Issue for cmt_training.py

[ting-chuan-wang]

I run your code, and got an error, please help me. Thank you.

python cmt_training.py --project_path "./results/0927/" --data_path "/home/tingw/mne_data/physionet-sleep-data/multi_epoch/" --train_data_list [0,1,2,3] --val_data_list [4] --model_type "Seq"

The error message is,
Traceback (most recent call last):
File "/home/tingw/CMTrans/Cross-Modal-Transformer/cmt_training.py", line 155, in
main()
File "/home/tingw/CMTrans/Cross-Modal-Transformer/cmt_training.py", line 92, in main
train_data_loader, val_data_loader = get_dataset(device,args)
File "/home/tingw/CMTrans/Cross-Modal-Transformer/datasets/sleep_edf.py", line 412, in get_dataset
train_dataset = SleepEDF_Seq_MultiChan_Dataset_Main(eeg_file = train_eeg_list ,
File "/home/tingw/CMTrans/Cross-Modal-Transformer/datasets/sleep_edf.py", line 223, in init
bin_labels = np.bincount(self.labels)
ValueError: object too deep for desired array

== LOG below == 
Torch Version : 1.10.0+cu113
Training Arguements ====================================>
project_path : ./results/0927/
data_path : /home/tingw/mne_data/physionet-sleep-data/multi_epoch/
train_data_list : ['[0,1,2,3]']
val_data_list : ['[4]']
is_retrain : False
model_path :
save_model_freq : 50
model_type : Seq
d_model : 128
dim_feedforward : 512
window_size : 50
num_seq : 15
batch_size : 32
weigths : [1.0, 2.0, 1.0, 2.0, 2.0]
lr : 0.001
beta_1 : 0.9
beta_2 : 0.999
eps : 1e-09
weight_decay : 0.0001
n_epochs : 200
step_size : 30
gamma : 0.5
is_neptune : False
nep_project :
nep_api :
Project directory created at ./results/0927/
Getting Dataset ===================================>
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x1.h5', '/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x2.h5', '/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x3.h5', '/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x4.h5', '/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x5.h5']
Training Data Files: ===========================>
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/y1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/y2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/y3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/y4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/std1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/std2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/std3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/std4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std4.h5']
Validation Data Files: ===========================>
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/y5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/std5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std5.h5']
Loading Train Data for many-to-many classification ==================================>
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/x1.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 43455
Shape of each data : (43455, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog1.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 43455
Shape of each data : (43455, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/y1.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 43455
Shape of each data : (43455, 5)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/x2.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 39919
Shape of each data : (39919, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog2.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 39919
Shape of each data : (39919, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/y2.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 39919
Shape of each data : (39919, 5)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/x3.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 35112
Shape of each data : (35112, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog3.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 35112
Shape of each data : (35112, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/y3.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 35112
Shape of each data : (35112, 5)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/x4.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 40211
Shape of each data : (40211, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog4.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 40211
Shape of each data : (40211, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/y4.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 40211
Shape of each data : (40211, 5)
Traceback (most recent call last):
File "/home/tingw/CMTrans/Cross-Modal-Transformer/cmt_training.py", line 155, in
main()
File "/home/tingw/CMTrans/Cross-Modal-Transformer/cmt_training.py", line 92, in main
train_data_loader, val_data_loader = get_dataset(device,args)
File "/home/tingw/CMTrans/Cross-Modal-Transformer/datasets/sleep_edf.py", line 412, in get_dataset
train_dataset = SleepEDF_Seq_MultiChan_Dataset_Main(eeg_file = train_eeg_list ,
File "/home/tingw/CMTrans/Cross-Modal-Transformer/datasets/sleep_edf.py", line 223, in init
bin_labels = np.bincount(self.labels)
ValueError: object too deep for desired array

[Jathurshan0330]

Can you comment out the bin_labels? It is there to check the counts of different classes. I hope it works

[ting-chuan-wang]

Thanks.
I commented out the bin_labels. Now it stops at another place.
b,_, _ = x.shape
ValueError: too many values to unpack (expected 3)

Below I found the issue is that x.shape become 4D data:
def forward(self, x: Tensor) -> Tensor:
print(">>> forward() In shape: ", x.shape) ### [32, 3000, 5]
if x.shape[0]!=1:
x = x.squeeze().unsqueeze(dim = 1) ### output [32, 1, 3000, 5]

If I change it to
b,_, _, _ = x.shape
The next line is not working.
x_1 = self.projection_1(x)

Please help. Thanks.

