search

kaniblu / pytorch-skipthoughts

MIT License
18 stars 7 forks source link

Runtime Error #4

Open JiangYunchao opened 6 years ago

JiangYunchao commented 6 years ago

When I start my training, I meet this error. Traceback (most recent call last): File "/share1/home/jiang/anaconda2/envs/pytorch/lib/python3.6/runpy.py", line 193, in _run_module_as_main "main", mod_spec) File "/share1/home/jiang/anaconda2/envs/pytorch/lib/python3.6/runpy.py", line 85, in _run_code exec(code, run_globals) File "/share1/home/jiang/anaconda2/envs/pytorch/lib/python3.6/site-packages/torchst/train.py", line 712, in main() File "/share1/home/jiang/anaconda2/envs/pytorch/lib/python3.6/site-packages/torchst/train.py", line 706, in main trainer.train() File "/share1/home/jiang/anaconda2/envs/pytorch/lib/python3.6/site-packages/torchst/train.py", line 415, in train loss_b, loss_s = self.step_train(step, data) File "/share1/home/jiang/anaconda2/envs/pytorch/lib/python3.6/site-packages/torchst/train.py", line 391, in step_train title="Training Loss") File "/share1/home/jiang/anaconda2/envs/pytorch/lib/python3.6/site-packages/torchst/train.py", line 358, in step volatile=volatile) File "/share1/home/jiang/anaconda2/envs/pytorch/lib/python3.6/site-packages/torchst/train.py", line 202, in prepare_batches data = torch.cat(data) RuntimeError: inconsistent tensor sizes at /pytorch/torch/lib/THC/generic/THCTensorMath.cu:157