Open ShihuaHuang95 opened 6 years ago
Thanks for your sharing. I have one question, after I have made some changes for running on my environment (Python 3.6, Pytorch 4.1), the training is easy to collapse ( loss be NAN), have you encountered it ? Best,
Thanks for your sharing. I have one question, after I have made some changes for running on my environment (Python 3.6, Pytorch 4.1), the training is easy to collapse ( loss be NAN), have you encountered it ? Best,