ros/LLaMA-Factory
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix baichuan templates

Browse Source 
Former-commit-id: f48a49e835b32f3991cfad8874c7b9c78953809f
This commit is contained in:
hiyouga
2023-09-07 18:54:14 +08:00
parent ed89e29bcc
commit eae7b331d3
9 changed files with 53 additions and 87 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/llmtuner/tuner/ppo/trainer.py
View File

@@ -76,7 +76,7 @@ class PPOPeftTrainer(PPOTrainer, PeftTrainer):
# Keyword arguments for `model.generate`
gen_kwargs = self.generating_args.to_dict()
gen_kwargs["eos_token_id"] = list(set([self.tokenizer.eos_token_id] + self.tokenizer.additional_special_tokens_ids))
gen_kwargs["eos_token_id"] = [self.tokenizer.eos_token_id] + self.tokenizer.additional_special_tokens_ids
gen_kwargs["pad_token_id"] = self.tokenizer.pad_token_id
gen_kwargs["logits_processor"] = get_logits_processor()