github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-09-11 05:49:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

[chatgpt] fix lora gemini conflict in RM training (#2984)

Browse Source 
* fix lora bug

* polish

* fix lora gemini
This commit is contained in:
BlueRum
2023-03-03 15:58:16 +08:00
committed by GitHub
parent 19ad49fb3b
commit f5ca0397dd
3 changed files with 3 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
applications/ChatGPT/examples/train_reward_model.py
View File

@@ -66,8 +66,6 @@ def train(args):
train_dataset = RewardDataset(train_data, tokenizer, max_len)
eval_dataset = RewardDataset(eval_data, tokenizer, max_len)
# batch_size here is expected to be C(k,2), k means # response of each prompt
# be limited with the format of dataset 'Dahoas/rm-static', we'd better use batch_size as 1
trainer = RewardModelTrainer(model=model,
strategy=strategy,
optim=optim,