demiseom / rnnt-pytorch Goto Github PK

train_loss = model(inputs, targets_list, input_sizes, target_sizes)
File "/home/Liangcd/anaconda3/envs/pyEnv/lib/python3.7/site-packages/torch/nn/modules/module.py", line 541, in call
result = self.forward(*input, **kwargs)
File "/home/Liangcd/code/pytorch/RNNT-pytorch-master/models/models.py", line 188, in forward
loss = self.loss(out, ys.int(), xlen, ylen)
File "/home/Liangcd/anaconda3/envs/pyEnv/lib/python3.7/site-packages/torch/nn/modules/module.py", line 541, in call
result = self.forward(*input, **kwargs)
File "/home/Liangcd/anaconda3/envs/pyEnv/lib/python3.7/site-packages/warprnnt_pytorch-0.1-py3.7-linux-x86_64.egg/warprnnt_pytorch/init.py", line 100, in forward
return self.loss(acts, labels, act_lens, label_lens, self.blank, self.reduction)
File "/home/Liangcd/anaconda3/envs/pyEnv/lib/python3.7/site-packages/warprnnt_pytorch-0.1-py3.7-linux-x86_64.egg/warprnnt_pytorch/init.py", line 40, in forward
grads /= minibatch_size
RuntimeError: CUDA error: an illegal memory access was encountered

논문에서는 bptt를 사용했다고 봤던 것 같은데, 해당 소스에는 bptt 구현은 없는 것 같네요?
layer가 한종류라면은 쉽게 구현 가능할텐데, transcript와 prediction이 따로 네트워크로 존재하다보니, 어떻게 해야할지 크게 난항을 겪고있습니다 ㅠㅠㅠ

혹시 좋은 아이디어 없으신지요?

I got an error on from data.SpecAugment import sparse_image_warp_zcaceres could you give me some advice to solve it.