Skip to content

🧠 两小时从 0 训出你的 64M 大模型。Train your own 64M LLM from scratch in 2 hours.

[refactor] remove redundant rollout update_policy after torch.compile

TOPSAIL\781900c91dea4a28e
diff --git a/trainer/train_agent.py b/trainer/train_agent.py
index 445101c..e18aada 100644
--- a/trainer/train_agent.py
+++ b/trainer/train_agent.py
@@ -470,7 +470,6 @@ if __name__ == "__main__":
if args.use_compile == 1:
model = torch.compile(model)
Logger('torch.compile enabled')
- rollout_engine.update_policy(model)
if dist.is_initialized():
model = DistributedDataParallel(model, device_ids=[local_rank])
rollout_engine.update_policy(model)
diff --git a/trainer/train_grpo.py b/trainer/train_grpo.py
index 200adb6..664e7e5 100755
--- a/trainer/train_grpo.py
+++ b/trainer/train_grpo.py
@@ -310,7 +310,6 @@ if __name__ == "__main__":
if args.use_compile == 1:
model = torch.compile(model)
Logger('torch.compile enabled')
- rollout_engine.update_policy(model)
if dist.is_initialized():
model = DistributedDataParallel(model, device_ids=[local_rank])
rollout_engine.update_policy(model)
diff --git a/trainer/train_ppo.py b/trainer/train_ppo.py
index 0cf844d..f0427b6 100644
--- a/trainer/train_ppo.py
+++ b/trainer/train_ppo.py
@@ -426,7 +426,6 @@ if __name__ == "__main__":
if args.use_compile == 1:
actor_model = torch.compile(actor_model)
Logger('torch.compile enabled')
- rollout_engine.update_policy(actor_model)
if dist.is_initialized():
actor_model = DistributedDataParallel(actor_model, device_ids=[local_rank])
critic_model = DistributedDataParallel(critic_model, device_ids=[local_rank])