-
Notifications
You must be signed in to change notification settings - Fork 70
Description
Hi there!
May I ask a question about the neural response prediction part?
I have a neural response array in firing rate (FRarray)along with camera stamps and neural stamps; both camera stamps(CSarray) and neural stamps(NSarray) are in frames. NSarray.shape==FRarray.shape(if ignore the number of neurons); CSarray.shape match with the input video frames (I input a 5 mins data, but the facemap only read in 4.2 mins data, I tried both the size of 5 and 4.2 mins); It gave me this error when I run the neural response prediction, Please help me about this, thank you very much!:
at non-singleton dimension 1
pose data shape (3, 15, 50531)
xy.shape (50531, 15, 2)
outlier filter shape (50531, 15, 2)
x.shape (50531, 30)
Neural target shape: (300002, 10)
Keypoints shape: (50531, 30)
Inputs: (50531, 30)
Neural: (300002, 10)
Using lr, n_iter, weight_decay: 0.001 300 0.0001
Ignoring window icon 4534162 exceeds maximum xcb request length 4194303
Ignoring window icon 4534162 exceeds maximum xcb request length 4194303
Traceback (most recent call last):
File "/home/isocortex/miniconda3/envs/facemap/lib/python3.8/site-packages/facemap/gui/neural_activity_window.py", line 537, in
lambda clicked: self.run_neural_predictions(clicked, dialog)
File "/home/isocortex/miniconda3/envs/facemap/lib/python3.8/site-packages/facemap/gui/neural_activity_window.py", line 653, in run_neural_predictions
) = prediction_utils.get_keypoints_to_neural_varexp(
File "/home/isocortex/miniconda3/envs/facemap/lib/python3.8/site-packages/facemap/neural_prediction/prediction_utils.py", line 448, in get_keypoints_to_neural_varexp
) = model.train_model(
File "/home/isocortex/miniconda3/envs/facemap/lib/python3.8/site-packages/facemap/neural_prediction/neural_model.py", line 184, in train_model
loss = ((y_pred - Y_train[i][n].unsqueeze(0)) ** 2).mean()
RuntimeError: The size of tensor a (4503) must match the size of tensor b (22500) at non-singleton dimension 1