[bench] Added vllm vs nano-vllm bench.
This commit is contained in:
@@ -3,7 +3,7 @@ from torch import nn
|
||||
import triton
|
||||
import triton.language as tl
|
||||
|
||||
from flash_attn import flash_attn_varlen_func, flash_attn_with_kvcache
|
||||
from flash_attn.flash_attn_interface import flash_attn_varlen_func, flash_attn_with_kvcache
|
||||
from nanovllm.utils.context import get_context
|
||||
|
||||
|
||||
|
||||
Reference in New Issue
Block a user