alibaba / ChatLearn

A flexible and efficient training framework for large-scale alignment tasks
Apache License 2.0
220 stars 17 forks source link

support qwen2-moe. #146

Closed charles9304 closed 2 weeks ago

charles9304 commented 2 weeks ago

Qwen-MoE模型后续可能需要补充额外的测试,例如Qwen+vLLM的回归测试

charles9304 commented 2 weeks ago

Qwen-MoE模型后续可能需要补充额外的测试,例如Qwen+vLLM的回归测试

后续在回归测试增加test