== LOG below ==
Torch Version : 1.10.0+cu113
Training Arguements ====================================>
project_path : ./results/0930/
data_path : /home/tingw/mne_data/physionet-sleep-data/multi_epoch/
train_data_list : ['[0,1,2,3]']
val_data_list : ['[4]']
is_retrain : False
model_path :
save_model_freq : 50
model_type : Seq
d_model : 128
dim_feedforward : 512
window_size : 50
num_seq : 15
batch_size : 32
weigths : [1.0, 2.0, 1.0, 2.0, 2.0]
lr : 0.001
beta_1 : 0.9
beta_2 : 0.999
eps : 1e-09
weight_decay : 0.0001
n_epochs : 200
step_size : 30
gamma : 0.5
is_neptune : False
nep_project :
nep_api :
Project directory already available at ./results/0930/
Getting Dataset ===================================>
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x1.h5', '/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x2.h5', '/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x3.h5', '/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x4.h5', '/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x5.h5']
Training Data Files: ===========================>
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/y1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/y2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/y3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/y4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/std1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/std2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/std3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/std4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m4.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std1.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std2.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std3.h5'
'/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std4.h5']
Validation Data Files: ===========================>
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/x5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/y5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/std5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m5.h5']
['/home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std5.h5']
Loading Train Data for many-to-many classification ==================================>
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/x1.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 43455
Shape of each data : (43455, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog1.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 43455
Shape of each data : (43455, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/y1.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 43455
Shape of each data : (43455, 5)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/x2.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 39919
Shape of each data : (39919, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog2.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 39919
Shape of each data : (39919, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/y2.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 39919
Shape of each data : (39919, 5)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/x3.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 35112
Shape of each data : (35112, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog3.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 35112
Shape of each data : (35112, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/y3.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 35112
Shape of each data : (35112, 5)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/x4.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 40211
Shape of each data : (40211, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog4.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 40211
Shape of each data : (40211, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/y4.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 40211
Shape of each data : (40211, 5)
Shape of EEG : (158697, 5, 3000) , EOG : (158697, 5, 3000)
Shape of Labels : torch.Size([158697, 5])
Reading Subject wise mean and sd
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean1.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 43455
Shape of each data : (43455,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/std1.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 43455
Shape of each data : (43455,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m1.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 43455
Shape of each data : (43455,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std1.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 43455
Shape of each data : (43455,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean2.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 39919
Shape of each data : (39919,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/std2.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 39919
Shape of each data : (39919,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m2.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 39919
Shape of each data : (39919,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std2.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 39919
Shape of each data : (39919,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean3.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 35112
Shape of each data : (35112,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/std3.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 35112
Shape of each data : (35112,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m3.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 35112
Shape of each data : (35112,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std3.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 35112
Shape of each data : (35112,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean4.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 40211
Shape of each data : (40211,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/std4.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 40211
Shape of each data : (40211,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m4.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 40211
Shape of each data : (40211,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std4.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 40211
Shape of each data : (40211,)
Shapes of Mean : EEG: (158697,), EOG : (158697,)
Shapes of Sd : EEG: (158697,), EOG : (158697,)
Loading Val Data for many-to-many classification ==================================>
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/x5.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 36485
Shape of each data : (36485, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog5.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 36485
Shape of each data : (36485, 5, 3000)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/y5.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 36485
Shape of each data : (36485, 5)
Shape of EEG : (36485, 5, 3000) , EOG : (36485, 5, 3000)
Shape of Labels : torch.Size([36485, 5])
Reading Subject wise mean and sd
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/mean5.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 36485
Shape of each data : (36485,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/std5.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 36485
Shape of each data : (36485,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_m5.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 36485
Shape of each data : (36485,)
Reading from /home/tingw/mne_data/physionet-sleep-data/multi_epoch/eog_std5.h5 ====================================================
Keys in the h5py file : <KeysViewHDF5 ['data']>
Number of samples : 36485
Shape of each data : (36485,)
Shapes of Mean : EEG: (36485,), EOG : (36485,)
Shapes of Sd : EEG: (36485,), EOG : (36485,)
EEG batch shape: torch.Size([32, 3000, 15, 5])
EOG batch shape: torch.Size([32, 3000, 15, 5])
Labels batch shape: torch.Size([32, 15, 5])
EEG batch shape: torch.Size([32, 3000, 15, 5])
EOG batch shape: torch.Size([32, 3000, 15, 5])
Labels batch shape: torch.Size([32, 15, 5])
EEG Minimum :-11.41333721330944
EEG Maximum :11.627864012828486
EOG Minimum :-13.291103170514944
EOG Maximum :12.179615041092605
EEG Mean :-9.066378188759094e-05
EEG Standard Deviation :0.9713700974582382
EOG Mean :4.682988494505859e-05
EOG Standard Deviation :0.9103014239304752
Initializing Sequence Cross Modal Transformer ==================>
============================= Training Epoch : [1/200]===============================>

