diff --git a/notebooks/inference.ipynb b/notebooks/inference.ipynb
index dbc84c2..59e44be 100644
--- a/notebooks/inference.ipynb
+++ b/notebooks/inference.ipynb
@@ -1,206 +1,430 @@
 {
- "cells": [
-  {
-   "cell_type": "markdown",
-   "id": "0",
-   "metadata": {},
-   "source": [
-    "# Alpamayo-R1 Demo"
-   ]
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "id": "0",
+      "metadata": {},
+      "source": [
+        "# Alpamayo 1 Demo"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "id": "1",
+      "metadata": {},
+      "source": [
+        "This notebook will load some example data from the NVIDIA [PhysicalAI-AV Dataset](https://huggingface.co/datasets/nvidia/PhysicalAI-Autonomous-Vehicles) and run the Alpamayo 1 model on it, producing and visualizing output trajectories and associated reasoning traces."
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "2",
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "import copy\n",
+        "import numpy as np\n",
+        "import pandas as pd\n",
+        "\n",
+        "import torch\n",
+        "from alpamayo_r1.models.alpamayo_r1 import AlpamayoR1\n",
+        "from alpamayo_r1.load_physical_aiavdataset import load_physical_aiavdataset\n",
+        "from alpamayo_r1 import helper"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "id": "3",
+      "metadata": {},
+      "source": [
+        "### Load model and construct data preprocessor"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "1199ecb4",
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "# check the gpu\n",
+        "!nvidia-smi"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "4",
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "model = AlpamayoR1.from_pretrained(\"nvidia/Alpamayo-R1-10B\", dtype=torch.bfloat16).to(\"cuda\")\n",
+        "processor = helper.get_processor(model.tokenizer)"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "id": "5",
+      "metadata": {},
+      "source": [
+        "### Load and prepare data"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "6",
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "clip_ids = pd.read_parquet(\"clip_ids.parquet\")[\"clip_id\"].tolist()\n",
+        "clip_id = clip_ids[774]\n",
+        "# examples\n",
+        "# 774 clip_id = '030c760c-ae38-49aa-9ad8-f5650a545d26'\n",
+        "\n",
+        "data = load_physical_aiavdataset(clip_id)\n",
+        "\n",
+        "messages = helper.create_message(data[\"image_frames\"].flatten(0, 1))\n",
+        "\n",
+        "inputs = processor.apply_chat_template(\n",
+        "    messages,\n",
+        "    tokenize=True,\n",
+        "    add_generation_prompt=False,\n",
+        "    continue_final_message=True,\n",
+        "    return_dict=True,\n",
+        "    return_tensors=\"pt\",\n",
+        ")\n",
+        "print(\"seq length:\", inputs.input_ids.shape)\n",
+        "model_inputs = {\n",
+        "    \"tokenized_data\": inputs,\n",
+        "    \"ego_history_xyz\": data[\"ego_history_xyz\"],\n",
+        "    \"ego_history_rot\": data[\"ego_history_rot\"],\n",
+        "}\n",
+        "model_inputs = helper.to_device(model_inputs, \"cuda\")"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "id": "e34c9783",
+      "metadata": {},
+      "source": [
+        "### Inspect input data"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "0afd0b70",
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "print(\"model_inputs:\")\n",
+        "# full input data\n",
+        "for key, value in model_inputs.items():\n",
+        "    if isinstance(value, torch.Tensor):\n",
+        "        print(key, value.shape)\n",
+        "    else:\n",
+        "        print(key, type(value))\n",
+        "\n",
+        "# data fed into model.vlm\n",
+        "print(\"\\nmodel_inputs['tokenized_data']:\")\n",
