search

facebookresearch / mae

PyTorch implementation of MAE https//arxiv.org/abs/2111.06377
Other
7.2k stars 1.2k forks source link

should i do nb_classes=1 or 2 for binary classification finetuning task? #138

Open ucalyptus2 opened 1 year ago

ucalyptus2 commented 1 year ago

Hi @endernewton, @KaimingHe great paper.

Let me know what is better. I get errors with nb_classes=1

../aten/src/ATen/native/cuda/ScatterGatherKernel.cu:111: operator(): block: [0,0,0], thread: [63,0,0] Assertion `idx_dim >= 0 && idx_dim < index_size && "index out of bounds"` failed.
terminate called after throwing an instance of 'c10::CUDAError'
  what():  CUDA error: device-side assert triggered
Exception raised from createEvent at ../aten/src/ATen/cuda/CUDAEvent.h:174 (most recent call first):