Adapt INCWeightOnlyLinear (#1686)

Signed-off-by: Kaihui-intel <[email protected]>
intel · Aug 11, 2024 · dbd809b · dbd809b
1 parent 03875a1
commit dbd809b
Show file tree

Hide file tree

Showing 3 changed files with 3 additions and 2 deletions.
diff --git a/intel_extension_for_transformers/transformers/llm/quantization/utils.py b/intel_extension_for_transformers/transformers/llm/quantization/utils.py
@@ -23,7 +23,7 @@
 from ....tools.utils import _ipex_version
 from accelerate import init_empty_weights
 from datasets import load_dataset
-from neural_compressor.torch.algorithms.weight_only.modules import WeightOnlyLinear
+from neural_compressor.torch.algorithms.weight_only.modules import INCWeightOnlyLinear as WeightOnlyLinear
 from neural_compressor.torch.quantization import (
     AutoRoundConfig,
     AWQConfig,

diff --git a/intel_extension_for_transformers/transformers/modeling/modeling_auto.py b/intel_extension_for_transformers/transformers/modeling/modeling_auto.py
@@ -70,7 +70,7 @@
 from ...tools.utils import is_intel_gpu_available, is_ipex_available, _neural_compressor_version
 from accelerate import init_empty_weights
 from huggingface_hub import hf_hub_download
-from neural_compressor.torch.algorithms.weight_only.modules import WeightOnlyLinear
+from neural_compressor.torch.algorithms.weight_only.modules import INCWeightOnlyLinear as WeightOnlyLinear
 from neural_compressor.model.torch_model import PyTorchFXModel
 from packaging import version
 from threading import Thread

diff --git a/tests/requirements.txt b/tests/requirements.txt
@@ -15,6 +15,7 @@ intel-tensorflow==2.14.0
 lm-eval==0.4.3
 mlflow
 nlpaug==1.1.9
+nltk==3.8.1
 onnx
 onnxruntime
 peft==0.6.2