+        "for key, value in model_inputs[\"tokenized_data\"].items():\n",
+        "    if isinstance(value, torch.Tensor):\n",
+        "        print(key, value.shape)\n",
+        "    else:\n",
+        "        print(key, type(value))\n"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "id": "440734a7",
+      "metadata": {},
+      "source": [
+        "### Visualization of the ego history in BEV"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "a816bd61",
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "import matplotlib.pyplot as plt\n",
+        "\n",
+        "# Extract ego history — (1, 1, T, 3) → (T, 3)\n",
+        "ego_xyz = model_inputs[\"ego_history_xyz\"].cpu().numpy().squeeze()  # (T, 3)\n",
+        "\n",
+        "x, y = ego_xyz[:, 0], ego_xyz[:, 1]  # ignore z\n",
+        "T = len(x)\n",
+        "\n",
+        "# BEV mapping: plot_x = -y (right), plot_y = x (forward up)\n",
+        "px, py = -y, x\n",
+        "\n",
+        "fig, ax = plt.subplots(figsize=(8, 8))\n",
+        "\n",
+        "# Trajectory line + markers\n",
+        "ax.plot(px, py, color=\"royalblue\", linewidth=2, alpha=0.6, zorder=2)\n",
+        "ax.scatter(px, py, c=np.arange(T), cmap=\"viridis\", s=30, zorder=3)\n",
+        "\n",
+        "# Start & End\n",
+        "ax.scatter(px[0], py[0], color=\"green\", s=100, marker=\"D\", zorder=4, label=\"Ego History Start\")\n",
+        "ax.scatter(px[-1], py[-1], color=\"red\", s=100, marker=\"D\", zorder=4, label=\"Ego History End (t0)\")\n",
+        "\n",
+        "ax.set_xlabel(\"Y (right) [m]\")\n",
+        "ax.set_ylabel(\"X (forward) [m]\")\n",
+        "ax.set_title(\"BEV Ego History\")\n",
+        "ax.set_aspect(\"equal\")\n",
+        "\n",
+        "# Make plot square\n",
+        "v_span = py.max() - py.min()\n",
+        "h_span = px.max() - px.min()\n",
+        "span = max(v_span, h_span) * 1.1  # 10% padding\n",
+        "h_mid = (px.max() + px.min()) / 2\n",
+        "v_mid = (py.max() + py.min()) / 2\n",
+        "ax.set_xlim(h_mid - span / 2, h_mid + span / 2)\n",
+        "ax.set_ylim(v_mid - span / 2, v_mid + span / 2)\n",
+        "\n",
+        "ax.legend()\n",
+        "\n",
+        "ax.grid(True, alpha=0.3)\n",
+        "plt.tight_layout()\n",
+        "plt.show()"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "ad7c2dc7",
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "import matplotlib.pyplot as plt\n",
+        "\n",
+        "# Camera names corresponding to the sorted camera indices\n",
+        "# Map camera indices to names (based on avdi.features.CAMERA enum)\n",
+        "CAMERA_INDEX_TO_NAME = {\n",
+        "    0: \"Cross Left\",\n",
+        "    1: \"Front Wide\",\n",
+        "    2: \"Cross Right\",\n",
+        "    6: \"Front Tele\",\n",
+        "}\n",
+        "frame_idx_to_time = [\n",
+        "    f\"t={t - data['relative_timestamps'][-1].tolist()[-1]:.1f}s\"\n",
+        "    for t in data[\"relative_timestamps\"][-1].tolist()\n",
+        "]\n",
+        "camera_names = [\n",
+        "    CAMERA_INDEX_TO_NAME.get(idx.item(), f\"Camera {idx.item()}\") for idx in data[\"camera_indices\"]\n",
+        "]\n",
+        "\n",
+        "# image_frames shape: (N_cameras, num_frames, 3, H, W)\n",
+        "images = (\n",
+        "    data[\"image_frames\"].permute(0, 1, 3, 4, 2).cpu().numpy()\n",
+        ")  # (N_cameras, num_frames, H, W, 3)\n",
+        "n_cameras, n_frames = images.shape[:2]\n",
+        "\n",
+        "fig, axes = plt.subplots(n_cameras, n_frames, figsize=(12, 8))\n",
+        "plt.subplots_adjust(wspace=0.02, hspace=0.1, left=0.12, right=0.98, top=0.92, bottom=0.02)\n",
+        "\n",
+        "for cam_idx in range(n_cameras):\n",
+        "    for frame_idx in range(n_frames):\n",
+        "        ax = axes[cam_idx, frame_idx]\n",
+        "        ax.imshow(images[cam_idx, frame_idx])\n",
