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

[NewFeatures] support eplb (#3547)

Browse Source 
* [NewFeatures] support eplb

* fix eplb
This commit is contained in:
xiaoxiaohehe001
2025-08-26 16:19:30 +08:00
committed by GitHub
parent 56e2d7e668
commit 9afa236e39
17 changed files with 174 additions and 67 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
fastdeploy/model_executor/layers/quantization/block_wise_fp8.py
View File

@@ -108,7 +108,7 @@ class BlockWiseFP8LinearMethod(QuantMethodBase):
layer.weight.copy_(quanted_weight_tensor, False)
layer.weight_scale.set_value(weight_block_scale_tensor)
def process_prequanted_weights(self, layer, state_dict):
def process_prequanted_weights(self, layer, state_dict, is_rearrange: bool = False):
"""
process_prequanted_weights
"""

2
fastdeploy/model_executor/layers/quantization/tensor_wise_fp8.py
View File

@@ -90,7 +90,7 @@ class TensorWiseFP8LinearMethod(QuantMethodBase):
default_initializer=paddle.nn.initializer.Constant(0),
)
def process_prequanted_weights(self, layer, state_dict) -> None:
def process_prequanted_weights(self, layer, state_dict, is_rearrange: bool = False) -> None:
"""
Process pre-quantized weights before applying them to the model
Args:

2
fastdeploy/model_executor/layers/quantization/weight_only.py
View File

@@ -305,7 +305,7 @@ class GPUWeightOnlyLinearMethod(WeightOnlyLinearMethod):
) -> None:
super().__init__(quant_config)
def process_prequanted_weights(self, layer, state_dict) -> None:
def process_prequanted_weights(self, layer, state_dict, is_rearrange: bool = False) -> None:
"""
Process pre-quantized weights before applying them to the model
Args: