2.5156e-02, -6.9439e-02, -3.3135e-02,\n", + " 7.7741e-02, 1.1933e-01, 1.0338e-01, -1.0786e-01, 1.4212e-02,\n", + " 1.0450e-01, -2.8758e-02, -8.2726e-02, -3.1254e-01, -7.7462e-02,\n", + " -1.4204e-02, -1.1961e-01, 7.7052e-02]),\n", + " 'res4.in2.weight': tensor([1.0928, 0.9725, 1.0602, 1.2067, 1.2930, 0.9388, 1.1911, 0.9966, 0.8497,\n", + " 0.9756, 1.2857, 1.1767, 1.0402, 0.9065, 1.0436, 1.0146, 1.0727, 1.0635,\n", + " 1.4121, 1.1513, 0.9846, 0.9381, 1.1464, 1.0167, 1.0085, 1.2724, 0.9456,\n", + " 0.9988, 1.1147, 1.1389, 1.0657, 1.1056, 1.1976, 0.9243, 0.7957, 1.0279,\n", + " 1.0200, 1.0084, 1.3275, 1.1229, 0.8371, 1.0449, 0.8926, 1.1834, 0.9072,\n", + " 1.2217, 0.9966, 1.0788, 0.9945, 0.9489, 0.8685, 0.8391, 0.8168, 0.9471,\n", + " 1.3837, 1.1091, 1.1895, 1.0461, 1.1324, 1.0119, 1.0353, 0.9313, 1.1704,\n", + " 1.0453, 1.0597, 0.7837, 1.0386, 1.0457, 0.9406, 1.0199, 1.1182, 1.1233,\n", + " 1.0364, 1.0384, 1.0447, 1.1456, 1.0476, 1.0525, 1.2928, 1.0734, 1.0834,\n", + " 1.1251, 1.0449, 1.0171, 1.0487, 0.9541, 0.8729, 1.0371, 1.0668, 1.0232,\n", + " 0.8239, 0.9734, 0.9768, 0.9437, 1.2728, 0.9210, 1.5454, 1.1437, 1.1067,\n", + " 1.0037, 1.1267, 1.1523, 1.1863, 0.9733, 0.8615, 1.0412, 1.0760, 0.9003,\n", + " 1.1645, 1.1805, 1.1557, 0.8678, 1.0172, 0.8218, 1.0786, 1.0114, 0.9397,\n", + " 0.9246, 1.1430, 1.0357, 1.0707, 1.0179, 1.1333, 1.0593, 1.2382, 0.8557,\n", + " 1.0077, 0.8916]),\n", + " 'res4.in2.bias': tensor([-2.1987e-01, -1.1996e-01, -2.7755e-01, -1.3667e-01, 1.8001e-01,\n", + " -1.5257e-01, 4.6085e-02, -1.9533e-02, 1.3018e-01, 2.7695e-02,\n", + " 2.1758e-01, 5.3916e-02, 2.5095e-01, 1.9328e-01, 9.5427e-02,\n", + " -1.2520e-01, -3.5298e-01, 1.8403e-01, 3.5003e-02, -8.4514e-02,\n", + " 1.8459e-01, 6.0320e-02, -3.1223e-01, -7.9493e-04, 1.7728e-02,\n", + " 2.7800e-02, -2.4330e-02, 2.5298e-03, 1.0903e-01, 9.8363e-02,\n", + " 1.3371e-02, -1.9750e-03, -2.3167e-01, 8.0130e-02, 1.2087e-02,\n", + " 1.7585e-01, 3.6889e-02, 1.0725e-02, -4.7857e-02, -7.5621e-02,\n", + " -1.9550e-01, 4.6517e-02, 2.0033e-02, -5.8428e-02, -9.4094e-02,\n", + " 4.6097e-02, 6.0073e-02, -6.6983e-02, 1.7378e-01, 1.1388e-01,\n", + " 1.3710e-01, -2.6752e-02, -3.2889e-03, -3.5461e-02, -8.6975e-02,\n", + " -1.7633e-01, -2.9299e-02, 7.4094e-02, -2.4311e-01, -1.2373e-01,\n", + " -7.7355e-04, 1.3536e-01, 1.2612e-01, 2.2187e-01, 1.0655e-01,\n", + " -1.2713e-01, 1.3038e-01, -6.1913e-02, -1.4248e-01, 2.5916e-01,\n", + " 2.9709e-01, -7.5170e-02, -1.7093e-02, -6.2882e-03, 1.3789e-02,\n", + " -3.7273e-01, -3.5688e-02, 1.3671e-01, -7.8031e-02, 2.0508e-01,\n", + " 3.5838e-01, -8.8170e-02, 1.5686e-02, 2.5902e-01, 3.5436e-02,\n", + " -3.8512e-02, -7.5531e-02, 8.2945e-02, 5.6526e-02, 1.6616e-01,\n", + " 3.6786e-02, -3.2084e-01, -1.3351e-02, -1.0194e-01, -8.4298e-02,\n", + " 1.8275e-01, 2.3885e-01, -2.5497e-01, -1.3487e-01, 1.7990e-01,\n", + " 8.5742e-02, 3.3304e-02, -1.0207e-01, 2.7566e-01, 6.7126e-02,\n", + " -1.6918e-02, -1.9922e-01, -5.4585e-02, 1.1258e-01, -3.0163e-02,\n", + " 9.8076e-02, -4.9950e-02, -7.6504e-02, 2.0575e-03, -8.1882e-02,\n", + " 2.9037e-01, -5.9135e-02, -2.4035e-01, -3.0977e-02, 5.9266e-02,\n", + " -6.2665e-02, -4.8456e-02, 1.4344e-01, 6.5920e-02, 2.3405e-01,\n", + " -9.8571e-02, -7.9812e-05, -3.6139e-03]),\n", + " 'res5.conv1.conv2d.weight': tensor([[[[ 0.1497, -0.0485, -0.0422],\n", + " [ 0.1031, -0.0646, 0.0606],\n", + " [-0.0085, 0.0170, 0.0383]],\n", + " \n", + " [[ 0.0297, 0.1259, 0.2383],\n", + " [-0.0329, -0.0237, 0.1544],\n", + " [ 0.0757, -0.0301, 0.0910]],\n", + " \n", + " [[-0.0042, 0.0692, 0.1077],\n", + " [-0.0959, -0.0643, -0.0293],\n", + " [-0.0302, 0.0509, 0.1082]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0384, 0.0187, -0.0725],\n", + " [ 0.1359, -0.1097, 0.0154],\n", + " [ 0.1615, 0.1061, 0.0296]],\n", + " \n", + " [[-0.1625, 0.0186, -0.1539],\n", + " [-0.0082, 0.1475, -0.0586],\n", + " [ 0.0316, 0.2062, 0.0110]],\n", + " \n", + " [[ 0.1104, 0.0874, 0.1613],\n", + " [ 0.0362, 0.0531, 0.1865],\n", + " [ 0.0177, 0.0089, 0.0687]]],\n", + " \n", + " \n", + " [[[ 0.0593, 0.0538, 0.0545],\n", + " [-0.0035, 0.0243, 0.0297],\n", + " [-0.1221, -0.0440, -0.0490]],\n", + " \n", + " [[-0.1212, -0.1158, -0.0689],\n", + " [-0.0462, -0.0985, -0.0994],\n", + " [-0.1207, -0.1730, -0.1919]],\n", + " \n", + " [[ 0.0971, 0.1131, 0.1040],\n", + " [ 0.1374, 0.1333, 0.2066],\n", + " [ 0.0663, -0.0434, 0.0830]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0377, 0.0253, 0.1514],\n", + " [-0.0717, -0.0874, 0.0842],\n", + " [ 0.0108, -0.1371, 0.0691]],\n", + " \n", + " [[ 0.1161, -0.0294, -0.0250],\n", + " [ 0.0364, -0.0022, 0.0204],\n", + " [ 0.1080, 0.0635, 0.0092]],\n", + " \n", + " [[-0.0680, -0.1211, 0.0262],\n", + " [ 0.0731, 0.0223, 0.1321],\n", + " [-0.0312, -0.0561, 0.1026]]],\n", + " \n", + " \n", + " [[[ 0.1841, 0.0379, 0.0728],\n", + " [-0.1259, -0.0482, 0.0620],\n", + " [-0.0788, 0.0081, -0.0027]],\n", + " \n", + " [[ 0.0701, 0.1364, 0.0117],\n", + " [-0.1286, -0.0650, 0.0568],\n", + " [-0.0983, -0.0290, -0.1767]],\n", + " \n", + " [[-0.0308, -0.2448, -0.0642],\n", + " [ 0.0349, -0.1007, -0.0178],\n", + " [ 0.2285, 0.1252, 0.1601]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0380, -0.0466, -0.1005],\n", + " [ 0.1221, -0.0284, 0.0272],\n", + " [ 0.1415, 0.0795, -0.0218]],\n", + " \n", + " [[ 0.1206, 0.1266, 0.0237],\n", + " [-0.1854, -0.0097, 0.1052],\n", + " [-0.1412, -0.1378, -0.0545]],\n", + " \n", + " [[ 0.0350, 0.0122, 0.1082],\n", + " [ 0.0521, 0.1387, 0.1164],\n", + " [ 0.0582, 0.1060, 0.1362]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[-0.0879, 0.0074, -0.0930],\n", + " [ 0.0971, 0.0157, 0.0073],\n", + " [ 0.1529, 0.1674, 0.0247]],\n", + " \n", + " [[ 0.0118, -0.0078, 0.0941],\n", + " [ 0.0128, -0.0539, -0.0833],\n", + " [ 0.1170, 0.0515, 0.0850]],\n", + " \n", + " [[ 0.1357, 0.0452, 0.2546],\n", + " [ 0.0290, -0.0004, 0.1520],\n", + " [ 0.0638, 0.0787, 0.2606]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.1589, 0.1387, 0.0077],\n", + " [ 0.0414, -0.0028, 0.0729],\n", + " [ 0.0450, -0.1621, 0.0103]],\n", + " \n", + " [[-0.0645, 0.0059, 0.1126],\n", + " [ 0.0326, 0.0344, 0.0768],\n", + " [ 0.1180, 0.0557, 0.1111]],\n", + " \n", + " [[ 0.0824, 0.0410, 0.0777],\n", + " [-0.0606, -0.0149, 0.0489],\n", + " [-0.1106, 0.0147, 0.1685]]],\n", + " \n", + " \n", + " [[[-0.0101, 0.0489, 0.1117],\n", + " [ 0.2809, 0.0236, -0.1037],\n", + " [-0.1490, 0.1291, 0.2597]],\n", + " \n", + " [[ 0.0613, 0.0329, 0.0334],\n", + " [ 0.0256, 0.2373, -0.0572],\n", + " [-0.1908, 0.0207, 0.0670]],\n", + " \n", + " [[-0.1722, -0.2452, 0.0414],\n", + " [ 0.1116, 0.0431, -0.0694],\n", + " [ 0.0651, -0.0262, -0.0169]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0390, -0.0471, 0.0867],\n", + " [-0.1411, 0.1293, -0.0441],\n", + " [ 0.1524, -0.1699, -0.0227]],\n", + " \n", + " [[-0.0037, 0.1237, 0.0181],\n", + " [-0.1650, -0.1294, -0.0041],\n", + " [-0.1595, -0.0245, -0.1342]],\n", + " \n", + " [[ 0.0010, 0.0697, -0.0865],\n", + " [-0.1534, 0.2002, 0.0978],\n", + " [ 0.0953, -0.0794, 0.1040]]],\n", + " \n", + " \n", + " [[[ 0.0500, 0.0806, -0.0627],\n", + " [-0.0482, -0.0336, 0.0266],\n", + " [-0.0649, 0.0498, 0.0642]],\n", + " \n", + " [[-0.0506, 0.0130, -0.0163],\n", + " [ 0.0261, 0.0120, 0.0052],\n", + " [-0.0130, 0.0174, 0.0325]],\n", + " \n", + " [[ 0.0549, -0.0160, -0.0495],\n", + " [-0.0784, -0.0082, -0.0134],\n", + " [-0.0514, -0.0132, -0.0722]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0682, 0.0993, -0.0380],\n", + " [-0.0072, 0.1701, 0.0194],\n", + " [-0.1082, 0.0750, -0.1863]],\n", + " \n", + " [[-0.1142, -0.0592, -0.1043],\n", + " [-0.2797, -0.1796, -0.0360],\n", + " [-0.2874, -0.2096, -0.1601]],\n", + " \n", + " [[-0.0742, 0.0284, 0.0604],\n", + " [ 0.0319, 0.1342, 0.0244],\n", + " [ 0.1900, 0.1673, 0.0474]]]]),\n", + " 'res5.conv1.conv2d.bias': tensor([ 0.0975, -0.1063, -0.0552, -0.2342, 0.1280, 0.1904, -0.1254, 0.0428,\n", + " -0.0067, 0.0784, -0.1138, 0.1245, 0.0892, 0.0685, -0.0812, -0.0235,\n", + " 0.3161, -0.0936, -0.0874, 0.0925, 0.0135, 0.0683, 0.0640, 0.0104,\n", + " -0.0146, 0.0145, -0.0226, 0.0368, -0.1899, 