Y.Y. comments

Results 7 comments of


                                            Y.Y.

怎么实现在多GPU 上微调训练？

已经确认可以使用 ``` accelerate config # 首先配置分布式环境 accelerate launch src/train_bash.py # 参数同上 ```

BUG: probability tensor contains either `inf`, `nan` or element < 0

加載 deepseek-coder-33b-instruct 模型也會出現此錯誤，能否可以自己加選項 model.bfloat16()

[Usage]: tensor-parallel-size=2，The program just kept hanging

Ollama 支持一个模型，加载到多卡嘛？发自我的 iPhone在 2025年2月18日，18:04，斓曦未央丶 ***@***.***> 写道： @gaoshangle 怎么验证不大懂啊但是用ollama多卡跑没问题—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you are subscribed to this thread.Message ID:...

[Usage]: tensor-parallel-size=2，The program just kept hanging

Ollama 多卡是怎么支持的，单模型可以跨卡运行吗？发自我的 iPhone在 2025年2月21日，11:21，Zoe Shay ***@***.***> 写道：有没有可能和驱动有关呢，我ollama跑多卡，是ok的—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: ***@***.***> xueshuai0922 left a...

Y.Y.

怎么实现在多GPU 上微调训练？

BUG: probability tensor contains either `inf`, `nan` or element < 0

[Usage]: tensor-parallel-size=2，The program just kept hanging

[Usage]: tensor-parallel-size=2，The program just kept hanging

[Usage]: tensor-parallel-size=2，The program just kept hanging

[Usage]: restarting vllm --> "WARNING: process group has NOT been destroyed before we destruct ProcessGroupNCCL"

是否支持离线部署？（使用自建的镜像仓库拉取镜像文件）