DeepSpeed
DeepSpeed copied to clipboard
Can you provide a startup script for deepspeed inference llama 65B MP=8?