Getting Segmentation fault while training on MNLI
Closed this issue · 1 comments
RishabhMaheshwary commented
==================== Training ESIM model on device: cuda:0 ====================
- Training epoch 1:
Avg. batch proc. time: 0.0299s, loss: 0.8712: 100%|█████████████████████████████████████████████████████████████| 49088/49088 [26:49<00:00, 30.49it/s]
Segmentation fault (core dumped)
coetaur0 commented
It's a bit difficult to guess what went wrong solely on the basis of this error message. But which version of PyTorch did you use? The model was developed with PyTorch 0.4 (although it worked with version 1.0 when I last tried). Maybe this has something to do with it.