Remove compress_to_fp16=False from examples (#2682)

### Changes Removed `compress_to_fp16=False` from `save_model` functions in examples. ### Reason for changes Using this parameter when saving OpenVINO model is no longer required ### Tests Changes were tested via pytest, using example tests in project. While testing, some errors occured, but all of them were about wrong development environment settings. Setting up the invironment took most of the time.
openvinotoolkit · May 24, 2024 · 936533e · 936533e
1 parent 4d7fd56
commit 936533e
Show file tree

Hide file tree

Showing 7 changed files with 7 additions and 7 deletions.
diff --git a/.../post_training_quantization/openvino/anomaly_stfpm_quantize_with_accuracy_control/main.py b/.../post_training_quantization/openvino/anomaly_stfpm_quantize_with_accuracy_control/main.py
@@ -177,7 +177,7 @@ def transform_fn(data_item):
     # nncf.quantize_with_accuracy_control(...) keeps the most impactful operations within
     # the model in the original precision to achieve the specified model accuracy.
     int8_ir_path = f"{ROOT}/stfpm_int8.xml"
-    ov.save_model(ov_quantized_model, int8_ir_path, compress_to_fp16=False)
+    ov.save_model(ov_quantized_model, int8_ir_path)
     print(f"[2/7] Save INT8 model: {int8_ir_path}")
     int8_size = get_model_size(int8_ir_path, verbose=True)
 

diff --git a/examples/post_training_quantization/openvino/mobilenet_v2/main.py b/examples/post_training_quantization/openvino/mobilenet_v2/main.py
@@ -142,7 +142,7 @@ def transform_fn(data_item):
 fp32_model_size = get_model_size(fp32_ir_path, verbose=True)
 
 int8_ir_path = ROOT / "mobilenet_v2_int8.xml"
-ov.save_model(ov_quantized_model, int8_ir_path, compress_to_fp16=False)
+ov.save_model(ov_quantized_model, int8_ir_path)
 print(f"[2/7] Save INT8 model: {int8_ir_path}")
 int8_model_size = get_model_size(int8_ir_path, verbose=True)
 

diff --git a/examples/post_training_quantization/openvino/yolov8/main.py b/examples/post_training_quantization/openvino/yolov8/main.py
@@ -152,7 +152,7 @@ def main():
     # Quantize mode in OpenVINO representation
     quantized_model = quantize(ov_model, data_loader, validator)
     quantized_model_path = Path(f"{ROOT}/{MODEL_NAME}_openvino_model/{MODEL_NAME}_quantized.xml")
-    ov.save_model(quantized_model, str(quantized_model_path), compress_to_fp16=False)
+    ov.save_model(quantized_model, str(quantized_model_path))
 
     # Validate FP32 model
     fp_stats, total_images, total_objects = validate(ov_model, tqdm(data_loader), validator)

diff --git a/examples/post_training_quantization/openvino/yolov8_quantize_with_accuracy_control/main.py b/examples/post_training_quantization/openvino/yolov8_quantize_with_accuracy_control/main.py
@@ -222,7 +222,7 @@ def main():
     quantized_model = quantize_ac(ov_model, data_loader, validator)
 
     quantized_model_path = Path(f"{ROOT}/{MODEL_NAME}_openvino_model/{MODEL_NAME}_quantized.xml")
-    ov.save_model(quantized_model, str(quantized_model_path), compress_to_fp16=False)
+    ov.save_model(quantized_model, str(quantized_model_path))
 
     # Validate FP32 model
     fp_stats, total_images, total_objects = validate(ov_model, tqdm(data_loader), validator)

diff --git a/examples/post_training_quantization/tensorflow/mobilenet_v2/main.py b/examples/post_training_quantization/tensorflow/mobilenet_v2/main.py
@@ -154,7 +154,7 @@ def transform_fn(data_item):
 fp32_model_size = get_model_size(fp32_ir_path, verbose=True)
 
 int8_ir_path = f"{ROOT}/mobilenet_v2_int8.xml"
-ov.save_model(ov_quantized_model, int8_ir_path, compress_to_fp16=False)
+ov.save_model(ov_quantized_model, int8_ir_path)
 print(f"[2/7] Save INT8 model: {int8_ir_path}")
 int8_model_size = get_model_size(int8_ir_path, verbose=True)
 

diff --git a/examples/post_training_quantization/torch/mobilenet_v2/main.py b/examples/post_training_quantization/torch/mobilenet_v2/main.py
@@ -159,7 +159,7 @@ def transform_fn(data_item: Tuple[torch.Tensor, int], device: torch.device) -> t
 fp32_model_size = get_model_size(fp32_ir_path, verbose=True)
 
 int8_ir_path = ROOT / "mobilenet_v2_int8.xml"
-ov.save_model(ov_quantized_model, int8_ir_path, compress_to_fp16=False)
+ov.save_model(ov_quantized_model, int8_ir_path)
 print(f"[2/7] Save INT8 model: {int8_ir_path}")
 int8_model_size = get_model_size(int8_ir_path, verbose=True)
 

diff --git a/examples/post_training_quantization/torch/ssd300_vgg16/main.py b/examples/post_training_quantization/torch/ssd300_vgg16/main.py
@@ -171,7 +171,7 @@ def main():
     fp32_model_size = get_model_size(fp32_ir_path, verbose=True)
 
     int8_ir_path = f"{ROOT}/ssd300_vgg16_int8.xml"
-    ov.save_model(ov_quantized_model, int8_ir_path, compress_to_fp16=False)
+    ov.save_model(ov_quantized_model, int8_ir_path)
     print(f"[2/7] Save INT8 model: {int8_ir_path}")
     int8_model_size = get_model_size(int8_ir_path, verbose=True)