fix

Author: echarlaix

optimum/commands/export/openvino.py

@@ -20,7 +20,6 @@
 from typing import TYPE_CHECKING, Optional
 
 from huggingface_hub.constants import HUGGINGFACE_HUB_CACHE
-from transformers.utils.quantization_config import QuantizationMethod
 
 from ...exporters import TasksManager
 from ...intel.utils.import_utils import DIFFUSERS_IMPORT_ERROR, is_diffusers_available
@@ -289,7 +288,7 @@ def _get_default_int4_config(model_id_or_path, library_name):
                     "all_layers": None if is_int8 else self.args.all_layers,
                     "dataset": self.args.dataset,
                     "num_samples": self.args.num_samples,
-                    "quant_method": QuantizationMethod.AWQ if self.args.awq else None,
+                    "quant_method": "awq" if self.args.awq else "default",
                     "sensitivity_metric": self.args.sensitivity_metric,
                     "scale_estimation": self.args.scale_estimation,
                 }

optimum/intel/openvino/configuration.py

@@ -200,7 +200,7 @@ def __init__(
         sensitivity_metric: Optional[str] = None,
         ignored_scope: Optional[dict] = None,
         num_samples: Optional[int] = None,
-        quant_method: Union[QuantizationMethod, OVQuantizationMethod] = OVQuantizationMethod.DEFAULT,
+        quant_method: Union[str, QuantizationMethod, OVQuantizationMethod] = OVQuantizationMethod.DEFAULT,
         scale_estimation: bool = None,
         **kwargs,
     ):

optimum/intel/openvino/quantization.py

@@ -38,7 +38,6 @@
 from transformers import AutoTokenizer, DataCollator, PreTrainedModel, default_data_collator
 from transformers.pytorch_utils import Conv1D
 from transformers.utils import is_accelerate_available
-from transformers.utils.quantization_config import QuantizationMethod
 
 from optimum.exporters.onnx.convert import check_dummy_inputs_are_allowed
 from optimum.exporters.tasks import TasksManager
@@ -828,7 +827,7 @@ def _weight_only_quantization(
         group_size=config.group_size,
         all_layers=config.all_layers,
         sensitivity_metric=sensitivity_metric,
-        awq=config.quant_method.lower() == "awq",
+        awq=getattr(config.quant_method, "name", "") == "AWQ",
         ignored_scope=config.get_ignored_scope_instance(),
         dataset=dataset,
         subset_size=config.num_samples if config.num_samples else 128,