train is ok，but evaluate OOM

[TheHonestBob]

thanks for your project, when I sh finetune.sh，OOM happen when evaluate，even though I set eval_batch_size=1，my gpu is 2080ti 11G.

[ZhuohanX]

Hi, Could you please help me to indicate the versions of packages you are using to run the code? I tried Python 3.7.0 with torch==1.4.0 but it seems that module 'torch.cuda' has no attribtue 'amp' because I think it is included in torch 1.6. And then I changed my package version to Python 3.8 with torch 1.7 but I got an error when running the fine-tune.

File "/home/zhuohanx/HINT/model/utils.py", line 294, in collate_fn batch_score.append([float(s.split()[2]) for s in score_list]) File "/home/zhuohanx/HINT/model/utils.py", line 294, in batch_score.append([float(s.split()[2]) for s in score_list]) IndexError: list index out of range

I am not sure if it is caused by the different version or an error in the code?

Thank you in advance!