+        "        ax.axis(\"off\")\n",
+        "\n",
+        "        # Add column labels (timesteps) at the top row\n",
+        "        if cam_idx == 0:\n",
+        "            ax.set_title(frame_idx_to_time[frame_idx], fontsize=11)\n",
+        "\n",
+        "        # Add row labels (camera names) on the left column\n",
+        "        if frame_idx == 0:\n",
+        "            ax.text(\n",
+        "                -0.05,\n",
+        "                0.5,\n",
+        "                camera_names[cam_idx],\n",
+        "                fontsize=10,\n",
+        "                rotation=0,\n",
+        "                ha=\"right\",\n",
+        "                va=\"center\",\n",
+        "                transform=ax.transAxes,\n",
+        "            )\n",
+        "\n",
+        "plt.show()"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "52a76e4c",
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "# check the input sequence\n",
+        "input_ids = model_inputs[\"tokenized_data\"][\"input_ids\"]\n",
+        "tokenizer = model.tokenizer\n",
+        "input_seq = tokenizer.decode(input_ids[0])\n",
+        "\n",
+        "# replacing special tokens for easier reading\n",
+        "input_seq = input_seq.replace(\"<|image_pad|>\", \"I\")\n",
+        "input_seq = input_seq.replace(\"<|traj_history|>\", \"H\")\n",
+        "input_seq = input_seq.replace(\"<|traj_future|>\", \"F\")\n",
+        "input_seq = input_seq.replace(\"<|vision_end|>\", \"<|vision_end|>\\n\")\n",
+        "input_seq = input_seq.replace(\"<|traj_history_end|>\", \"<|traj_history_end|>\\n\")\n",
+        "\n",
+        "print(input_seq)\n"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "id": "7",
+      "metadata": {},
+      "source": [
+        "### Model inference"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "8",
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "torch.cuda.manual_seed_all(42)\n",
+        "with torch.autocast(\"cuda\", dtype=torch.bfloat16):\n",
+        "    pred_xyz, pred_rot, extra = model.sample_trajectories_from_data_with_vlm_rollout(\n",
+        "        data=copy.deepcopy(model_inputs),\n",
+        "        top_p=0.98,\n",
+        "        temperature=0.6,\n",
+        "        num_traj_samples=4,  # Feel free to raise this for more output trajectories and CoC traces.\n",
+        "        max_generation_length=256,\n",
+        "        return_extra=True,\n",
+        "    )\n",
+        "\n",
+        "# the size is [batch_size, num_traj_sets, num_traj_samples]\n",
+        "print(\"Chain-of-Causation (per trajectory):\\n\", extra[\"cot\"][0])"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "id": "9",
+      "metadata": {},
+      "source": [
+        "## Visualizing data and results"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "11",
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "import matplotlib.pyplot as plt\n",
+        "\n",
+        "# --- Ego history (BEV) ---\n",
+        "hist_xyz = model_inputs[\"ego_history_xyz\"].cpu().numpy().squeeze()  # (T_h, 3)\n",
+        "hx, hy = -hist_xyz[:, 1], hist_xyz[:, 0]  # BEV: plot_x=-y, plot_y=x\n",
+        "\n",
+        "# --- GT future ---\n",
+        "gt_xyz = data[\"ego_future_xyz\"].cpu().numpy().squeeze()  # (T_f, 3)\n",
+        "gx, gy = -gt_xyz[:, 1], gt_xyz[:, 0]\n",
+        "\n",
+        "# --- Predicted futures ---\n",
+        "n_samples = pred_xyz.shape[2]\n",
+        "pred_np = pred_xyz.cpu().numpy()[0, 0]  # (n_samples, T_f, 3)\n",
+        "\n",
+        "fig, ax = plt.subplots(figsize=(8, 8))\n",
+        "\n",
+        "# Ego history\n",
+        "ax.plot(hx, hy, color=\"royalblue\", linewidth=2, alpha=0.6, zorder=2)\n",
+        "ax.scatter(hx, hy, c=np.arange(len(hx)), cmap=\"viridis\", s=30, zorder=3)\n",
