Merge 0c91dea4a2 into 7a6fddd63a

README.md

@@ -1909,6 +1909,8 @@ python llmexport.py --path /path/to/模型路径/ --export mnn --hqq --dst_path
 
 * [@Nijikadesu](https://github.com/Nijikadesu)：[🔗以交互笔记本方式分解项目代码](https://github.com/jingyaogong/minimind/issues/213)
 
+* [@jaylearnstocode](https://github.com/jaylearnstocode)：[🔗模型结构、注意力机制与训练流程可视化](https://llm-visualization-minimind.vercel.app/)
+
 
 致谢以下优秀的论文与项目：
 

README_en.md

@@ -1909,6 +1909,8 @@ Thanks to the following contributors for sharing training notes, data processing
 
 * [@Nijikadesu](https://github.com/Nijikadesu): [🔗Breaking Down Project Code in Interactive Notebook Format](https://github.com/jingyaogong/minimind/issues/213)
 
+* [@jaylearnstocode](https://github.com/jaylearnstocode): [🔗Visualization of Model Architecture, Attention Mechanisms, and Training Pipelines](https://llm-visualization-minimind.vercel.app/)
+
 
 Thanks also to the following papers and projects:
 

requirements.txt

@@ -23,10 +23,10 @@ jinja2==3.1.2
 jsonlines==4.0.0
 trl==0.13.0
 ujson==5.1.0
-wandb==0.18.3
+wandb==0.22.3
 streamlit==1.50.0
 einops==0.8.1
-swanlab==0.7.11
+swanlab==0.9.8
 modelscope==1.37.0
 # torch==2.6.0
 # torchvision==0.21.0

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)

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)

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])