0.0257, -0.1321, -0.1139,\n", + " 0.4685, -0.0510, 0.0460, 0.1339, -0.2061, 0.0132, 0.0811, -0.1034,\n", + " 0.0731, 0.0472, 0.1575, 0.1255, 0.0753, -0.0894, -0.0281, 0.0200,\n", + " 0.2396, -0.1223, 0.1120, 0.1937, -0.0532, -0.0900, 0.0454, 0.0455,\n", + " 0.0536, 0.1517, -0.0415, 0.1079, -0.0375, 0.1390, -0.0404, 0.2141,\n", + " 0.0978, 0.1668, -0.0857, -0.1006, -0.0960, 0.1088, -0.0152, 0.1428,\n", + " 0.0710, -0.1000, 0.2503, 0.1392, -0.1800, -0.0880, -0.0299, -0.0237,\n", + " 0.0677, -0.0278, 0.1236, -0.0598, 0.1536, 0.0456, -0.1412, -0.0523,\n", + " -0.0637, -0.1652, -0.2016, -0.0416, 0.1714, 0.1428, 0.0183, 0.0617,\n", + " 0.0987, -0.1384, -0.0132, 0.2158, 0.1386, -0.1154, 0.0578, 0.0081,\n", + " -0.1293, -0.0301, -0.2198, 0.0189, 0.0534, 0.0239, 0.3195, -0.0757,\n", + " -0.2069, 0.2919, 0.1811, -0.1450, -0.1362, 0.0046, 0.1313, -0.1825,\n", + " 0.1456, -0.0315, -0.1155, -0.1170, -0.0209, -0.0664, -0.0580, 0.1148]),\n", + " 'res5.in1.weight': tensor([0.9105, 0.8891, 0.9172, 0.9010, 0.7742, 1.0159, 0.8422, 1.0576, 0.8667,\n", + " 1.0234, 0.8899, 0.7961, 0.9635, 0.9723, 0.9432, 0.9406, 1.0680, 1.0246,\n", + " 1.0256, 0.9505, 1.0361, 0.8681, 0.9961, 0.9207, 0.9726, 0.8807, 0.8783,\n", + " 1.0245, 0.9810, 1.0342, 0.8267, 1.0847, 1.0510, 0.7651, 0.9732, 0.9573,\n", + " 0.9696, 1.0485, 0.9372, 0.9105, 0.9884, 1.0138, 1.0093, 1.0015, 0.9275,\n", + " 0.9290, 0.8801, 0.9825, 0.8630, 1.0340, 0.8781, 0.9369, 0.7593, 1.0792,\n", + " 0.9494, 0.9590, 0.8981, 0.9383, 0.8386, 1.0399, 0.8869, 0.8246, 0.9914,\n", + " 0.9628, 1.0170, 0.8212, 1.0089, 1.0657, 0.9189, 0.8719, 0.9845, 0.7620,\n", + " 0.9712, 1.0427, 1.0275, 0.9036, 1.0655, 1.0676, 0.9538, 0.9186, 1.0528,\n", + " 0.9321, 0.8853, 0.8784, 0.8363, 1.0169, 0.9714, 1.0055, 1.0345, 0.9179,\n", + " 0.8389, 0.9655, 1.0731, 0.9526, 0.8830, 0.8735, 0.8292, 0.9473, 0.8312,\n", + " 1.1299, 1.1231, 0.8642, 1.0147, 0.8299, 1.0186, 0.8957, 0.9711, 0.9175,\n", + " 1.0343, 0.7942, 1.0407, 0.9432, 0.9335, 0.9391, 0.9243, 0.9211, 1.0421,\n", + " 0.9095, 1.0800, 0.9388, 0.9252, 0.8624, 0.9051, 1.0106, 0.9748, 0.9619,\n", + " 0.9086, 1.0272]),\n", + " 'res5.in1.bias': tensor([-1.1276e-01, -2.2821e-01, -4.1278e-01, -4.1976e-01, 1.5072e-02,\n", + " -9.1067e-02, -2.4177e-01, -1.6506e-01, -2.1493e-01, -1.3746e-01,\n", + " -5.5018e-01, -3.4816e-01, -1.3112e-01, -2.8190e-01, -3.3469e-01,\n", + " 6.9852e-02, -2.0382e-01, -7.2904e-02, -4.7958e-01, -2.7999e-02,\n", + " -1.1905e-01, -4.6233e-01, -1.9651e-01, -1.9468e-01, -4.9792e-01,\n", + " -1.0292e-01, -1.6671e-01, -4.4316e-01, -7.8815e-02, 1.6966e-01,\n", + " 3.1691e-01, -1.5052e-01, 8.3080e-03, 5.9455e-03, -2.7360e-01,\n", + " -3.2051e-01, -1.2319e-01, -5.1170e-02, -9.0688e-02, -4.1057e-01,\n", + " 2.9393e-03, -2.0905e-01, -8.9482e-04, -3.1948e-01, -9.5748e-02,\n", + " -2.5678e-01, -5.7448e-01, -1.7210e-01, -3.7475e-01, -3.7897e-01,\n", + " -2.3035e-01, -3.1518e-01, 9.8196e-02, -2.5523e-01, -1.3790e-01,\n", + " -1.4951e-01, -1.7529e-02, -1.3103e-01, -2.0883e-01, -3.4232e-01,\n", + " -5.0471e-01, -2.0050e-01, -1.1361e+00, -5.9761e-01, -1.2953e-01,\n", + " -1.6340e-01, -3.9104e-02, 2.1731e-02, -7.3480e-02, -2.5612e-01,\n", + " -1.8789e-01, -3.8581e-01, -1.7138e-01, 7.3673e-02, -4.4502e-01,\n", + " -2.7481e-01, -1.4830e-01, -3.1053e-01, -3.3897e-01, -2.8116e-01,\n", + " -2.2930e-01, -1.9837e-01, 1.4789e-01, -2.5408e-01, -2.6871e-01,\n", + " 1.8476e-01, -2.8801e-01, -1.3169e-01, -1.5295e-01, -1.4171e-01,\n", + " -2.1653e-01, -4.1082e-02, -5.5823e-02, -1.8656e-01, -4.0274e-01,\n", + " -2.0226e-01, -3.4199e-01, 2.8629e-02, -3.7990e-01, 2.4943e-01,\n", + " 1.1146e-01, -1.4150e-01, -4.7382e-01, -3.5612e-01, -4.3581e-01,\n", + " -2.7769e-01, -2.8862e-02, -1.0752e-01, -5.7473e-01, -4.3865e-01,\n", + " -1.8146e-01, -4.1974e-02, -1.7978e-01, -8.3879e-02, -1.1708e-01,\n", + " -1.8915e-01, -6.5724e-02, -3.0576e-01, -1.1370e-01, -4.7036e-01,\n", + " -1.8508e-01, -3.3302e-01, -2.2694e-01, -3.5280e-01, -3.4376e-01,\n", + " -2.3032e-01, -2.2713e-01, -1.5740e-01]),\n", + " 'res5.conv2.conv2d.weight': tensor([[[[-1.1258e-01, 1.8926e-01, 2.1939e-01],\n", + " [-6.5452e-02, -1.5452e-02, 3.0658e-02],\n", + " [-1.1323e-01, -2.1200e-01, -2.7608e-01]],\n", + " \n", + " [[ 2.8932e-02, -4.9939e-02, 6.2161e-02],\n", + " [ 5.1844e-02, -6.3734e-02, 1.6338e-02],\n", + " [ 5.1675e-02, 3.5503e-03, 2.4636e-02]],\n", + " \n", + " [[-2.7239e-02, 5.9370e-02, 2.4674e-03],\n", + " [ 1.0191e-01, 1.3740e-01, -8.4796e-02],\n", + " [ 9.0092e-02, 1.0403e-02, -1.1046e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[ 5.6042e-02, 6.3660e-03, 2.6939e-02],\n", + " [ 6.9044e-03, -6.3799e-02, -2.1449e-02],\n", + " [ 6.3264e-02, 5.2505e-02, 1.6944e-01]],\n", + " \n", + " [[-1.2313e-01, -9.9095e-02, 8.3774e-02],\n", + " [ 7.5081e-02, 1.7479e-02, -2.2470e-01],\n", + " [ 5.6162e-03, -1.8715e-01, -3.6923e-02]],\n", + " \n", + " [[ 1.3085e-02, -1.1160e-02, 2.0811e-02],\n", + " [ 4.9260e-02, 2.4341e-03, 2.1013e-02],\n", + " [-9.0254e-02, -1.1770e-02, -1.4888e-01]]],\n", + " \n", + " \n", + " [[[ 4.8852e-02, -4.6847e-02, -1.6551e-02],\n", + " [-1.1348e-01, -4.0781e-02, -9.5942e-02],\n", + " [-7.3796e-02, -1.1241e-01, 4.0000e-02]],\n", + " \n", + " [[ 2.1469e-03, 1.2651e-02, 2.3102e-02],\n", + " [ 1.7362e-01, 1.4920e-01, 9.0281e-02],\n", + " [ 3.2078e-02, -1.0752e-02, -6.1617e-03]],\n", + " \n", + " [[ 6.3447e-02, 1.4091e-01, 1.1906e-01],\n", + " [-9.2516e-02, -2.1015e-01, 7.9315e-02],\n", + " [-1.0749e-01, -5.2522e-02, 1.3510e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[ 2.1800e-02, -1.2169e-01, -1.3495e-01],\n", + " [ 1.4844e-02, 6.1664e-02, 7.4860e-02],\n", + " [ 8.3647e-02, 1.8926e-01, 1.0273e-01]],\n", + " \n", + " [[ 4.5791e-02, 7.5916e-02, -1.7618e-02],\n", + " [ 4.7607e-02, -2.5137e-01, -6.4249e-02],\n", + " [ 4.8016e-02, 1.3564e-01, -5.5259e-02]],\n", + " \n", + " [[ 2.7397e-02, 8.9506e-02, 1.3309e-01],\n", + " [-1.6431e-02, -1.2000e-01, -8.3625e-03],\n", + " [ 1.0706e-01, 1.3362e-01, 4.5475e-02]]],\n", + " \n", + " \n", + " [[[-1.1256e-01, -1.8189e-01, -1.7307e-01],\n", + " [ 2.3487e-02, 4.0728e-02, -2.2292e-02],\n", + " [ 8.8698e-03, 2.8883e-04, -3.9491e-02]],\n", + " \n", + " [[-8.4035e-03, -6.8641e-02, -3.8875e-02],\n", + " [-1.7779e-01, -1.1572e-01, -1.2614e-01],\n", + " [-8.6486e-02, -1.3366e-01, -1.0242e-01]],\n", + " \n", + " [[-4.7217e-02, 6.2274e-03, 9.9785e-03],\n", + " [-9.1322e-02, 5.7652e-02, 2.2170e-02],\n", + " [-9.6496e-02, 1.2082e-02, -9.7870e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[-1.2841e-01, -1.2115e-01, -1.2286e-01],\n", + " [-1.0762e-01, -3.5285e-02, -1.3253e-01],\n", + " [-8.8942e-02, -3.5659e-02, -7.3297e-02]],\n", + " \n", + " [[ 4.7532e-02, 2.0160e-03, -1.4784e-01],\n", + " [ 8.6376e-04, -2.9197e-02, 7.0961e-03],\n", + " [ 9.8280e-02, 8.6355e-02, 3.2334e-02]],\n", + " \n", + " [[-5.6197e-02, 1.4437e-02, 4.7333e-02],\n", + " [-1.6208e-02, -2.0676e-02, 2.1752e-02],\n", + " [-8.8742e-02, -1.1375e-01, -1.0689e-01]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[-5.2641e-02, -7.0825e-03, -1.5740e-01],\n", + " [ 4.4263e-02, -4.2839e-02, -2.6836e-02],\n", + " [ 1.1101e-01, 3.8058e-02, -1.0893e-01]],\n", + " \n", + " [[-1.3123e-01, 6.8032e-02, -1.8249e-02],\n", + " [-1.1192e-01, 5.4630e-02, 2.3104e-02],\n", + " [-1.0022e-01, 2.0534e-02, 5.8267e-02]],\n", + " \n", + " [[-9.6796e-02, -8.9647e-02, -1.6216e-01],\n", + " [ 5.3962e-03, -1.3253e-02, -2.1710e-02],\n", + " [ 1.4507e-01, 2.4262e-01, 4.1603e-03]],\n", + " \n", + " ...,\n", + " \n", + " [[-3.6778e-02, 1.4039e-02, -3.5288e-02],\n", + " [-2.1010e-01, 1.3462e-01, -7.6285e-02],\n", + " [-6.6927e-02, -2.6431e-02, 7.6774e-03]],\n", + " \n", + " [[-1.4710e-02, 8.5993e-02, -1.3376e-01],\n", + " [-2.6065e-02, -2.5235e-01, 1.1854e-01],\n", + " [-5.8822e-02, 5.3471e-02, -5.4296e-02]],\n", + " \n", + " [[ 3.5981e-02, -1.3440e-01, -2.7591e-01],\n", + " [ 3.3236e-02, -2.2145e-01, -3.1565e-01],\n", + " [-6.1641e-02, -1.9711e-01, -1.5434e-01]]],\n", + " \n", + " \n", + " [[[-1.0675e-01, -4.3466e-02, -8.1393e-02],\n", + " [-6.6610e-02, -1.7445e-02, -8.7192e-02],\n", + " [-9.0055e-02, -6.0648e-02, -1.0110e-01]],\n", + " \n", + " [[ 