EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

sglang quant module remove vllm dependency (#4507)

Browse Source
This commit is contained in:
Xiaoyu Zhang
2025-03-18 06:51:59 +08:00
committed by GitHub
parent f81a27f65e
commit 9b81f9bd34
8 changed files with 907 additions and 238 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
python/sglang/srt/layers/quantization/modelopt_quant.py
View File

@@ -5,12 +5,6 @@ from typing import Any, Dict, List, Optional
import torch
from torch.nn.parameter import Parameter
from vllm.model_executor.layers.quantization.kv_cache import BaseKVCacheMethod
from vllm.model_executor.layers.quantization.utils.w8a8_utils import (
convert_to_channelwise,
cutlass_fp8_supported,
requantize_with_max_scale,
)
from sglang.srt.layers.attention.base_attn_backend import AttentionBackend
from sglang.srt.layers.linear import LinearBase, LinearMethodBase
@@ -19,7 +13,15 @@ from sglang.srt.layers.quantization.base_config import (
QuantizationConfig,
QuantizeMethodBase,
)
from sglang.srt.layers.quantization.fp8_utils import apply_fp8_linear
from sglang.srt.layers.quantization.fp8_utils import (
apply_fp8_linear,
cutlass_fp8_supported,
)
from sglang.srt.layers.quantization.kv_cache import BaseKVCacheMethod
from sglang.srt.layers.quantization.utils import (
convert_to_channelwise,
requantize_with_max_scale,
)
# Initialize logger for the module
logger = logging.getLogger(__name__)