apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2026-04-23 08:21:53 +08:00
Code Issues Actions 19 Packages Projects Releases Wiki Activity

[Models][OP][Optimization] Support DeepSeek-v3.2 model, integrate DSA & Indexer architecture with FlashMLA/DeepGEMM (#6689)

Browse Source 
* Support DeepSeek-v3.2 model, integrate DSA & Indexer architecture with FlashMLA/DeepGEMM
This commit is contained in:
AIbin
2026-03-10 15:05:14 +08:00
committed by GitHub
parent 25c479312d
commit c3aceb6bdc
22 changed files with 8022 additions and 143 deletions
Download Patch File Download Diff File Expand all files Collapse all files
fastdeploy/model_executor/layers/quantization/weight_only.py
+1 -2
View File
@@ -20,7 +20,6 @@ from typing import Optional
import paddle
from paddle.nn.quant import weight_quantize
from paddleformers.utils.log import logger
from fastdeploy import envs
from fastdeploy.model_executor.layers.linear import (
@@ -181,7 +180,7 @@ class WeightOnlyConfig(QuantConfigBase):
and check_machete_supports_shape(layer.weight_shape[0], layer.weight_shape[1])
):
self.group_size = query_machete_supported_group_size(layer.weight_shape[0])
logger.info(f"Using Machete kernel for WeightOnlyLinearMethod, group size: {self.group_size}")
# logger.info(f"Using Machete kernel for WeightOnlyLinearMethod, group size: {self.group_size}")
return MacheteWeightOnlyLinearMethod(self)
return GPUWeightOnlyLinearMethod(self)