8.5274e-02, 5.0048e-02, 3.5418e-02],\n", + " [-4.7481e-02, -4.4621e-02, -4.4108e-02],\n", + " [-2.6946e-02, -5.7325e-02, -3.5740e-02]],\n", + " \n", + " [[-7.8864e-03, -1.3265e-01, -1.6295e-01],\n", + " [-1.6522e-03, -6.4357e-02, -6.5443e-02],\n", + " [-5.1132e-02, -8.7688e-02, -1.3786e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[ 6.4589e-02, 5.9859e-03, 1.6947e-01],\n", + " [-4.2377e-02, -1.4142e-01, 5.0437e-02],\n", + " [ 4.6514e-02, 5.1288e-02, 2.1415e-01]],\n", + " \n", + " [[-2.8740e-02, 1.4892e-04, -9.6771e-02],\n", + " [-6.3713e-02, 6.9677e-04, -2.3969e-02],\n", + " [-4.6390e-02, -7.1786e-02, -5.4185e-02]],\n", + " \n", + " [[ 1.8604e-01, 2.7153e-01, 2.5050e-01],\n", + " [ 1.7946e-01, 2.2380e-01, 2.2644e-01],\n", + " [ 1.0978e-01, 8.3211e-02, 8.4003e-02]]],\n", + " \n", + " \n", + " [[[ 5.7402e-02, 8.3387e-02, -2.0938e-02],\n", + " [-5.5700e-02, -6.5512e-02, -3.1942e-02],\n", + " [ 2.8187e-02, 1.4586e-01, 1.7195e-01]],\n", + " \n", + " [[-5.4430e-02, -7.3594e-02, -1.0204e-01],\n", + " [-1.1361e-01, -1.0786e-01, -1.4435e-01],\n", + " [-8.7002e-02, -9.9509e-03, -9.6679e-02]],\n", + " \n", + " [[-3.2413e-02, 8.8038e-02, -1.0923e-02],\n", + " [-7.0665e-02, -9.0624e-02, -2.0977e-01],\n", + " [ 4.3102e-02, 2.1430e-02, -1.0736e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[-5.2192e-02, 1.2107e-01, 1.7081e-01],\n", + " [ 2.2189e-01, 1.0084e-01, 1.4915e-01],\n", + " [ 1.1905e-01, 4.0403e-02, -6.4328e-02]],\n", + " \n", + " [[-1.2837e-01, 4.7181e-02, -1.5476e-02],\n", + " [ 7.0383e-02, -1.6827e-02, 5.2418e-02],\n", + " [ 1.0580e-01, 1.0931e-01, 1.8803e-02]],\n", + " \n", + " [[-2.0055e-02, -4.9056e-02, 1.3585e-01],\n", + " [-9.9666e-02, -9.5586e-02, 4.7421e-02],\n", + " [ 7.9093e-02, 1.1410e-01, 2.3538e-01]]]]),\n", + " 'res5.conv2.conv2d.bias': tensor([-0.2420, 0.0847, 0.0181, -0.0509, 0.2133, 0.0331, 0.0660, -0.0284,\n", + " -0.0828, 0.1394, 0.0541, 0.0076, 0.0963, 0.1306, -0.2814, 0.0150,\n", + " 0.0045, 0.1431, -0.1203, 0.0743, 0.0617, 0.0278, -0.1506, -0.0065,\n", + " 0.0094, 0.0009, 0.0053, -0.1278, 0.0612, -0.0560, -0.1502, -0.1588,\n", + " -0.0473, -0.1532, 0.1224, 0.2504, 0.0676, 0.0373, 0.1680, -0.0549,\n", + " 0.0328, -0.0119, 0.2394, -0.1817, 0.4952, -0.0589, 0.0750, 0.1576,\n", + " 0.1058, 0.1073, 0.1487, -0.0844, 0.1677, -0.1663, -0.0638, -0.0894,\n", + " 0.0534, -0.1931, -0.0111, -0.0101, 0.0373, -0.0852, 0.1715, 0.0507,\n", + " -0.0111, 0.0316, 0.0034, 0.0653, 0.1634, 0.1688, -0.2819, 0.1935,\n", + " 0.0884, -0.0579, -0.2298, -0.1241, 0.0451, 0.0465, -0.0688, 0.0424,\n", + " -0.1026, 0.0440, -0.1010, -0.1090, 0.0681, -0.0039, -0.1813, 0.1672,\n", + " -0.0856, -0.2648, 0.1920, 0.0601, -0.0992, 0.1019, -0.0569, 0.3002,\n", + " -0.1250, -0.2872, -0.0678, -0.0288, -0.0246, 0.2779, 0.1023, 0.0006,\n", + " 0.3165, 0.0361, 0.1798, -0.0592, 0.1190, -0.1068, -0.0288, -0.0375,\n", + " -0.0848, -0.0246, -0.0212, -0.0545, 0.1120, 0.0394, -0.1108, -0.0070,\n", + " 0.0082, 0.0398, 0.0269, -0.2069, 0.0711, 0.1666, 0.1282, 0.1355]),\n", + " 'res5.in2.weight': tensor([1.3993, 1.1355, 1.1450, 1.1937, 1.3117, 1.2002, 1.3688, 1.1001, 1.0248,\n", + " 1.2118, 1.3297, 1.3135, 0.9588, 0.9512, 1.0657, 1.0919, 1.2105, 1.3276,\n", + " 1.1991, 1.0544, 1.1360, 1.3764, 1.1669, 1.1038, 1.1211, 1.2504, 1.1048,\n", + " 1.2078, 1.0024, 1.2304, 1.3249, 1.2139, 1.0967, 1.1354, 0.9694, 1.1348,\n", + " 1.1234, 1.0613, 1.3629, 1.0437, 1.2400, 1.2838, 1.1275, 0.9635, 1.1311,\n", + " 1.2330, 1.1727, 1.1354, 1.1728, 1.0593, 1.2356, 1.1613, 1.2037, 1.2730,\n", + " 1.4041, 0.8310, 1.0301, 1.1150, 1.2088, 1.1890, 1.1474, 1.1404, 1.3967,\n", + " 1.1439, 1.1394, 1.1124, 1.1378, 1.1184, 1.1519, 1.1790, 1.1773, 1.2587,\n", + " 1.2223, 1.0439, 1.2656, 1.2736, 1.3075, 1.1752, 1.6114, 1.1593, 1.2341,\n", + " 1.0457, 1.2959, 1.1165, 1.1556, 1.3303, 1.1619, 1.1953, 1.2901, 1.0919,\n", + " 1.3273, 1.0691, 1.1831, 1.2258, 1.4424, 1.2168, 1.3488, 1.1385, 1.2099,\n", + " 1.1935, 1.1255, 1.2476, 1.1882, 1.1218, 1.1213, 1.0368, 1.3854, 1.1863,\n", + " 1.1656, 1.3210, 1.2381, 1.2143, 1.1449, 1.1805, 1.5149, 1.1286, 1.0308,\n", + " 1.0947, 1.2880, 1.2814, 1.1052, 1.1948, 1.2060, 1.3477, 1.2810, 1.2086,\n", + " 1.1298, 1.1495]),\n", + " 'res5.in2.bias': tensor([-0.1472, 0.0328, -0.2884, 0.0791, 0.1536, 0.0026, 0.1078, 0.0316,\n", + " -0.0164, -0.1926, 0.0861, -0.0146, 0.3491, -0.0218, 0.0737, 0.0200,\n", + " -0.1683, 0.1437, 0.1188, 0.0482, 0.0668, -0.0975, -0.1927, 0.1252,\n", + " 0.0706, -0.0997, 0.0112, 0.0292, 0.1187, 0.1608, 0.0813, -0.1116,\n", + " -0.2416, -0.0692, -0.2155, 0.1388, -0.0512, -0.0849, 0.0313, -0.0525,\n", + " -0.0967, -0.0520, -0.0537, -0.0315, -0.2610, 0.1141, 0.2241, -0.0532,\n", + " 0.1662, 0.2526, 0.0206, -0.1228, -0.0682, -0.0583, 0.0367, -0.0546,\n", + " -0.1482, -0.0777, -0.0531, 0.1883, 0.0896, 0.0948, 0.0466, -0.0757,\n", + " -0.0239, 0.0323, 0.1288, -0.1593, -0.1723, 0.1703, 0.0104, -0.1099,\n", + " -0.1312, -0.0938, 0.1332, -0.2516, 0.0204, 0.2359, 0.1006, 0.2336,\n", + " 0.2047, -0.0278, -0.1024, 0.1919, -0.1187, -0.0100, -0.0960, -0.1736,\n", + " -0.0308, 0.0264, -0.0329, -0.1007, 0.0404, -0.1072, -0.0701, -0.0030,\n", + " -0.0574, 0.1282, 0.0217, -0.0332, 0.0628, 0.0119, -0.2008, -0.0086,\n", + " -0.0158, -0.0850, 0.1969, 0.0008, -0.0020, 0.0383, 0.0013, 0.1310,\n", + " 0.0458, 0.0727, -0.0208, 0.0901, 0.0298, -0.2636, 0.0443, 0.0939,\n", + " 0.0679, -0.1464, -0.0330, -0.2032, 0.0786, -0.0129, -0.0634, 0.0829]),\n", + " 'deconv1.conv2d.weight': tensor([[[[ 0.0074, -0.0080, 0.0563],\n", + " [ 0.0300, -0.0281, 0.0839],\n", + " [ 0.0875, 0.0106, 0.0340]],\n", + " \n", + " [[-0.0359, -0.1401, 0.0110],\n", + " [ 0.0863, 0.0306, 0.0293],\n", + " [ 0.1120, 0.0400, -0.0055]],\n", + " \n", + " [[-0.1424, -0.0842, -0.0410],\n", + " [-0.0737, -0.1111, -0.0882],\n", + " [-0.0528, -0.0964, -0.0624]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0825, -0.1105, 0.0241],\n", + " [-0.1166, -0.0405, 0.0107],\n", + " [-0.1086, -0.0827, -0.0105]],\n", + " \n", + " [[ 0.2305, 0.1935, 0.2049],\n", + " [ 0.1596, 0.1737, 0.2176],\n", + " [ 0.2052, 0.2296, 0.2445]],\n", + " \n", + " [[-0.0327, 0.0400, 0.0094],\n", + " [ 0.0009, 0.0339, 0.0219],\n", + " [ 0.0356, 0.0152, 0.0250]]],\n", + " \n", + " \n", + " [[[-0.0328, -0.0636, 0.0699],\n", + " [-0.0658, -0.1910, 0.0217],\n", + " [ 0.1004, -0.1528, 0.0772]],\n", + " \n", + " [[-0.1394, 0.1039, 0.0091],\n", + " [ 0.0585, 0.1688, -0.0774],\n", + " [ 0.0306, 0.0790, -0.0607]],\n", + " \n", + " [[-0.0198, -0.0839, -0.0294],\n", + " [-0.1098, -0.0922, 0.0433],\n", + " [-0.0395, -0.0459, 0.1468]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0316, -0.0889, -0.0160],\n", + " [-0.1173, -0.0987, 0.2230],\n", + " [ 0.0124, -0.0787, 0.0213]],\n", + " \n", + " [[-0.0099, 0.1572, 0.0850],\n", + " [-0.0024, 0.2078, -0.0063],\n", + " [ 0.0960, 0.1615, -0.1318]],\n", + " \n", + " [[-0.1432, -0.0744, -0.0489],\n", + " [ 0.0031, 0.1143, 0.0612],\n", + " [-0.1192, 0.1147, 0.1158]]],\n", + " \n", + " \n", + " [[[-0.0199, 0.0010, -0.1273],\n", + " [ 0.0152, 0.0751, 0.0619],\n", + " [-0.0541, 0.0617, -0.0923]],\n", + " \n", + " [[ 0.0433, 0.1262, 0.0567],\n", + " [ 0.1222, 0.0608, -0.0708],\n", + " [-0.0523, -0.1826, -0.1291]],\n", + " \n", + " [[-0.0155, -0.0272, -0.0513],\n", + " [ 0.0730, 0.0022, -0.0443],\n", + " [-0.0263, -0.0024, -0.0823]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0340, 0.0471, 0.0066],\n", + " [ 0.0898, 0.0217, -0.0115],\n", + " [ 0.0757, -0.0248, 0.1197]],\n", + " \n", + " [[-0.0853, -0.0442, -0.1247],\n", + " [-0.0482, -0.0243, -0.0471],\n", + " [-0.1092, -0.0807, -0.1375]],\n", + " \n", + " [[ 0.0191, 0.0574, 0.0012],\n", + " [-0.0536, -0.0823, -0.0232],\n", + " [ 0.0540, 0.0324, 0.1118]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[-0.1025, 0.1595, -0.0329],\n", + " [-0.0937, 0.2239, -0.2700],\n", + " [-0.0097, 0.1013, -0.0566]],\n", + " \n", + " [[ 0.1219, -0.0009, -0.0971],\n", + " [ 0.0472, 0.0056, -0.0363],\n", + " [ 0.1275, -0.1338, 0.0894]],\n", + " \n", + " [[-0.0055, 0.1493, -0.1246],\n", + " [ 0.1512, 0.1091, -0.0270],\n", + " [-0.0390, -0.0033, -0.0354]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0153, 