forward() In shape: torch.Size([32, 3000, 5])
forward() shape: torch.Size([32, 1, 3000, 5])
Traceback (most recent call last):
File "/home/tingw/CMTrans/Cross-Modal-Transformer/cmt_training.py", line 155, in
main()
File "/home/tingw/CMTrans/Cross-Modal-Transformer/cmt_training.py", line 152, in main
train_seq_cmt(Net, train_data_loader, val_data_loader, criterion, optimizer, lr_scheduler, device, args)
File "/home/tingw/CMTrans/Cross-Modal-Transformer/models/sequence_cmt.py", line 328, in train_seq_cmt
outputs,_ = Net(eeg.float().to(device), eog.float().to(device))
File "/home/tingw/anaconda3/envs/CMTrans/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/home/tingw/CMTrans/Cross-Modal-Transformer/models/sequence_cmt.py", line 196, in forward
epoch_1,feat_1 = self.epoch_1(eeg[:,:,0,:],eog[:,:,0,:])#[0]
File "/home/tingw/anaconda3/envs/CMTrans/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/home/tingw/CMTrans/Cross-Modal-Transformer/models/sequence_cmt.py", line 39, in forward
self_eeg = self.eeg_atten(eeg)
File "/home/tingw/anaconda3/envs/CMTrans/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
return forward_call(*input, **kwargs)
File "/home/tingw/CMTrans/Cross-Modal-Transformer/models/model_blocks.py", line 147, in forward
src = self.window_embed(x)
File "/home/tingw/anaconda3/envs/CMTrans/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1102, in call_impl
return forward_call(*input, **kwargs)
File "/home/tingw/CMTrans/Cross-Modal-Transformer/models/model_blocks.py", line 107, in forward
b,, _ = x.shape
ValueError: too many values to unpack (expected 3)

[Jathurshan0330]

x should be three dimensional, can you share the shape of x given as input

[ting-chuan-wang]

The input x was [32, 3000, 5]
but
x = x.squeeze().unsqueeze(dim = 1)
change it to 4D [32, 1, 3000, 5]

[Yumi0606]

same problem,
Our team is reproducing the seq_model. However, when using the dataset processed by muti_epoch.py, we encounter this error. But the dataset processed by single_epoch.py can be used for training both the epoch_model and seq_model. So, is muti_epoch.py error-free?