What is the expected structure of ImageNet dataset? I'm trying to pass train folder wi

Imagenet training about pytorchnet HOT 2 CLOSED

human-analysis commented on July 17, 2024

Imagenet training

from pytorchnet.

Comments (2)

michaelklachko commented on July 17, 2024

Ok, so it seems like it expects val images to be in folders per class, as explained here: https://github.com/pytorch/examples/tree/master/imagenet

Yet I still get the following "Assertion t >= 0 && t < n_classes failed" error:

/opt/conda/conda-bld/pytorch_1532579245307/work/aten/src/THCUNN/ClassNLLCriterion.cu:105: void cunn_ClassNLLCriterion_updateOutput_kernel(Dtype *, Dtype *, Dtype *, long *, Dtype *, int, int, int, int, long) [with Dtype = float, Acctype = float]: block: [0,0,0], thread: [31,0,0] Assertion `t >= 0 && t < n_classes` failed.
Traceback (most recent call last):
  File "main.py", line 64, in <module>
    main()
  File "main.py", line 51, in main
    loss_train = trainer.train(epoch, loaders)
  File "/home/michael/pytorchnet/train.py", line 138, in train
    loss.backward()
  File "/home/michael/miniconda2/envs/pt/lib/python3.6/site-packages/torch/tensor.py", line 93, in backward
    torch.autograd.backward(self, gradient, retain_graph, create_graph)
  File "/home/michael/miniconda2/envs/pt/lib/python3.6/site-packages/torch/autograd/__init__.py", line 90, in backward
    allow_unreachable=True)  # allow_unreachable flag
RuntimeError: cuda runtime error (59) : device-side assert triggered at /opt/conda/conda-bld/pytorch_1532579245307/work/aten/src/THC/generic/THCTensorMath.cu:26
NoneType: None

I can run https://github.com/pytorch/examples/blob/master/imagenet/main.py just fine, so must be something missing in your code...

from pytorchnet.

michaelklachko commented on July 17, 2024

Oh, I forgot to change number of classes in args file, my bad!

from pytorchnet.

Recommend Projects