data_downloader.py is giving error.

[himanshubeniwal]

While downloading the data, it is giving an error: python data_downloader.py
The error I am getting:

UnboundLocalError: local variable 'train_data' referenced before assignment

Also, if directly using the python main.py:
Then the following error is encountered:

RuntimeError: Expected one of cpu, cuda, mkldnn, opengl, opencl, ideep, hip, msnpu device type at start of device string: mps

[verazuo]

Hi himanshubeniwal,

I fixed the bug in data_downloader.py, please git pull to get the update and re-run it.

And for the RuntimeError in main.py, it is caused by the device setting.
As I am using MBP M1, I set the device as mps.
You can feel free to change it to cuda or cpu, depending on your device type.

If you encounter other errors, do not hesitate to let me know. :D

Best wishes,
Vera

[himanshubeniwal]

Thanks @verazuo!
I have pulled the new code. Data has been downloaded.
But when I am running the python main.py --device cuda:1, here I am using cuda:1 to send it to the first cuda device. It is returning the following error:

RuntimeError: Input type (torch.FloatTensor) and weight type (torch.cuda.FloatTensor) should be the same or input should be a MKLDNN tensor and weight is a dense tensor

[verazuo]

Hi,

This is because the model is on the GPU, but the data is on the CPU.
To solve it, you can send the data to the GPU.

batch_x = batch_x.to(device, non_blocking=True)

I updated a fixed version and tested it on my GPU server.
Would you please try it on your machine again?

Best,
Vera

[himanshubeniwal]

This worked! Thanks! :)
Edit: I have updated the requirements for me with the latest versions. As there were dependencies issues. New versions works fine.