apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 08:37:06 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

polish code with new pre-commit rule (#2923)

Browse Source
This commit is contained in:
Zero Rains
2025-07-19 23:19:27 +08:00
committed by GitHub
parent b8676d71a8
commit 25698d56d1
424 changed files with 14307 additions and 13518 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
fastdeploy/model_executor/layers/quantization/mix_quant.py
View File

@@ -13,6 +13,7 @@
# See the License for the specific language governing permissions and
# limitations under the License.
"""
from typing import Optional
from fastdeploy.model_executor.layers.attention.attention import Attention
@@ -51,26 +52,23 @@ class MixQuantConfig(QuantConfigBase):
@classmethod
def from_config(cls, config: dict) -> "MixQuantConfig":
return cls(config['dense_quant_type'], config['moe_quant_type'],
config.get('kv_cache_quant_type', None),
config.get('image_moe_quant_type', None))
return cls(
config["dense_quant_type"],
config["moe_quant_type"],
config.get("kv_cache_quant_type", None),
config.get("image_moe_quant_type", None),
)
def get_quant_method(self, layer) -> Optional[QuantMethodBase]:
if isinstance(layer, FusedMoE):
if layer.moe_tag == "Image":
return get_quantization_config(
self.image_moe_quant_type).from_config(
{}).get_quant_method(layer)
return get_quantization_config(self.image_moe_quant_type).from_config({}).get_quant_method(layer)
else:
return get_quantization_config(
self.moe_quant_type).from_config(
{}).get_quant_method(layer)
return get_quantization_config(self.moe_quant_type).from_config({}).get_quant_method(layer)
elif isinstance(layer, Attention):
if self.kv_cache_quant_type is not None:
return (get_quantization_config("kvcache").from_config(
self.kv_cache_quant_type).get_quant_method(layer))
return get_quantization_config("kvcache").from_config(self.kv_cache_quant_type).get_quant_method(layer)
else:
return None
else:
return get_quantization_config(self.dense_quant_type).from_config(
{}).get_quant_method(layer)
return get_quantization_config(self.dense_quant_type).from_config({}).get_quant_method(layer)