jackchen556

Results 1 comments of jackchen556

you can add parameters in deepspeed.init_inference(), replace_with_kernel_inject=True, and i fix the bug in the same problem