+        "ax.scatter(hx[0], hy[0], color=\"green\", s=100, marker=\"D\", zorder=4, label=\"History Start\")\n",
+        "ax.scatter(hx[-1], hy[-1], color=\"red\", s=100, marker=\"D\", zorder=4, label=\"t0\")\n",
+        "\n",
+        "# Predicted trajectories\n",
+        "colors = plt.cm.tab10(np.linspace(0, 1, n_samples))\n",
+        "for i in range(n_samples):\n",
+        "    px_pred = -pred_np[i, :, 1]\n",
+        "    py_pred = pred_np[i, :, 0]\n",
+        "    ax.plot(\n",
+        "        px_pred,\n",
+        "        py_pred,\n",
+        "        \"o-\",\n",
+        "        color=colors[i],\n",
+        "        markersize=4,\n",
+        "        linewidth=1.5,\n",
+        "        alpha=0.8,\n",
+        "        zorder=5,\n",
+        "        label=f\"Pred #{i + 1}\",\n",
+        "    )\n",
+        "\n",
+        "# GT future\n",
+        "ax.plot(gx, gy, \"s-\", color=\"red\", markersize=4, linewidth=2, zorder=6, label=\"GT Future\")\n",
+        "\n",
+        "ax.set_xlabel(\"Y (right) [m]\")\n",
+        "ax.set_ylabel(\"X (forward) [m]\")\n",
+        "ax.set_title(\"BEV: Ego History + Predicted & GT Trajectories\")\n",
+        "ax.set_aspect(\"equal\")\n",
+        "\n",
+        "# Make plot square\n",
+        "all_px = np.concatenate([hx, gx] + [-pred_np[i, :, 1] for i in range(n_samples)])  # horizontal\n",
+        "all_py = np.concatenate([hy, gy] + [pred_np[i, :, 0] for i in range(n_samples)])  # vertical\n",
+        "v_span = all_py.max() - all_py.min()\n",
+        "h_span = all_px.max() - all_px.min()\n",
+        "span = max(v_span, h_span) * 1.1  # 10% padding\n",
+        "h_mid = (all_px.max() + all_px.min()) / 2\n",
+        "v_mid = (all_py.max() + all_py.min()) / 2\n",
+        "ax.set_xlim(h_mid - span / 2, h_mid + span / 2)\n",
+        "ax.set_ylim(v_mid - span / 2, v_mid + span / 2)\n",
+        "\n",
+        "ax.legend()\n",
+        "ax.grid(True, alpha=0.3)\n",
+        "plt.tight_layout()\n",
+        "plt.show()"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "12",
+      "metadata": {},
+      "outputs": [],
+      "source": [
+        "pred_xy = pred_xyz.cpu().numpy()[0, 0, :, :, :2].transpose(0, 2, 1)\n",
+        "gt_xy = data[\"ego_future_xyz\"].cpu().numpy()[0, 0, :, :2].T\n",
+        "diff = np.linalg.norm(pred_xy - gt_xy[None, ...], axis=1).mean(-1)\n",
+        "print(\"minADE:\", diff.min(), \"meters\")"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "id": "3fde18ba",
+      "metadata": {},
+      "outputs": [],
+      "source": []
+    }
+  ],
+  "metadata": {
+    "kernelspec": {
+      "display_name": "ar1_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": "3.12.3"
+    }
   },
-  {
-   "cell_type": "markdown",
-   "id": "1",
-   "metadata": {},
-   "source": [
-    "This notebook will load some example data from the NVIDIA [PhysicalAI-AV Dataset](https://huggingface.co/datasets/nvidia/PhysicalAI-Autonomous-Vehicles) and run the Alpamayo-R1 model on it, producing and visualizing output trajectories and associated reasoning traces."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "2",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import copy\n",
-    "import numpy as np\n",
-    "import mediapy as mp\n",
-    "import pandas as pd\n",
-    "\n",
-    "import torch\n",
-    "from alpamayo_r1.models.alpamayo_r1 import AlpamayoR1\n",
-    "from alpamayo_r1.load_physical_aiavdataset import load_physical_aiavdataset\n",
-    "from alpamayo_r1 import helper"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "3",
-   "metadata": {},
-   "source": [
-    "### Load model and construct data preprocessor"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "4",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "model = AlpamayoR1.from_pretrained(\"nvidia/Alpamayo-R1-10B\", dtype=torch.bfloat16).to(\"cuda\")\n",
