rename var

huggingface · Oct 3, 2023 · 745da20 · 745da20
1 parent 7505a19
commit 745da20
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 10 deletions.
diff --git a/optimum/exporters/openvino/__main__.py b/optimum/exporters/openvino/__main__.py
@@ -320,6 +320,6 @@ def main_export(
         input_shapes=input_shapes,
         device=device,
         fp16=fp16,
-        load_in_8bit=int8,
+        int8=int8,
         model_kwargs=model_kwargs,
     )
diff --git a/optimum/exporters/openvino/convert.py b/optimum/exporters/openvino/convert.py
@@ -75,7 +75,7 @@ def export(
     input_shapes: Optional[Dict] = None,
     model_kwargs: Optional[Dict[str, Any]] = None,
     fp16: bool = False,
-    load_in_8bit: bool = False,
+    int8: bool = False,
 ) -> Tuple[List[str], List[str]]:
     """
     Exports a Pytorch or TensorFlow model to an OpenVINO Intermediate Representation.
@@ -118,7 +118,7 @@ def export(
             input_shapes=input_shapes,
             model_kwargs=model_kwargs,
             fp16=fp16,
-            load_in_8bit=load_in_8bit,
+            int8=int8,
         )
 
     elif is_tf_available() and issubclass(type(model), TFPreTrainedModel):
@@ -173,7 +173,7 @@ def export_pytorch_via_onnx(
     input_shapes: Optional[Dict] = None,
     model_kwargs: Optional[Dict[str, Any]] = None,
     fp16: bool = False,
-    load_in_8bit: bool = False,
+    int8: bool = False,
 ):
     """
     Exports a PyTorch model to an OpenVINO Intermediate Representation via ONNX export.
@@ -216,7 +216,7 @@ def export_pytorch_via_onnx(
         ov_model,
         output.parent / OV_XML_FILE_NAME if output.suffix != ".xml" else output,
         compress_to_fp16=fp16,
-        load_in_8bit=load_in_8bit,
+        load_in_8bit=int8,
     )
     return input_names, output_names, True
 
@@ -230,7 +230,7 @@ def export_pytorch(
     input_shapes: Optional[Dict] = None,
     model_kwargs: Optional[Dict[str, Any]] = None,
     fp16: bool = False,
-    load_in_8bit: bool = False,
+    int8: bool = False,
 ) -> Tuple[List[str], List[str]]:
     """
     Exports a PyTorch model to an OpenVINO Intermediate Representation.
@@ -325,7 +325,7 @@ def ts_patched_forward(*args, **kwargs):
                 ov_model = convert_model(model, example_input=dummy_inputs, input=input_info)
         except Exception as ex:
             logger.warning(f"Export model to OpenVINO directly failed with: \n{ex}.\nModel will be exported to ONNX")
-            return export_pytorch_via_onnx(model, config, opset, output, device, input_shapes, model_kwargs, fp16=fp16, load_in_8bit=load_in_8bit)
+            return export_pytorch_via_onnx(model, config, opset, output, device, input_shapes, model_kwargs, fp16=fp16, int8=int8)
         ordered_dummy_inputs = {param: dummy_inputs[param] for param in sig.parameters if param in dummy_inputs}
         ordered_input_names = list(inputs)
         flatten_inputs = flattenize_inputs(ordered_dummy_inputs.values())
@@ -346,7 +346,7 @@ def ts_patched_forward(*args, **kwargs):
             inp_tensor.get_node().set_partial_shape(static_shape)
             inp_tensor.get_node().set_element_type(get_element_type(inp_data.cpu().numpy().dtype))
         ov_model.validate_nodes_and_infer_types()
-        _save_model(ov_model, output, compress_to_fp16=fp16, load_in_8bit=load_in_8bit)
+        _save_model(ov_model, output, compress_to_fp16=fp16, load_in_8bit=int8)
         clear_class_registry()
         del model
         gc.collect()
@@ -364,7 +364,7 @@ def export_models(
     input_shapes: Optional[Dict] = None,
     model_kwargs: Optional[Dict[str, Any]] = None,
     fp16: bool = False,
-    load_in_8bit: bool = False,
+    int8: bool = False,
 ) -> Tuple[List[List[str]], List[List[str]]]:
     """
     Export the models to OpenVINO IR format
@@ -410,7 +410,7 @@ def export_models(
                 input_shapes=input_shapes,
                 model_kwargs=model_kwargs,
                 fp16=fp16,
-                load_in_8bit=load_in_8bit,
+                int8=int8,
             )
         )
 

diff --git a/optimum/intel/openvino/modeling_base.py b/optimum/intel/openvino/modeling_base.py
@@ -219,6 +219,7 @@ def _from_transformers(
         local_files_only: bool = False,
         task: Optional[str] = None,
         trust_remote_code: bool = False,
+        # load_in_8bit: bool = False, # TODO : add int8
         **kwargs,
     ):
         """