0.1162, -0.0310],\n", + " [-0.1093, 0.0023, -0.1599],\n", + " [-0.0837, 0.1180, 0.0730]],\n", + " \n", + " [[ 0.0210, 0.0230, -0.0544],\n", + " [ 0.0205, -0.0111, -0.0643],\n", + " [-0.0632, -0.0291, 0.1018]],\n", + " \n", + " [[-0.0051, -0.0339, -0.0091],\n", + " [ 0.0962, -0.0301, -0.0320],\n", + " [-0.0693, 0.1252, -0.0362]]],\n", + " \n", + " \n", + " [[[ 0.1262, 0.1224, -0.0272],\n", + " [-0.0523, -0.0509, -0.0821],\n", + " [ 0.0291, -0.0312, -0.0981]],\n", + " \n", + " [[-0.0092, -0.0344, 0.0792],\n", + " [ 0.0681, -0.0277, -0.0066],\n", + " [ 0.0342, 0.0663, -0.1561]],\n", + " \n", + " [[-0.1496, -0.1297, -0.0718],\n", + " [-0.1047, -0.0639, -0.0328],\n", + " [-0.1356, -0.0171, -0.0081]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0608, 0.0180, -0.0642],\n", + " [ 0.0061, 0.0417, 0.0135],\n", + " [-0.0220, -0.0262, -0.1027]],\n", + " \n", + " [[-0.0911, 0.0286, -0.0238],\n", + " [-0.0537, 0.0497, 0.0269],\n", + " [-0.0606, -0.0807, 0.0166]],\n", + " \n", + " [[ 0.0614, -0.0254, 0.0975],\n", + " [ 0.0372, -0.0654, -0.0363],\n", + " [ 0.0335, -0.0773, 0.0418]]],\n", + " \n", + " \n", + " [[[-0.0347, -0.0242, -0.0247],\n", + " [-0.0086, 0.0444, 0.1129],\n", + " [ 0.0015, -0.1345, 0.0696]],\n", + " \n", + " [[-0.0912, -0.2072, -0.1738],\n", + " [ 0.1960, 0.1062, -0.0834],\n", + " [-0.0271, 0.2286, 0.0008]],\n", + " \n", + " [[ 0.0130, -0.0177, 0.0366],\n", + " [-0.0860, -0.1761, -0.0536],\n", + " [-0.0113, -0.0982, -0.0238]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.1703, -0.0006, -0.2810],\n", + " [-0.1438, 0.1932, 0.0624],\n", + " [-0.0357, 0.0339, 0.1690]],\n", + " \n", + " [[-0.0552, 0.0778, 0.0974],\n", + " [-0.0171, -0.0076, 0.0155],\n", + " [ 0.0284, -0.0253, -0.0969]],\n", + " \n", + " [[-0.0555, 0.3445, 0.1059],\n", + " [-0.0927, -0.0800, -0.1623],\n", + " [-0.0357, -0.0626, -0.0984]]]]),\n", + " 'deconv1.conv2d.bias': tensor([-0.0936, -0.1012, -0.0937, -0.1521, -0.0538, 0.1818, 0.0224, 0.1838,\n", + " -0.1147, -0.0070, -0.0593, -0.0933, -0.2525, 0.2333, 0.0060, -0.0628,\n", + " -0.3178, 0.2844, 0.0767, -0.0764, 0.1967, 0.0955, 0.0376, -0.1202,\n", + " -0.0433, -0.0173, -0.0900, -0.2419, -0.1772, -0.1507, -0.1090, 0.0614,\n", + " 0.0200, -0.0538, -0.0441, 0.0899, 0.2438, 0.0218, -0.0604, 0.2176,\n", + " 0.1383, -0.3305, -0.0719, 0.0633, 0.1462, 0.3388, 0.0531, 0.0588,\n", + " -0.0996, -0.0105, 0.1507, -0.2575, -0.0583, 0.2523, 0.2232, -0.1308,\n", + " 0.1204, 0.1276, -0.1185, 0.0845, -0.1090, 0.1203, -0.0934, -0.1089]),\n", + " 'in4.weight': tensor([0.9842, 0.9061, 0.9193, 1.0175, 0.8906, 0.9393, 0.8579, 1.0401, 1.1450,\n", + " 0.9735, 0.9343, 1.0383, 0.9441, 0.8709, 1.0394, 0.8908, 0.9042, 0.9830,\n", + " 0.9651, 0.9732, 0.9410, 0.9941, 1.0039, 0.7828, 1.0612, 0.8760, 0.9070,\n", + " 1.0103, 1.0069, 1.0929, 0.8136, 0.9772, 1.0105, 0.9768, 1.1602, 0.8824,\n", + " 0.7700, 1.0808, 1.0157, 1.1412, 0.9401, 0.9697, 1.2021, 0.9920, 0.7854,\n", + " 1.0669, 1.0688, 1.0977, 1.0001, 0.8514, 0.9041, 0.9751, 1.1644, 0.9256,\n", + " 0.9852, 0.8742, 0.9090, 0.9114, 0.8424, 1.1880, 0.9281, 1.0215, 1.0222,\n", + " 1.1841]),\n", + " 'in4.bias': tensor([-1.9221e-01, 1.0746e-01, -7.6189e-02, 1.3272e-01, 2.3797e-01,\n", + " -9.6581e-02, 1.3485e-01, 1.6731e-01, -3.6485e-01, -3.5545e-01,\n", + " -2.6448e-02, 9.2171e-02, -9.7275e-02, 3.2626e-01, -6.2046e-02,\n", + " 4.6572e-02, -1.9281e-02, -2.5059e-01, 2.4123e-01, -1.9033e-01,\n", + " -1.3048e-01, 6.7093e-02, -2.0236e-01, -7.9008e-02, 5.3519e-02,\n", + " -5.7277e-01, -1.0522e-01, -7.4963e-02, 8.8941e-02, -1.3756e-01,\n", + " -2.2325e-01, -2.6285e-02, -5.7634e-01, -6.9413e-02, 1.2463e-01,\n", + " -2.4769e-01, -2.1116e-01, -9.0390e-02, 4.3229e-03, 6.9607e-02,\n", + " -5.7818e-02, -5.6572e-04, -3.1495e-01, -3.1120e-02, -5.0814e-02,\n", + " 1.1488e-01, 1.8677e-01, 9.9703e-02, 1.2226e-01, -1.2130e-01,\n", + " 3.9999e-01, -3.4431e-01, 1.5797e-01, -2.0906e-02, -7.7865e-02,\n", + " 7.1772e-02, 2.0697e-01, -3.0306e-01, 1.3181e-01, -2.2648e-02,\n", + " 1.3879e-02, -1.6085e-02, 2.3846e-01, 2.2598e-01]),\n", + " 'deconv2.conv2d.weight': tensor([[[[-0.1294, -0.0545, -0.0947],\n", + " [-0.1109, -0.0267, -0.0283],\n", + " [-0.0668, -0.0116, 0.0041]],\n", + " \n", + " [[-0.0476, 0.0341, -0.1268],\n", + " [ 0.0197, 0.1444, 0.0285],\n", + " [ 0.0171, 0.0961, 0.0788]],\n", + " \n", + " [[ 0.1752, 0.0919, -0.0078],\n", + " [-0.0228, -0.0708, -0.1354],\n", + " [-0.1405, -0.0886, -0.0758]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0227, -0.1297, -0.0466],\n", + " [-0.0161, -0.0181, -0.0690],\n", + " [ 0.0381, 0.1122, 0.0128]],\n", + " \n", + " [[-0.1556, -0.1674, -0.1508],\n", + " [-0.0689, -0.1055, -0.0453],\n", + " [-0.0940, -0.0712, -0.0777]],\n", + " \n", + " [[-0.0548, -0.1258, -0.1213],\n", + " [ 0.0281, 0.1993, 0.2585],\n", + " [-0.0299, -0.0850, -0.1206]]],\n", + " \n", + " \n", + " [[[-0.0835, -0.0870, -0.0597],\n", + " [-0.0343, -0.0447, -0.0895],\n", + " [-0.0313, -0.0777, -0.1094]],\n", + " \n", + " [[-0.0837, -0.0016, -0.0866],\n", + " [-0.1200, 0.0237, -0.0565],\n", + " [-0.0434, 0.1095, 0.0875]],\n", + " \n", + " [[-0.0800, -0.0771, -0.1880],\n", + " [-0.0618, -0.0927, -0.1587],\n", + " [ 0.0223, -0.1193, -0.2389]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.0398, 0.0359, 0.1678],\n", + " [-0.0777, 0.0068, 0.0935],\n", + " [-0.0123, 0.0359, 0.1157]],\n", + " \n", + " [[-0.0716, -0.1095, -0.1566],\n", + " [-0.0540, -0.0680, -0.0566],\n", + " [-0.0665, -0.0828, -0.1148]],\n", + " \n", + " [[ 0.0871, 0.1051, 0.0564],\n", + " [-0.2262, -0.1608, -0.1427],\n", + " [ 0.1764, 0.0806, 0.0579]]],\n", + " \n", + " \n", + " [[[-0.0580, -0.0578, -0.0673],\n", + " [-0.1029, -0.0357, -0.0614],\n", + " [-0.0582, -0.0618, -0.0191]],\n", + " \n", + " [[ 0.0898, -0.0353, 0.0187],\n", + " [ 0.0588, -0.0630, -0.0422],\n", + " [ 0.1290, -0.0233, -0.0051]],\n", + " \n", + " [[ 0.2155, 0.0683, 0.0051],\n", + " [ 0.1271, 0.0863, -0.0488],\n", + " [ 0.1342, 0.0737, -0.0170]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0342, -0.0455, -0.0536],\n", + " [ 0.0806, -0.0095, 0.0020],\n", + " [-0.0102, -0.0293, -0.0239]],\n", + " \n", + " [[-0.1156, -0.0687, -0.1052],\n", + " [-0.1734, -0.1273, -0.0886],\n", + " [-0.0934, -0.1375, -0.0399]],\n", + " \n", + " [[-0.0125, -0.0738, -0.0495],\n", + " [ 0.0251, 0.0441, 0.0523],\n", + " [-0.0696, -0.0793, -0.0698]]],\n", + " \n", + " \n", + " ...,\n", + " \n", + " \n", + " [[[-0.0430, 0.0210, -0.0303],\n", + " [-0.0088, -0.0213, -0.0068],\n", + " [-0.0705, -0.0391, -0.0855]],\n", + " \n", + " [[ 0.0147, -0.0655, 0.0174],\n", + " [ 0.0657, 0.0145, 0.0653],\n", + " [ 0.1065, 0.0830, 0.0338]],\n", + " \n", + " [[-0.0161, -0.0159, -0.1039],\n", + " [-0.0861, 0.0230, -0.0762],\n", + " [-0.0428, 0.0078, -0.0474]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0637, 0.0093, -0.0524],\n", + " [ 0.0123, -0.0259, -0.0857],\n", + " [ 0.0115, -0.0004, 0.0857]],\n", + " \n", + " [[-0.1463, -0.0966, -0.0767],\n", + " [-0.1275, -0.0999, -0.0952],\n", + " [-0.1363, -0.0577, -0.0587]],\n", + " \n", + " [[ 0.1206, 0.0709, 0.0235],\n", + " [-0.3078, -0.2512, -0.2316],\n", + " [ 0.2659, 0.2739, 0.1726]]],\n", + " \n", + " \n", + " [[[-0.0536, -0.0532, -0.0172],\n", + " [-0.1134, -0.0450, -0.0807],\n", + " [-0.0527, -0.0361, -0.0804]],\n", + " \n", + " [[-0.0923, -0.0874, 0.0786],\n", + " [ 0.0296, -0.1247, -0.0026],\n", + " [ 0.0287, -0.0784, -0.1571]],\n", + " \n", + " [[-0.0452, 0.0627, 0.0968],\n", + " [-0.0223, 0.0719, 0.0500],\n", + " [-0.1468, 0.0422, 0.0718]],\n", + " \n", + " ...,\n", + " \n", + " [[-0.1289, -0.0942, 0.0370],\n", + " [-0.1846, -0.1195, 0.0744],\n", + " [-0.0638, -0.0511, 0.0192]],\n", + " \n", + " [[-0.0569, -0.0404, -0.0457],\n", + " [-0.0759, -0.0913, 0.0213],\n", + " [-0.0128, -0.1342, -0.0337]],\n", + " \n", + " [[-0.0372, -0.0190, 0.0513],\n", + " [ 0.0405, 0.0358, -0.0073],\n", + " [ 0.0841, 0.0026, -0.1044]]],\n", + " \n", + " \n", + " [[[ 0.0266, 0.0543, 0.0431],\n", + " [ 0.0025, 0.0508, 0.0237],\n", + " [-0.0439, 0.0165, 0.0106]],\n", + " \n", + " [[ 0.0058, 0.0548, -0.0092],\n", + " [-0.0281, 