EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[1/N][Refactor][Quantization] remove redundant quantizer class (#2680)

Browse Source 
### What this PR does / why we need it?

AscendQuantizer/LLMQuantizer class is used to select quant method based
on quant config and some other arguments,
but it is more simple and clean replacing these classes with map. So i
remove them.

### Does this PR introduce _any_ user-facing change?
No 

### How was this patch tested?

ut and e2e test


- vLLM version: v0.10.1.1
- vLLM main:
6997a25ac6

Signed-off-by: 22dimensions <waitingwind@foxmail.com>
This commit is contained in:
22dimensions
2025-09-04 11:35:14 +08:00
committed by GitHub
parent d4370ebc42
commit 37f5a29cd4
10 changed files with 321 additions and 554 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
vllm_ascend/torchair/utils.py
View File

@@ -180,15 +180,22 @@ def register_torchair_model():
def torchair_quant_method_register():
from vllm_ascend.quantization.quantizer import \
SUPPORT_ASCEND_QUANTIZER_TYPE
from vllm_ascend.torchair.quantization.torchair_quantizer import (
TorchairW4A8DYNAMICQuantizer, TorchairW8A8DYNAMICQuantizer)
from vllm_ascend.quantization.utils import ASCEND_QUANTIZATION_METHOD_MAP
from vllm_ascend.torchair.quantization.torchair_w4a8_dynamic import (
TorchairAscendW4A8DynamicFusedMoEMethod,
TorchairAscendW4A8DynamicLinearMethod)
from vllm_ascend.torchair.quantization.torchair_w8a8_dynamic import (
TorchairAscendW8A8DynamicFusedMoEMethod,
TorchairAscendW8A8DynamicLinearMethod)
SUPPORT_ASCEND_QUANTIZER_TYPE[
"W8A8_DYNAMIC"] = TorchairW8A8DYNAMICQuantizer
SUPPORT_ASCEND_QUANTIZER_TYPE[
"W4A8_DYNAMIC"] = TorchairW4A8DYNAMICQuantizer
ASCEND_QUANTIZATION_METHOD_MAP["W8A8_DYNAMIC"][
"linear"] = TorchairAscendW8A8DynamicLinearMethod
ASCEND_QUANTIZATION_METHOD_MAP["W8A8_DYNAMIC"][
"moe"] = TorchairAscendW8A8DynamicFusedMoEMethod
ASCEND_QUANTIZATION_METHOD_MAP["W4A8_DYNAMIC"][
"linear"] = TorchairAscendW4A8DynamicLinearMethod
ASCEND_QUANTIZATION_METHOD_MAP["W4A8_DYNAMIC"][
"moe"] = TorchairAscendW4A8DynamicFusedMoEMethod
def torchair_ops_patch():