chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

Clean up import vllm in quantization/__init__.py (#4834)

Browse Source
This commit is contained in:
Lianmin Zheng
2025-03-28 10:34:10 -07:00
committed by GitHub
parent ef9a378a20
commit 74e0ac1dbd
14 changed files with 191 additions and 254 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
python/sglang/srt/layers/quantization/fp8_kernel.py
View File

@@ -24,6 +24,7 @@ import triton.language as tl
from sglang.srt.utils import (
direct_register_custom_op,
get_bool_env_var,
get_device_core_count,
get_device_name,
get_device_sm,
@@ -43,7 +44,7 @@ if _is_cuda:
from sgl_kernel import sgl_per_token_group_quant_fp8, sgl_per_token_quant_fp8
sm_version = get_device_sm()
if sm_version >= 90 and int(os.getenv("SGL_ENABLE_JIT_DEEPGEMM", "1")):
if sm_version >= 90 and get_bool_env_var("SGL_ENABLE_JIT_DEEPGEMM", default="true"):
_enable_jit_deepgemm = True