Update quality tooling for formatting (#760)

Author: regisss

.github/generate_dev_tests.py

@@ -1,4 +1,3 @@
-import shutil
 from pathlib import Path
 
 import yaml

.github/workflows/check_code_quality.yml

@@ -47,7 +47,7 @@ jobs:
       run: |
         source venv/bin/activate
         black --check .
-    - name: Check style with isort
+    - name: Check style with ruff
       run: |
         source venv/bin/activate
-        isort --check .
+        ruff .

CONTRIBUTING.md

@@ -51,7 +51,7 @@ If you would like to work on any of the open Issues:
 
 5. Develop the features on your branch.
 
-6. Format your code. Run black and isort so that your newly added files look nice with the following command:
+6. Format your code. Run black and ruff so that your newly added files look nice with the following command:
 
 	```bash
 	make style

Makefile

@@ -23,11 +23,11 @@ REAL_CLONE_URL = $(if $(CLONE_URL),$(CLONE_URL),$(DEFAULT_CLONE_URL))
 # Run code quality checks
 style_check:
 	black --check .
-	isort --check .
+	ruff .
 
 style:
 	black .
-	isort .
+	ruff . --fix
 
 # Run tests for the library
 test:

examples/onnxruntime/optimization/multiple-choice/run_swag.py

@@ -25,20 +25,18 @@
 from dataclasses import dataclass, field
 from functools import partial
 from itertools import chain
-from typing import Optional, Union
+from typing import Optional
 
 import datasets
 import numpy as np
-import torch
 import transformers
 from datasets import load_dataset
 from transformers import AutoTokenizer, HfArgumentParser, TrainingArguments
 from transformers.tokenization_utils_base import PreTrainedTokenizerBase
-from transformers.trainer_utils import get_last_checkpoint
 from transformers.utils import check_min_version
 
 from optimum.onnxruntime import ORTModelForMultipleChoice, ORTOptimizer
-from optimum.onnxruntime.configuration import OptimizationConfig, ORTConfig
+from optimum.onnxruntime.configuration import OptimizationConfig
 from optimum.onnxruntime.model import ORTModel
 
 
@@ -238,7 +236,6 @@ def main():
         )
 
     os.makedirs(training_args.output_dir, exist_ok=True)
-    model_path = os.path.join(training_args.output_dir, "model.onnx")
     optimized_model_path = os.path.join(training_args.output_dir, "model_optimized.onnx")
 
     tokenizer = AutoTokenizer.from_pretrained(model_args.tokenizer_name or model_args.model_name_or_path)

examples/onnxruntime/optimization/question-answering/run_qa.py

@@ -24,22 +24,20 @@
 import sys
 from dataclasses import dataclass, field
 from functools import partial
-from pathlib import Path
 from typing import Optional
 
 import datasets
 import transformers
 from datasets import load_dataset
+from evaluate import load
 from transformers import AutoTokenizer, EvalPrediction, HfArgumentParser, PreTrainedTokenizer, TrainingArguments
 from transformers.utils import check_min_version
 from transformers.utils.versions import require_version
+from utils_qa import postprocess_qa_predictions
 
-from evaluate import load
 from optimum.onnxruntime import ORTModelForQuestionAnswering, ORTOptimizer
-from optimum.onnxruntime.configuration import OptimizationConfig, ORTConfig
+from optimum.onnxruntime.configuration import OptimizationConfig
 from optimum.onnxruntime.model import ORTModel
-from trainer_qa import QuestionAnsweringTrainer
-from utils_qa import postprocess_qa_predictions
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
@@ -307,7 +305,6 @@ def main():
         )
 
     os.makedirs(training_args.output_dir, exist_ok=True)
-    model_path = os.path.join(training_args.output_dir, "model.onnx")
     optimized_model_path = os.path.join(training_args.output_dir, "model_optimized.onnx")
 
     tokenizer = AutoTokenizer.from_pretrained(model_args.tokenizer_name or model_args.model_name_or_path)
