Torch not compiled with CUDA enabled

[danielsoy]

Hi Marco, thanks for sharing your project.

I was trying to run train_student.py from CPU with device=cpu in config.py, but I have got:

Torch not compiled with CUDA enabled error.

Is there another device config that i should change, to train the model from CPU?

I could run all the scripts in colab, from GPU by the way.

[marco-rudolph]

Hey,

sorry for the inconvenience. There were ".cuda()"s which I changed to ".to(c.devices)"s in the latest commit. Please report if there are still issues with this.

With "cpu" for device in config.py everything should be set.

Hope that helps, Marco

[danielsoy]

the error poped, training with the latest commit. the one with :(c.devices).

running on colab with cpu enabled in config.py, and cpu as resource from google, fails too and shows: RuntimeError: No CUDA GPUs are available.

As it was trying to train with gpu, despite the selected device is cpu.

[danielsoy]

heres the full log:

Train class cable_gland Traceback (most recent call last): File "train_student.py", line 109, in train_dataset(train) File "/content/AST/utils.py", line 25, in train_dataset mean_sc, max_sc = train_function(train_loader, test_loader) File "train_student.py", line 13, in train student = Model(nf=not c.asymmetric_student, channels_hidden=c.channels_hidden_student, n_blocks=c.n_st_blocks) File "/content/AST/model.py", line 95, in init self.pos_enc = positionalencoding2d(c.pos_enc_dim, c.map_len, c.map_len) File "/content/AST/model.py", line 79, in positionalencoding2d return P.cuda()[None] File "/usr/local/lib/python3.8/dist-packages/torch/cuda/init.py", line 229, in _lazy_init torch._C._cuda_init() RuntimeError: No CUDA GPUs are available

[marco-rudolph]

It seems like you don't use the latest commit. line 79 in model.py is https://github.com/marco-rudolph/AST/blob/47adbdce2f17ffe238a52e0bcb4c277b87652e8b/model.py#L79

[danielsoy]

bingo! mine says CUDA instead of c.device. Thanks , I see how it goes now...

[marco-rudolph]

no problem :)