99-WSJ
99-WSJ
hello i met this problem in last day , terrible images. so could you solve it and get better work?
> indeed hello can i ask you some questions in your gitpage
> For me I deleted all MPI related code and use `torchrun` because I found it generally easier to work with. (No slot or port or what not) > For...
maybe you check the dropout== 0.3 in readme,
hello, have you solved this question?
hello, have you solved this issue?
> what device did you use in training, I use 512 per V100 16GB lead to an OOM error. but if I use a small batch, the loss go to...