0.0374, 0.0143],\n", + " [ 0.0008, -0.0576, 0.0194]],\n", + " \n", + " [[-0.0478, -0.0173, -0.0736],\n", + " [-0.0718, -0.0058, -0.0232],\n", + " [-0.0794, 0.0489, 0.0180]],\n", + " \n", + " ...,\n", + " \n", + " [[ 0.0404, 0.0368, 0.0134],\n", + " [-0.0253, 0.0564, 0.0180],\n", + " [-0.0495, 0.0625, 0.0238]],\n", + " \n", + " [[-0.2002, -0.1695, -0.1422],\n", + " [-0.1445, -0.1611, -0.1104],\n", + " [-0.1192, -0.1530, -0.1182]],\n", + " \n", + " [[ 0.0011, 0.0556, -0.0029],\n", + " [ 0.0397, 0.0098, 0.0147],\n", + " [-0.0595, -0.0434, -0.0358]]]]),\n", + " 'deconv2.conv2d.bias': tensor([ 0.0915, -0.0942, 0.1146, 0.1183, -0.2425, 0.0437, 0.0394, -0.1087,\n", + " -0.3324, 0.0209, -0.0304, -0.1474, -0.0498, -0.0493, 0.0274, 0.1026,\n", + " 0.0975, -0.0864, -0.0162, 0.1119, -0.0217, 0.2456, -0.0538, 0.0914,\n", + " -0.2083, -0.0944, -0.1761, 0.0291, 0.0237, 0.0315, 0.0222, 0.0130]),\n", + " 'in5.weight': tensor([1.1269, 1.2278, 1.3141, 1.6012, 1.2806, 1.5737, 1.2854, 1.3499, 1.6401,\n", + " 1.8579, 1.0621, 1.0790, 1.6552, 1.2020, 1.0735, 1.2732, 1.2408, 1.2734,\n", + " 1.3268, 1.2310, 1.2582, 1.5684, 1.2350, 1.2635, 1.2419, 1.1329, 1.2269,\n", + " 1.2569, 1.5506, 1.1232, 1.4963, 1.1877]),\n", + " 'in5.bias': tensor([0.8869, 0.5009, 0.5812, 0.6170, 0.3923, 0.5126, 1.3038, 0.6838, 0.5469,\n", + " 0.4946, 1.2467, 1.1618, 0.4959, 0.6230, 0.6535, 0.6131, 0.3895, 0.7427,\n", + " 0.5899, 0.3722, 0.3141, 0.6118, 0.4792, 0.5788, 0.2973, 1.2467, 0.3557,\n", + " 0.4963, 0.6114, 0.5701, 0.6839, 0.4126]),\n", + " 'deconv3.conv2d.weight': tensor([[[[ 3.2771e-02, -1.1656e-02, -6.8142e-02, ..., -8.3614e-02,\n", + " -8.0133e-03, 2.2606e-02],\n", + " [ 8.4824e-02, 8.8802e-02, 8.1544e-02, ..., 1.8808e-01,\n", + " 1.6263e-01, 8.5595e-02],\n", + " [ 9.3750e-02, 8.4448e-03, 8.0531e-02, ..., 2.2060e-01,\n", + " 2.0762e-01, 1.5631e-01],\n", + " ...,\n", + " [ 3.0062e-02, -5.1876e-02, -1.4505e-01, ..., -2.9930e-01,\n", + " -2.3552e-04, 1.3036e-01],\n", + " [-6.5396e-02, -1.2577e-02, -1.0025e-01, ..., -2.3994e-01,\n", + " -3.9920e-03, 1.3184e-02],\n", + " [-5.8952e-02, -2.3119e-02, 8.0088e-02, ..., -6.3957e-02,\n", + " 8.3895e-02, -7.7074e-03]],\n", + " \n", + " [[ 1.1961e-01, 1.4349e-01, 1.5482e-01, ..., 1.6078e-02,\n", + " 5.2004e-02, 1.0234e-02],\n", + " [ 6.5094e-02, 1.1747e-01, 1.5367e-01, ..., -5.5684e-02,\n", + " 8.1633e-03, -2.9713e-02],\n", + " [ 6.3225e-03, 1.0664e-01, 1.9482e-01, ..., -1.4038e-01,\n", + " 2.7422e-02, -5.5842e-02],\n", + " ...,\n", + " [ 6.3132e-02, 1.9397e-01, 2.4338e-01, ..., -1.0526e-01,\n", + " 2.7870e-02, -4.0618e-02],\n", + " [ 6.6475e-02, 1.4960e-01, 1.4160e-01, ..., 4.4794e-02,\n", + " 1.0822e-01, -4.0280e-02],\n", + " [ 1.5867e-01, 2.4400e-01, 2.0004e-01, ..., 1.1847e-01,\n", + " 1.1883e-01, 3.6181e-02]],\n", + " \n", + " [[-2.3321e-02, 8.0748e-03, 9.7943e-03, ..., 8.4530e-02,\n", + " 3.0065e-02, -9.0375e-02],\n", + " [ 3.1290e-03, 1.3776e-01, 1.4799e-01, ..., 1.1675e-01,\n", + " 1.3147e-01, 4.7408e-03],\n", + " [-5.2472e-02, 6.4026e-02, 7.4648e-02, ..., -4.7169e-02,\n", + " 1.7588e-02, -1.9598e-02],\n", + " ...,\n", + " [-9.1378e-02, 4.2844e-02, 3.7021e-02, ..., -1.3178e-01,\n", + " 6.6164e-02, 2.2903e-02],\n", + " [-6.0196e-02, 7.0224e-02, -4.5455e-02, ..., -2.2338e-01,\n", + " 3.3437e-02, -5.2640e-02],\n", + " [-1.6823e-01, 3.6807e-02, 8.6206e-02, ..., 7.3992e-02,\n", + " 1.0765e-01, -5.5084e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[ 6.5849e-02, 1.1591e-01, 6.4292e-02, ..., 1.2117e-01,\n", + " 5.9087e-02, 8.6197e-02],\n", + " [ 7.6181e-02, 9.3731e-02, -3.9322e-02, ..., 8.2448e-02,\n", + " -6.0789e-03, 6.2521e-02],\n", + " [ 8.6458e-02, 7.3733e-02, 3.7398e-02, ..., 1.7427e-01,\n", + " 7.7054e-02, 1.4943e-01],\n", + " ...,\n", + " [ 1.2649e-01, 5.7835e-02, 5.1641e-02, ..., -1.2165e-01,\n", + " 1.7386e-02, 1.1838e-01],\n", + " [ 1.3228e-01, 8.0972e-02, 1.0069e-01, ..., 5.0541e-02,\n", + " 1.2525e-01, 1.8117e-01],\n", + " [ 1.2774e-01, 1.1652e-01, 1.1474e-01, ..., 5.8662e-04,\n", + " 1.4742e-01, 1.1876e-01]],\n", + " \n", + " [[ 2.6491e-02, 2.4479e-02, -9.9262e-03, ..., -4.4809e-02,\n", + " -1.3972e-02, 8.6580e-02],\n", + " [ 7.4275e-02, 8.3362e-02, 1.0486e-01, ..., 9.7586e-02,\n", + " 3.6410e-02, 1.7422e-01],\n", + " [ 3.1006e-02, 8.3711e-02, 4.4571e-02, ..., 6.8200e-02,\n", + " -6.3988e-02, 5.7015e-02],\n", + " ...,\n", + " [-1.2853e-01, 9.0864e-02, 2.0010e-01, ..., 1.8399e-01,\n", + " -9.2196e-02, -4.6385e-02],\n", + " [ 1.1811e-01, 5.9988e-02, -2.2721e-01, ..., 4.3318e-02,\n", + " 1.5362e-02, 7.4826e-02],\n", + " [ 1.0364e-02, 2.0778e-01, -1.5964e-01, ..., 4.0748e-02,\n", + " 6.0082e-03, -8.2132e-03]],\n", + " \n", + " [[ 3.3584e-02, 3.9030e-02, 5.7761e-02, ..., 1.2216e-02,\n", + " -7.0585e-03, 4.3796e-02],\n", + " [ 5.2266e-02, 6.5033e-02, 5.2559e-02, ..., 3.0077e-02,\n", + " 2.8688e-02, 8.0254e-02],\n", + " [ 6.7801e-02, 7.0803e-02, 2.2669e-02, ..., 5.6638e-02,\n", + " 3.2285e-02, 8.2530e-02],\n", + " ...,\n", + " [ 4.1627e-02, 4.9316e-02, 3.5680e-02, ..., 2.0071e-02,\n", + " 7.0274e-03, 1.6493e-02],\n", + " [ 6.8174e-02, 8.6090e-02, 6.1226e-02, ..., -1.5565e-02,\n", + " 6.0356e-02, 6.1023e-02],\n", + " [ 5.5528e-02, 7.7656e-02, 1.7746e-02, ..., -1.2501e-02,\n", + " 1.1452e-02, 3.8445e-02]]],\n", + " \n", + " \n", + " [[[ 7.4018e-03, 4.0419e-03, 5.1836e-02, ..., 3.4645e-02,\n", + " 1.3005e-01, 7.1236e-02],\n", + " [-5.1866e-02, 4.7196e-02, 1.3559e-01, ..., 1.3815e-01,\n", + " 1.7068e-01, 8.0958e-02],\n", + " [-6.5827e-03, 3.7686e-02, 1.1213e-01, ..., 1.1429e-01,\n", + " 1.9054e-01, 1.2207e-01],\n", + " ...,\n", + " [-1.0165e-02, 1.6161e-02, -5.1698e-02, ..., -1.9688e-02,\n", + " 4.1872e-02, 7.1493e-02],\n", + " [ 7.9446e-02, 1.1139e-01, 4.6507e-02, ..., 4.4623e-02,\n", + " 9.9029e-02, 7.5998e-02],\n", + " [-9.4745e-03, 1.0169e-02, -2.6547e-02, ..., 1.3507e-03,\n", + " 1.6619e-02, 9.0237e-03]],\n", + " \n", + " [[ 3.9285e-03, -2.7422e-02, 3.0922e-02, ..., 1.3179e-02,\n", + " 4.8536e-02, 3.8512e-02],\n", + " [ 6.0031e-04, 1.9361e-02, 5.9657e-02, ..., -3.0622e-02,\n", + " 1.9341e-03, 1.4554e-03],\n", + " [ 1.5134e-02, 7.2711e-02, 1.3580e-01, ..., -2.3448e-03,\n", + " 5.4212e-02, 1.9527e-02],\n", + " ...,\n", + " [ 4.8095e-02, 7.0361e-02, 1.3038e-01, ..., 3.1442e-02,\n", + " 9.3217e-03, -2.4357e-03],\n", + " [ 1.0117e-02, 4.6312e-03, 5.8446e-02, ..., -2.4674e-02,\n", + " -1.1924e-03, 3.1652e-03],\n", + " [ 1.0487e-02, 4.2419e-02, 5.6233e-02, ..., -1.7863e-02,\n", + " 1.1510e-02, 5.6531e-02]],\n", + " \n", + " [[ 8.6201e-02, 2.2736e-02, 1.4923e-02, ..., 4.8925e-02,\n", + " 6.9612e-02, 4.1760e-02],\n", + " [ 6.6875e-02, 3.3861e-02, 2.8449e-02, ..., -1.3802e-03,\n", + " -2.0041e-03, 4.2423e-02],\n", + " [ 4.2806e-02, 6.2963e-04, 4.1867e-02, ..., -5.5489e-02,\n", + " -4.4326e-02, 1.5638e-02],\n", + " ...,\n", + " [ 9.3927e-02, 2.0650e-02, 8.0071e-02, ..., 1.3091e-01,\n", + " 8.0053e-02, 6.0861e-02],\n", + " [ 6.4131e-02, -8.1969e-03, 1.6075e-02, ..., 3.1647e-02,\n", + " 3.2516e-02, 2.3509e-02],\n", + " [ 3.5929e-02, -4.2648e-02, 1.5922e-02, ..., 4.0720e-02,\n", + " -1.5319e-02, -2.0371e-02]],\n", + " \n", + " ...,\n", + " \n", + " [[ 2.6757e-02, 5.4513e-02, 2.6361e-02, ..., 7.2981e-02,\n", + " 8.6459e-02, 1.0534e-01],\n", + " [ 1.2251e-02, 5.8692e-02, 2.1836e-02, ..., 1.3197e-02,\n", + " 4.2997e-02, 4.9866e-02],\n", + " [ 5.5434e-03, 5.0646e-02, 9.6724e-02, ..., 5.1262e-02,\n", + " 5.7809e-02, 7.7184e-02],\n", + " ...,\n", + " [ 6.6736e-02, 5.2741e-02, 1.5601e-01, ..., 1.0632e-01,\n", + " 7.7346e-02, 1.2898e-01],\n", + " [ 2.5531e-02, -7.3291e-03, 2.6494e-02, ..., 2.6832e-02,\n", + " 3.1126e-02, 6.3864e-02],\n", + " [ 7.2615e-02, 4.9367e-02, 8.6340e-02, ..., -7.4520e-03,\n", + " 7.6799e-02, 9.7371e-02]],\n", + " \n", + " [[ 4.4071e-02, -9.1801e-04, 6.5188e-02, ..., 1.3349e-01,\n", + " 8.2732e-02, -1.9904e-02],\n", + " [ 6.6533e-02, -1.7603e-02, 1.1545e-02, ..., 4.6236e-02,\n", + " 4.9954e-02, 2.9793e-02],\n", + " [ 6.8090e-02, 3.8963e-02, -1.2739e-02, ..., 3.6751e-02,\n", + " 1.9449e-02, 4.0692e-02],\n", + " ...,\n", + " [-3.3542e-02, 4.2381e-02, 1.1486e-01, ..., 1.6422e-01,\n", + " 1.8518e-02, 