423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-11-05 18:32:14 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix #809

Browse Source 
Former-commit-id: 2783ca75365d7c373cefba039788a48f0b8f35fc
This commit is contained in:
hiyouga 2023-09-07 19:04:32 +08:00
parent 47d9325873
commit c426dd6d69
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

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

@ -42,7 +42,7 @@ class DPOPeftTrainer(PeftModelMixin, DPOTrainer):
if ref_model is not None: if ref_model is not None:
if self.is_deepspeed_enabled: if self.is_deepspeed_enabled:
self.ref_model = self.accelerator._prepare_deepspeed(self.ref_model) self.ref_model, = self.accelerator._prepare_deepspeed(self.ref_model)
self.ref_model.eval() self.ref_model.eval()
else: else:
self.ref_model = self.accelerator.prepare_model(self.ref_model, evaluation_mode=True) self.ref_model = self.accelerator.prepare_model(self.ref_model, evaluation_mode=True)