about ppo unwrap_model
Author: likecakeCreated Aug 10, 2026Updated Sep 8, 2026
Labelsbugpending
Reminder
- I have read the above rules and searched the existing issues.
System Info
with unwrap_model_for_generation(self.model, self.accelerator) as unwrapped_model: unwrapped_model: AutoModelForCausalLMWithValueHead = self.accelerator.unwrap_model(self.model) 在src/llamafactory/train/ppo/trainer.py中为什么要对self.model做两次unwrap呢，这样是不是重复了？
Reproduction
Put your message here.
Others
No response
Source: hiyouga/LlamaFactory