423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-08-03 20:22:49 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix #809

Browse Source 
Former-commit-id: e2bf7c3badbd5d2fd513ca7a00bd74d9c0d62d07
This commit is contained in:
hiyouga 2023-09-07 19:04:32 +08:00
parent f74b980650
commit e6fa0229f4
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)