Cola Chan

icon indicating a website link https://141forever.github.io

icon company East China Normal University icon location Shanghai icon location I am now a joint Ph.D. student of East China Normal University and the China Academy of Artificial Intelligence.

Results 17 comments of Cola Chan

Issue about ray: ray.exceptions.RayTaskError(OSError)

I cannot remember, but I did solve it. It is nothing about VLLM or SGlang. I remember that our server was under maintenance at that time. Once they finished the...

Issue about ray: ray.exceptions.RayTaskError(OSError)

Good luck mate!

ray.exceptions.ActorDiedError: The actor died unexpectedly before finishing this task.

same problem

ray.exceptions.ActorDiedError: The actor died unexpectedly before finishing this task.

> I finally found that it is a cluster communication bug, you can test it with vllm for inference on multimachine. If it works, then the cluster communication is ok....

[Bug] 微调过的Qwen2.5-7b输出全是感叹号/The output of fine-tuning Qwen2.5-7b is '！！！！！！'

这个问题我自己这边的解决方法是，首先检查fp16的精度问题，然后我升级了cuda为121，transformers从3.34升级到了3.37，然后就好了。

assert "expandable_segments:True" not in conf报错

same error

RuntimeError: 0 active drivers ([]). There should only be one.

> 不兼容，我也遇到了，必须卸载才能用 我也是遇到了这个问题，卸载deepspeed后解决了，甚至是vllm和triton的兼容性问题也是这样解决的，属实逆天