DeepSpeed
Add fp16 support of Qwen1.5MoE models (A2.7B) to DeepSpeed-FastGen
#5403
Merged

Loading