1.9617e-02],\n", + " [ 2.8308e-02, -5.0646e-02, -1.0231e-01, ..., 8.1270e-02,\n", + " 3.0508e-02, -4.3493e-02],\n", + " [ 3.9462e-02, 1.2977e-01, 8.8224e-02, ..., 1.2952e-01,\n", + " 1.1492e-01, 1.0218e-02]],\n", + " \n", + " [[ 1.3572e-01, 1.0765e-01, 1.2613e-01, ..., 1.0066e-01,\n", + " 1.0864e-01, 1.4372e-01],\n", + " [ 8.2709e-02, 1.1740e-01, 1.3475e-01, ..., 1.0653e-01,\n", + " 1.5032e-01, 1.3986e-01],\n", + " [ 8.3571e-02, 1.1900e-01, 1.4205e-01, ..., 5.7168e-02,\n", + " 9.5233e-02, 1.2814e-01],\n", + " ...,\n", + " [ 1.0619e-01, 1.0715e-01, 1.2485e-01, ..., 9.4612e-02,\n", + " 1.6047e-01, 1.3917e-01],\n", + " [ 1.0966e-01, 1.2234e-01, 1.3195e-01, ..., 8.9528e-02,\n", + " 1.5922e-01, 1.4495e-01],\n", + " [ 1.2299e-01, 1.2449e-01, 1.2864e-01, ..., 9.7505e-02,\n", + " 1.5168e-01, 1.4385e-01]]],\n", + " \n", + " \n", + " [[[ 9.3081e-02, 7.3211e-02, 9.2631e-02, ..., 7.8207e-03,\n", + " 7.0456e-03, 7.2446e-02],\n", + " [ 3.8194e-02, 6.2228e-02, 2.5160e-02, ..., -4.7167e-02,\n", + " -3.2355e-02, 6.7824e-02],\n", + " [-3.3938e-02, -1.9700e-02, -4.8784e-02, ..., -9.4935e-02,\n", + " -7.2434e-02, 1.2301e-01],\n", + " ...,\n", + " [ 8.3838e-02, 5.1790e-02, 2.7209e-02, ..., 1.2126e-01,\n", + " 1.7845e-02, 8.3741e-02],\n", + " [ 1.4282e-02, 2.0798e-02, -1.9180e-02, ..., -2.0645e-02,\n", + " -3.9852e-02, 2.1974e-02],\n", + " [-1.2676e-02, -4.6993e-02, -8.1847e-02, ..., -5.1035e-02,\n", + " -5.2540e-02, -8.0928e-03]],\n", + " \n", + " [[-1.9673e-02, -3.9708e-03, 5.2304e-02, ..., 4.6471e-02,\n", + " 1.0894e-01, -2.1526e-02],\n", + " [-1.5173e-02, -4.2693e-03, 4.7255e-02, ..., -3.5673e-02,\n", + " 6.2218e-02, -1.8960e-02],\n", + " [-5.9138e-04, -1.3347e-02, -2.2374e-03, ..., -3.2764e-02,\n", + " 1.1044e-01, 2.0177e-02],\n", + " ...,\n", + " [ 2.9323e-02, -2.6600e-02, -1.0026e-02, ..., 1.1013e-01,\n", + " 1.1494e-01, -1.9608e-02],\n", + " [ 5.1470e-02, -2.3202e-03, 1.6107e-02, ..., 1.3337e-02,\n", + " 5.9104e-02, -3.6030e-03],\n", + " [ 2.1413e-03, 3.1370e-02, 6.8175e-02, ..., -6.0787e-03,\n", + " 8.5091e-02, 2.7318e-02]],\n", + " \n", + " [[ 1.7577e-01, 1.8333e-01, 1.6445e-01, ..., 2.1188e-01,\n", + " 2.2973e-01, 1.7488e-01],\n", + " [ 1.0611e-01, 1.4949e-01, 1.2594e-01, ..., 1.0798e-01,\n", + " 1.8054e-01, 2.3116e-01],\n", + " [ 9.3080e-02, 1.4338e-01, 1.3039e-01, ..., 8.5919e-02,\n", + " 1.4873e-01, 2.1318e-01],\n", + " ...,\n", + " [ 2.0594e-01, 1.8537e-01, 1.7957e-01, ..., 2.5187e-01,\n", + " 2.3976e-01, 1.5556e-01],\n", + " [ 2.3226e-01, 2.0303e-01, 1.0204e-01, ..., 2.5203e-01,\n", + " 2.6102e-01, 1.7734e-01],\n", + " [ 2.7248e-01, 1.9309e-01, 1.1104e-01, ..., 2.7160e-01,\n", + " 2.1623e-01, 1.9764e-01]],\n", + " \n", + " ...,\n", + " \n", + " [[-6.5619e-02, -5.0082e-02, -2.6704e-02, ..., 7.1071e-02,\n", + " 1.3986e-02, -1.1090e-02],\n", + " [ 2.4437e-03, 2.1230e-02, -1.7267e-02, ..., 1.0551e-01,\n", + " -2.2639e-03, 5.9378e-02],\n", + " [ 2.3235e-04, -1.5483e-03, -3.3116e-02, ..., 8.4449e-02,\n", + " -1.3242e-02, 9.0661e-02],\n", + " ...,\n", + " [ 3.6619e-02, -1.7030e-02, 7.2965e-02, ..., -3.4544e-02,\n", + " -1.6459e-02, 1.3115e-01],\n", + " [ 5.0467e-02, 2.3102e-03, 1.0794e-01, ..., -1.4375e-02,\n", + " -9.6894e-04, 9.6666e-02],\n", + " [ 6.6078e-02, 6.9323e-02, 9.1618e-02, ..., -4.4344e-02,\n", + " 4.2789e-02, 1.2422e-01]],\n", + " \n", + " [[-2.9407e-02, -2.0045e-02, 3.1147e-02, ..., 1.2786e-01,\n", + " 6.0353e-02, -1.2973e-02],\n", + " [ 1.0998e-01, 1.1936e-02, 1.2633e-01, ..., 6.6870e-03,\n", + " 1.0851e-01, 7.3025e-02],\n", + " [ 5.9735e-02, 1.0367e-01, 1.1030e-01, ..., -3.0993e-03,\n", + " 1.0475e-02, 1.4001e-01],\n", + " ...,\n", + " [-9.0474e-03, 2.2540e-01, 3.4040e-01, ..., 2.3693e-01,\n", + " 1.1339e-01, 1.0653e-01],\n", + " [-2.9985e-02, -2.3190e-02, -1.2345e-01, ..., 9.8696e-03,\n", + " -4.9102e-02, -6.0093e-02],\n", + " [-1.6822e-03, 6.7063e-02, -6.0226e-02, ..., 2.2233e-02,\n", + " 4.3452e-03, -4.0438e-02]],\n", + " \n", + " [[-1.0691e-01, -5.6675e-02, -4.5410e-02, ..., -8.4990e-02,\n", + " -5.5654e-02, -7.7370e-02],\n", + " [-8.9281e-02, -1.3529e-03, 2.5868e-03, ..., -7.0592e-02,\n", + " -2.9084e-02, -4.2487e-02],\n", + " [-1.1462e-01, -5.0914e-02, -8.2663e-02, ..., -1.0681e-01,\n", + " -5.4781e-02, -3.6654e-02],\n", + " ...,\n", + " [-9.7046e-02, -6.1813e-02, -2.8259e-02, ..., -4.3565e-02,\n", + " -2.9207e-02, -2.6191e-02],\n", + " [-6.1087e-02, 2.2289e-02, -1.0171e-03, ..., -9.0176e-02,\n", + " -4.9931e-02, -4.9930e-02],\n", + " [-5.0141e-02, -1.5549e-02, -5.3237e-02, ..., -6.3770e-02,\n", + " -6.9067e-02, -8.0556e-02]]]]),\n", + " 'deconv3.conv2d.bias': tensor([0.1373, 0.1139, 0.1075])}" + ] + }, + "execution_count": 4, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "dict(style_model.state_dict())" + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "id": "59050ad9", + "metadata": {}, + "outputs": [], + "source": [ + "torch.save(dict(style_model.state_dict()), \"state_dict_raw.pt\")\n" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "id": "2aad69bd", + "metadata": {}, + "outputs": [], + "source": [] + } + ], + "metadata": { + "kernelspec": { + "display_name": ".venv", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.12.9" + } + }, + "nbformat": 4, + "nbformat_minor": 5 +} diff --git a/demos/video/cpp-model-construction/net.cpp b/demos/video/cpp-model-construction/net.cpp new file mode 100644 index 00000000..3ccc4767 --- /dev/null +++ b/demos/video/cpp-model-construction/net.cpp @@ -0,0 +1,28 @@ +#include "transformer_net.hpp" + + + + + +int main() { + torch::manual_seed(0); + std::cout << "set seed\n"; + TransformerNet model; + std::cout << "TransformerNet model created.\n"; + + model->load_parameters("/Users/iainmoncrief/Documents/Github/ChAI/demos/video/cpp-model-construction/state_dict_raw.pt"); + // torch::serialize::InputArchive archive; + // archive.load_from("/Users/iainmoncrief/Documents/Github/ChAI/demos/video/cpp-model-construction/incomplete_sunday_afternoon.model"); // load the raw weights :contentReference[oaicite:2]{index=2} + // std::cout << "Loading model from archive...\n"; + // model->load(archive); + // std::cout << "Model loaded successfully!\n"; + // model->eval(); + std::cout << "Model is in evaluation mode.\n"; + + // dummy input + auto input = torch::randn({1,3,256,256}); + std::cout << "Input shape: " << input.sizes() << "\n"; + auto output = model->forward(input); + + std::cout << "Output shape: " << output.sizes() << "\n"; +} \ No newline at end of file diff --git a/demos/video/cpp-model-construction/state_dict_raw.pt b/demos/video/cpp-model-construction/state_dict_raw.pt new file mode 100644 index 00000000..ee9baeeb Binary files /dev/null and b/demos/video/cpp-model-construction/state_dict_raw.pt differ diff --git a/demos/video/cpp-model-construction/transformer_net.hpp b/demos/video/cpp-model-construction/transformer_net.hpp new file mode 100644 index 00000000..e7296595 --- /dev/null +++ b/demos/video/cpp-model-construction/transformer_net.hpp @@ -0,0 +1,207 @@ +// transformer_net.h +#pragma once +#include +#include +#include +#include +#include +// +// --- ConvLayer ------------------------------------------------------------- +// +struct ConvLayerImpl : torch::nn::Module { + torch::nn::ReflectionPad2d reflection_pad{nullptr}; + torch::nn::Conv2d conv2d{nullptr}; + + ConvLayerImpl(int64_t in_channels, + int64_t out_channels, + int64_t kernel_size, + int64_t stride) + : reflection_pad(torch::nn::ReflectionPad2dOptions(kernel_size / 2)), + conv2d(torch::nn::Conv2dOptions(in_channels, out_channels, kernel_size) + .stride(stride)) + { + register_module("reflection_pad", reflection_pad); + register_module("conv2d", conv2d); + } + + torch::Tensor forward(torch::Tensor x) { + x = reflection_pad->forward(x); + x = conv2d->forward(x); + return x; + } +}; +TORCH_MODULE(ConvLayer); + +// +// --- ResidualBlock --------------------------------------------------------- +// +struct ResidualBlockImpl : torch::nn::Module { + ConvLayer conv1{nullptr}; + torch::nn::InstanceNorm2d in1{nullptr}; + ConvLayer conv2{nullptr}; + torch::nn::InstanceNorm2d in2{nullptr}; + torch::nn::ReLU