lucywang720
lucywang720
it seems that you should modify that: class DataCollatorRLHF in data_utils.py batch["prompt"] = F.pad(prompt, # pad=(0, pad_length), pad=(pad_length, 0), mode='constant', value=pad_token_id) that you should keep it a right-padding style. it...
> Thanks for your reply! I made the modifications: > > ```python > # utils/data/data_utils.py > > if pad_length > 0: > batch["prompt"] = F.pad(prompt, > # pad=(0, pad_length), >...
and also, I am wondering whether you use hybrid_engine. if yes, please shut it down
thank you for your share! How about its training cost? > Thanks for your patience, now the actor model can generate answers :) > > some of the answers are...
Hi, if i just want it to generate a video instead of matplotlib rendering, is this issue suitable for my problem?
i ran into the same problem, very strange
> > > I meet the same issue ! When "Actor" is llama2 and "Critic" is opt-350m. > > > Maybe the "Actor" and "Critic" is difference model, it look...
i met the same problem, have you solved it? thank you:)