-    "processor = helper.get_processor(model.tokenizer)"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "5",
-   "metadata": {},
-   "source": [
-    "### Load and prepare data"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "6",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "clip_ids = pd.read_parquet(\"clip_ids.parquet\")[\"clip_id\"].tolist()\n",
-    "clip_id = clip_ids[774]\n",
-    "# clip_id = '030c760c-ae38-49aa-9ad8-f5650a545d26'\n",
-    "\n",
-    "data = load_physical_aiavdataset(clip_id)\n",
-    "\n",
-    "messages = helper.create_message(data[\"image_frames\"].flatten(0, 1))\n",
-    "\n",
-    "inputs = processor.apply_chat_template(\n",
-    "    messages,\n",
-    "    tokenize=True,\n",
-    "    add_generation_prompt=False,\n",
-    "    continue_final_message=True,\n",
-    "    return_dict=True,\n",
-    "    return_tensors=\"pt\",\n",
-    ")\n",
-    "print(\"seq length:\", inputs.input_ids.shape)\n",
-    "model_inputs = {\n",
-    "    \"tokenized_data\": inputs,\n",
-    "    \"ego_history_xyz\": data[\"ego_history_xyz\"],\n",
-    "    \"ego_history_rot\": data[\"ego_history_rot\"],\n",
-    "}\n",
-    "model_inputs = helper.to_device(model_inputs, \"cuda\")"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "7",
-   "metadata": {},
-   "source": [
-    "### Model inference"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "8",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "torch.cuda.manual_seed_all(42)\n",
-    "with torch.autocast(\"cuda\", dtype=torch.bfloat16):\n",
-    "    pred_xyz, pred_rot, extra = model.sample_trajectories_from_data_with_vlm_rollout(\n",
-    "        data=copy.deepcopy(model_inputs),\n",
-    "        top_p=0.98,\n",
-    "        temperature=0.6,\n",
-    "        num_traj_samples=1,  # Feel free to raise this for more output trajectories and CoC traces.\n",
-    "        max_generation_length=256,\n",
-    "        return_extra=True,\n",
-    "    )\n",
-    "\n",
-    "# the size is [batch_size, num_traj_sets, num_traj_samples]\n",
-    "print(\"Chain-of-Causation (per trajectory):\\n\", extra[\"cot\"][0])"
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "id": "9",
-   "metadata": {},
-   "source": [
-    "## Visualizing data and results"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "10",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "mp.show_images(data[\"image_frames\"].flatten(0, 1).permute(0, 2, 3, 1), columns=4, width=200)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "11",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "import matplotlib.pyplot as plt\n",
-    "\n",
-    "\n",
-    "def rotate_90cc(xy):\n",
-    "    # Rotate (x, y) by 90 deg CCW -> (y, -x)\n",
-    "    return np.stack([-xy[1], xy[0]], axis=0)\n",
-    "\n",
-    "\n",
-    "for i in range(pred_xyz.shape[2]):\n",
-    "    pred_xy = pred_xyz.cpu()[0, 0, i, :, :2].T.numpy()\n",
-    "    pred_xy_rot = rotate_90cc(pred_xy)\n",
-    "    gt_xy = data[\"ego_future_xyz\"].cpu()[0, 0, :, :2].T.numpy()\n",
-    "    gt_xy_rot = rotate_90cc(gt_xy)\n",
-    "    plt.plot(*pred_xy_rot, \"o-\", label=f\"Predicted Trajectory #{i + 1}\")\n",
-    "plt.ylabel(\"y coordinate (meters)\")\n",
-    "plt.xlabel(\"x coordinate (meters)\")\n",
-    "plt.plot(*gt_xy_rot, \"r-\", label=\"Ground Truth Trajectory\")\n",
-    "plt.legend(loc=\"best\")\n",
-    "plt.axis(\"equal\")"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "12",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "pred_xy = pred_xyz.cpu().numpy()[0, 0, :, :, :2].transpose(0, 2, 1)\n",
-    "diff = np.linalg.norm(pred_xy - gt_xy[None, ...], axis=1).mean(-1)\n",
-    "print(\"minADE:\", diff.min(), \"meters\")"
-   ]
-  }
- ],
- "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": "3.12.3"
-  }
- },
- "nbformat": 4,
- "nbformat_minor": 5
+  "nbformat": 4,
+  "nbformat_minor": 5
 }