relu{nullptr}; + + ResidualBlockImpl(int64_t channels) + : conv1(ConvLayer(channels, channels, 3, 1)), + in1(torch::nn::InstanceNorm2dOptions(channels).affine(true)), + conv2(ConvLayer(channels, channels, 3, 1)), + in2(torch::nn::InstanceNorm2dOptions(channels).affine(true)), + relu(torch::nn::ReLUOptions(true)) + { + register_module("conv1", conv1); + register_module("in1", in1); + register_module("conv2", conv2); + register_module("in2", in2); + register_module("relu", relu); + } + + torch::Tensor forward(torch::Tensor x) { + auto residual = x; + auto out = relu->forward(in1->forward(conv1->forward(x))); + out = in2->forward(conv2->forward(out)); + return out + residual; + } +}; +TORCH_MODULE(ResidualBlock); + +// +// --- UpsampleConvLayer ----------------------------------------------------- +// +struct UpsampleConvLayerImpl : torch::nn::Module { + torch::nn::Upsample upsample{nullptr}; + torch::nn::ReflectionPad2d reflection_pad{nullptr}; + torch::nn::Conv2d conv2d{nullptr}; + + UpsampleConvLayerImpl(int64_t in_channels, + int64_t out_channels, + int64_t kernel_size, + int64_t stride, + int64_t upsample_scale) + : upsample(torch::nn::UpsampleOptions() + .scale_factor(std::vector{(double)upsample_scale, (double)upsample_scale}) + .mode(torch::kNearest)), + reflection_pad(torch::nn::ReflectionPad2dOptions(kernel_size / 2)), + conv2d(torch::nn::Conv2dOptions(in_channels, out_channels, kernel_size) + .stride(stride)) + { + register_module("upsample", upsample); + register_module("reflection_pad", reflection_pad); + register_module("conv2d", conv2d); + } + + torch::Tensor forward(torch::Tensor x) { + x = upsample->forward(x); + x = reflection_pad->forward(x); + x = conv2d->forward(x); + return x; + } +}; +TORCH_MODULE(UpsampleConvLayer); + +// +// --- TransformerNet -------------------------------------------------------- +// +struct TransformerNetImpl : torch::nn::Module { + ConvLayer conv1{nullptr}; + torch::nn::InstanceNorm2d in1{nullptr}; + ConvLayer conv2{nullptr}; + torch::nn::InstanceNorm2d in2{nullptr}; + ConvLayer conv3{nullptr}; + torch::nn::InstanceNorm2d in3{nullptr}; + ResidualBlock res1{nullptr}, res2{nullptr}, res3{nullptr}, + res4{nullptr}, res5{nullptr}; + UpsampleConvLayer deconv1{nullptr}, deconv2{nullptr}; + torch::nn::InstanceNorm2d in4{nullptr}, in5{nullptr}; + ConvLayer deconv3{nullptr}; + torch::nn::ReLU relu{nullptr}; + + TransformerNetImpl() + : conv1(ConvLayer( 3, 32, 9, 1)), + in1 (torch::nn::InstanceNorm2dOptions(32).affine(true)), + conv2(ConvLayer( 32, 64, 3, 2)), + in2 (torch::nn::InstanceNorm2dOptions(64).affine(true)), + conv3(ConvLayer( 64, 128, 3, 2)), + in3 (torch::nn::InstanceNorm2dOptions(128).affine(true)), + res1 (ResidualBlock(128)), + res2 (ResidualBlock(128)), + res3 (ResidualBlock(128)), + res4 (ResidualBlock(128)), + res5 (ResidualBlock(128)), + deconv1(UpsampleConvLayer(128, 64, 3, 1, 2)), + in4 (torch::nn::InstanceNorm2dOptions(64).affine(true)), + deconv2(UpsampleConvLayer( 64, 32, 3, 1, 2)), + in5 (torch::nn::InstanceNorm2dOptions(32).affine(true)), + deconv3(ConvLayer( 32, 3, 9, 1)), + relu (torch::nn::ReLUOptions(true)) + { + register_module("conv1", conv1); + register_module("in1", in1); + register_module("conv2", conv2); + register_module("in2", in2); + register_module("conv3", conv3); + register_module("in3", in3); + register_module("res1", res1); + register_module("res2", res2); + register_module("res3", res3); + register_module("res4", res4); + register_module("res5", res5); + register_module("deconv1", deconv1); + register_module("in4", in4); + register_module("deconv2", deconv2); + register_module("in5", in5); + register_module("deconv3", deconv3); + register_module("relu", relu); + } + + torch::Tensor forward(torch::Tensor x) { + x = relu->forward(in1->forward(conv1->forward(x))); + x = relu->forward(in2->forward(conv2->forward(x))); + x = relu->forward(in3->forward(conv3->forward(x))); + x = res1->forward(x); + x = res2->forward(x); + x = res3->forward(x); + x = res4->forward(x); + x = res5->forward(x); + x = relu->forward(in4->forward(deconv1->forward(x))); + x = relu->forward(in5->forward(deconv2->forward(x))); + x = deconv3->forward(x); + return x; + } + + // Model class is inherited from public nn::Module + std::vector get_the_bytes(std::string filename) { + std::ifstream input(filename, std::ios::binary); + std::vector bytes( + (std::istreambuf_iterator(input)), + (std::istreambuf_iterator())); + + input.close(); + return bytes; + } + + void load_parameters(std::string pt_pth) { + std::vector f = this->get_the_bytes(pt_pth); + c10::Dict weights = torch::pickle_load(f).toGenericDict(); + + const torch::OrderedDict& model_params = this->named_parameters(); + std::vector param_names; + for (auto const& w : model_params) { + param_names.push_back(w.key()); + } + + torch::NoGradGuard no_grad; + for (auto const& w : weights) { + std::string name = w.key().toStringRef(); + at::Tensor param = w.value().toTensor(); + + if (std::find(param_names.begin(), param_names.end(), name) != param_names.end()){ + model_params.find(name)->copy_(param); + } else { + std::cout << name << " does not exist among model parameters." << std::endl; + }; + + } + } +}; +TORCH_MODULE(TransformerNet); diff --git a/demos/video/cpp-model-construction/transformer_net.py b/demos/video/cpp-model-construction/transformer_net.py new file mode 100644 index 00000000..26d4e1c6 --- /dev/null +++ b/demos/video/cpp-model-construction/transformer_net.py @@ -0,0 +1,104 @@ +import torch + + +class TransformerNet(torch.nn.Module): + def __init__(self): + super(TransformerNet, self).__init__() + # Initial convolution layers + self.conv1 = ConvLayer(3, 32, kernel_size=9, stride=1) + self.in1 = torch.nn.InstanceNorm2d(32, affine=True) + self.conv2 = ConvLayer(32, 64, kernel_size=3, stride=2) + self.in2 = torch.nn.InstanceNorm2d(64, affine=True) + self.conv3 = ConvLayer(64, 128, kernel_size=3, stride=2) + self.in3 = torch.nn.InstanceNorm2d(128, affine=True) + # Residual layers + self.res1 = ResidualBlock(128) + self.res2 = ResidualBlock(128) + self.res3 = ResidualBlock(128) + self.res4 = ResidualBlock(128) + self.res5 = ResidualBlock(128) + # Upsampling Layers + self.deconv1 = UpsampleConvLayer(128, 64, kernel_size=3, stride=1, upsample=2) + self.in4 = torch.nn.InstanceNorm2d(64, affine=True) + self.deconv2 = UpsampleConvLayer(64, 32, kernel_size=3, stride=1, upsample=2) + self.in5 = torch.nn.InstanceNorm2d(32, affine=True) + self.deconv3 = ConvLayer(32, 3, kernel_size=9, stride=1) + # Non-linearities + self.relu = torch.nn.ReLU() + + def forward(self, X): + y = self.relu(self.in1(self.conv1(X))) + y = self.relu(self.in2(self.conv2(y))) + y = self.relu(self.in3(self.conv3(y))) + y = self.res1(y) + y = self.res2(y) + y = self.res3(y) + y = self.res4(y) + y = self.res5(y) + y = self.relu(self.in4(self.deconv1(y))) + y = self.relu(self.in5(self.deconv2(y))) + y = self.deconv3(y) + return y + + +class ConvLayer(torch.nn.Module): + def __init__(self, in_channels, out_channels, kernel_size, stride): + super(ConvLayer, self).__init__() + reflection_padding = kernel_size // 2 + self.reflection_pad = torch.nn.ReflectionPad2d(reflection_padding) + self.conv2d = torch.nn.Conv2d(in_channels, out_channels, kernel_size, stride) + + def forward(self, x): + out = self.reflection_pad(x) + out = self.conv2d(out) + return out + + +class ResidualBlock(torch.nn.Module): + """ResidualBlock + introduced in: https://arxiv.org/abs/1512.03385 + recommended architecture: http://torch.ch/blog/2016/02/04/resnets.html + """ + + def __init__(self, channels): + super(ResidualBlock, self).__init__() + self.conv1 = ConvLayer(channels, channels, kernel_size=3, stride=1) + self.in1 = torch.nn.InstanceNorm2d(channels, affine=True) + self.conv2 = ConvLayer(channels, channels, kernel_size=3, stride=1) + self.in2 = torch.nn.InstanceNorm2d(channels, affine=True) + self.relu = torch.nn.ReLU() + + def forward(self, x): + residual = x + out = self.relu(self.in1(self.conv1(x))) + out = self.in2(self.conv2(out)) + out = out + residual + return out + + +class UpsampleConvLayer(torch.nn.Module): + """UpsampleConvLayer + Upsamples the input and then does a convolution. This method gives better results + compared to ConvTranspose2d. + ref: http://distill.pub/2016/deconv-checkerboard/ + """ + + def __init__(self, in_channels, out_channels, kernel_size, stride, upsample): + super(UpsampleConvLayer, self).