Peng Wu pengwu22

## test_varlen_cg.py
from typing import List, Optional, Tuple

import pytest
import torch
from vllm_flash_attn import flash_attn_varlen_func, flash_attn_with_kvcache


NUM_HEADS = [(64, 8)]
HEAD_SIZES = [128]
BLOCK_SIZES = [16]
	from typing import List, Optional, Tuple

	import pytest
	import torch
	from vllm_flash_attn import flash_attn_varlen_func, flash_attn_with_kvcache


	NUM_HEADS = [(64, 8)]
	HEAD_SIZES = [128]
	BLOCK_SIZES = [16]