Merge pull request #5163 from liu-zichen/fix_ppo_optim

fix lr not change Former-commit-id: da335d42c30df2c2898516e290c0059250d57b63
2025-08-22 22:02:51 +08:00 · 2024-08-19 23:56:24 +08:00 · 2024-08-19 23:56:24 +08:00 · 033dd1f1dc
commit 033dd1f1dc
parent 5e19a604a6 8a7ab8ab21
1 changed files with 1 additions and 0 deletions
--- a/src/llamafactory/train/ppo/trainer.py
+++ b/src/llamafactory/train/ppo/trainer.py
@ -135,6 +135,7 @@ class CustomPPOTrainer(PPOTrainer, Trainer):
            dataset=train_dataset,
            data_collator=data_collator,
            lr_scheduler=scheduler,
+            optimizer=optimizer,
        )

        self.args = training_args