MedARC-AI / fMRI-reconstruction-NSD

fMRI-to-image reconstruction on the NSD dataset.
MIT License
308 stars 44 forks source link

Help wanted #37

Open Chris36839 opened 11 months ago

Chris36839 commented 11 months ago

Hi, I'm encountering the following error while running "Train_MindEye.py" and am unable to resolve it. I hope someone can help me identify the issue and suggest possible solutions. Firstly, I got the following error: Traceback (most recent call last): File "/mnt/d/Files/Codes/fMRI-reconstruction-NSD/src/Train_MindEye.py", line 585, in for train_i, (voxel, image, coco) in enumerate(train_dl): File "/root/miniconda3/envs/mindeye/lib/python3.10/site-packages/accelerate/data_loader.py", line 630, in iter self.set_epoch(self.iteration) File "/root/miniconda3/envs/mindeye/lib/python3.10/site-packages/accelerate/data_loader.py", line 707, in set_epoch if hasattr(self.batch_sampler.sampler, "set_epoch"): AttributeError: 'NoneType' object has no attribute 'sampler'

So, I changed "utilis.py" train_dl = torch.utils.data.DataLoader(train_data, batch_size=32, num_workers=1, shuffle=False), set batch_size to 32 but I still can not run the training process image

DavisMeee commented 10 months ago

Hey, you can see my solution in #34 And I think this may help you to solve this

Rui00000 commented 7 months ago

Hey,have you resolved this problem?