__init__() + # self.upsample = upsample + self.upsample = torch.nn.Upsample(scale_factor=2, mode='nearest') + reflection_padding = kernel_size // 2 + self.reflection_pad = torch.nn.ReflectionPad2d(reflection_padding) + self.conv2d = torch.nn.Conv2d(in_channels, out_channels, kernel_size, stride) + + def forward(self, x): + x_in = x + # print('upsample', self.upsample) + # x_in = torch.nn.functional.interpolate(x_in, mode='nearest', scale_factor=self.upsample) + # if self.upsample: + # x_in = torch.nn.functional.interpolate(x_in, mode='nearest', scale_factor=self.upsample) + out = self.upsample(x_in) + out = self.reflection_pad(out) + # out = self.reflection_pad(out.to(torch.float32)).to(x.dtype) + out = self.conv2d(out) + return out diff --git a/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float16.pt b/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float16.pt new file mode 100644 index 00000000..89c23f70 Binary files /dev/null and b/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float16.pt differ diff --git a/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float32.pt b/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float32.pt new file mode 100644 index 00000000..0ec6834f Binary files /dev/null and b/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float32.pt differ diff --git a/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float16.pt b/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float16.pt new file mode 100644 index 00000000..469eb9d8 Binary files /dev/null and b/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float16.pt differ diff --git a/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float32.pt b/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float32.pt new file mode 100644 index 00000000..59584ebe Binary files /dev/null and b/demos/video/style-transfer/models/exports/cpu/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float32.pt differ diff --git a/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float16.pt b/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float16.pt new file mode 100644 index 00000000..4ba2e334 Binary files /dev/null and b/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float16.pt differ diff --git a/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float32.pt b/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float32.pt new file mode 100644 index 00000000..38dffab2 Binary files /dev/null and b/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000_float32.pt differ diff --git a/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float16.pt b/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float16.pt new file mode 100644 index 00000000..d56fb917 Binary files /dev/null and b/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float16.pt differ diff --git a/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float32.pt b/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float32.pt new file mode 100644 index 00000000..85fe26ba Binary files /dev/null and b/demos/video/style-transfer/models/exports/mps/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000_float32.pt differ diff --git a/demos/video/style-transfer/saved_models/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000.model b/demos/video/style-transfer/saved_models/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000.model new file mode 100644 index 00000000..3d4e59d6 Binary files /dev/null and b/demos/video/style-transfer/saved_models/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_12000.model differ diff --git a/demos/video/style-transfer/saved_models/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000.model b/demos/video/style-transfer/saved_models/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000.model new file mode 100644 index 00000000..2576c9aa Binary files /dev/null and b/demos/video/style-transfer/saved_models/sunday_afternoon_ep3_bt4_sw3e10_cw_1e5_flash_epoch_0_batch_id_20000.model differ diff --git a/demos/video/style-transfer/style_transfer_test.py b/demos/video/style-transfer/style_transfer_test.py index 7bc5436b..a993d059 100644 --- a/demos/video/style-transfer/style_transfer_test.py +++ b/demos/video/style-transfer/style_transfer_test.py @@ -139,7 +139,19 @@ def torch_device(device_name): - frame_rgb = cv2.cvtColor(frame_bgr, cv2.COLOR_BGR2RGB) + + height, width = frame_bgr.shape[:2] + + scale_factor = 0.5 + + # Calculate the new dimensions + new_width = int(width * scale_factor) + new_height = int(height * scale_factor) + + + + frame_rgb = cv2.cvtColor(cv2.resize(frame_bgr,(new_width,new_height)), cv2.COLOR_BGR2RGB) + # 3) Ensure the array is contiguous (torch needs it) ------------------------- frame_rgb = np.ascontiguousarray(frame_rgb) @@ -185,6 +197,7 @@ def torch_device(device_name): # frame_bgr_out = tensor_to_bgr(output_tensor, undo_normalise=True,mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) frame_bgr_out = utils.tensor_to_bgr(output_tensor) + frame_bgr_out = cv2.resize(frame_bgr_out, (width, height)) if args.show_output or args.use_webcam: diff --git a/lib/Bridge.chpl b/lib/Bridge.chpl index f0b813c0..afd46edd 100644 --- a/lib/Bridge.chpl +++ b/lib/Bridge.chpl @@ -10,14 +10,17 @@ module Bridge { use CTypes; - extern record bridge_tensor_t { var data: c_ptr(real(32)); - var sizes: c_ptr(int(32)); - var dim: int(32); + var sizes: c_ptr(uint(32)); + var dim: uint(32); var created_by_c: bool; + var was_freed: bool; } + extern "free_bridge_tensor" proc freeBridgeTensorHandle( + in tensor: bridge_tensor_t): void; + extern record bridge_pt_model_t { var pt_module: c_ptr(void); } @@ -73,10 +76,6 @@ module Bridge { in model: bridge_pt_model_t, in input: bridge_tensor_t): bridge_tensor_t; - extern "model_forward_style_transfer" proc modelForwardStyleTransfer( - in model: bridge_pt_model_t, - in input: bridge_tensor_t): bridge_tensor_t; - extern "accelerator_available" proc acceleratorAvailable(): bool; @@ -141,7 +140,9 @@ module Bridge { return shape; } - proc bridgeTensorToArray(param rank: int, package: bridge_tensor_t): [] real(32) { + proc bridgeTensorToArray(param rank: int,package: bridge_tensor_t): [] real(32) { + if package.was_freed then + util.err("BridgeTensorToArray: Tensor has already been freed"); const shape = bridgeTensorShape(rank, package); const dom = util.domainFromShape((...shape)); var result: [dom] real(32); @@ -152,27 +153,38 @@ module Bridge { // deallocate(package.data); // deallocate(package.sizes); // } + freeBridgeTensor(package); return result; } - proc bridgeTensorToExistingArray(ref existing: [] real(32), package: bridge_tensor_t) { + proc bridgeTensorToExistingArray(ref existing: [] real(32),package: bridge_tensor_t) { + if package.was_freed then + util.err("BridgeTensorToArray: Tensor has already been freed"); const shape = bridgeTensorShape(existing.rank, package); if existing.shape != shape then util.err("BridgeTensorToExistingArray: Shape mismatch"); const dom = existing.domain; forall (i,idx) in dom.everyZip() do existing[idx] = package.data[i]; - if package.created_by_c { - deallocate(package.data); - deallocate(package.sizes); + freeBridgeTensor(package); + // if package.created_by_c { + // deallocate(package.data); + // deallocate(package.sizes); + // } + } + + proc freeBridgeTensor(handle: bridge_tensor_t) { + if handle.created_by_c && !handle.was_freed { + freeBridgeTensorHandle(handle); + // handle.was_freed = true; } } proc createBridgeTensor(const ref data: [] real(32)): bridge_tensor_t { var result: bridge_tensor_t; result.data = c_ptrToConst(data) : c_ptr(real(32)); - result.sizes = allocate(int(32),data.rank); + result.sizes = allocate(uint(32),data.rank); result.created_by_c = false; const sizeArr = getSizeArray(data); for i in 0..