@@ -492,7 +489,7 @@ def compute_metrics(p: EvalPrediction):
         metrics = compute_metrics(predictions)
 
         # Save metrics
-        with open(os.path.join(training_args.output_dir, f"eval_results.json"), "w") as f:
+        with open(os.path.join(training_args.output_dir, "eval_results.json"), "w") as f:
             json.dump(metrics, f, indent=4, sort_keys=True)
 
     # Prediction
@@ -527,7 +524,7 @@ def compute_metrics(p: EvalPrediction):
         metrics = compute_metrics(predictions)
 
         # Save metrics
-        with open(os.path.join(training_args.output_dir, f"predict_results.json"), "w") as f:
+        with open(os.path.join(training_args.output_dir, "predict_results.json"), "w") as f:
             json.dump(metrics, f, indent=4, sort_keys=True)
 
 

examples/onnxruntime/optimization/text-classification/run_glue.py

@@ -23,15 +23,14 @@
 import sys
 from dataclasses import dataclass, field
 from functools import partial
-from pathlib import Path
 from typing import Optional
 
 import datasets
 import numpy as np
 import transformers
 from datasets import load_dataset
+from evaluate import load
 from transformers import (
-    AutoConfig,
     AutoTokenizer,
     EvalPrediction,
     HfArgumentParser,
@@ -41,9 +40,8 @@
 from transformers.utils import check_min_version
 from transformers.utils.versions import require_version
 
-from evaluate import load
 from optimum.onnxruntime import ORTModelForSequenceClassification, ORTOptimizer
-from optimum.onnxruntime.configuration import OptimizationConfig, ORTConfig
+from optimum.onnxruntime.configuration import OptimizationConfig
 from optimum.onnxruntime.model import ORTModel
 
 
@@ -395,7 +393,7 @@ def compute_metrics(p: EvalPrediction):
 
         try:
             eval_dataset = eval_dataset.align_labels_with_mapping(label2id=model.config.label2id, label_column="label")
-        except Exception as e:
+        except Exception:
             logger.warning(
                 f"\nModel label mapping: {model.config.label2id}"
                 f"\nDataset label features: {eval_dataset.features['label']}"
@@ -418,7 +416,7 @@ def compute_metrics(p: EvalPrediction):
         )
         outputs = ort_model.evaluation_loop(eval_dataset)
         # Save metrics
-        with open(os.path.join(training_args.output_dir, f"eval_results.json"), "w") as f:
+        with open(os.path.join(training_args.output_dir, "eval_results.json"), "w") as f:
             json.dump(outputs.metrics, f, indent=4, sort_keys=True)
 
     # Prediction
@@ -445,7 +443,7 @@ def compute_metrics(p: EvalPrediction):
         predictions = np.squeeze(outputs.predictions) if is_regression else np.argmax(outputs.predictions, axis=1)
 
         # Save predictions
-        output_predictions_file = os.path.join(training_args.output_dir, f"prediction.txt")
+        output_predictions_file = os.path.join(training_args.output_dir, "prediction.txt")
         with open(output_predictions_file, "w") as writer:
             logger.info(f"***** Predict results {data_args.task_name} *****")
             writer.write("index\tprediction\n")

examples/onnxruntime/optimization/token-classification/run_ner.py

@@ -31,13 +31,13 @@
 import numpy as np
 import transformers
 from datasets import ClassLabel, load_dataset
+from evaluate import load
 from transformers import AutoTokenizer, HfArgumentParser, PreTrainedTokenizer, TrainingArguments
 from transformers.utils import check_min_version
 from transformers.utils.versions import require_version
 
-from evaluate import load
 from optimum.onnxruntime import ORTModelForTokenClassification, ORTOptimizer
-from optimum.onnxruntime.configuration import OptimizationConfig, ORTConfig
+from optimum.onnxruntime.configuration import OptimizationConfig
 from optimum.onnxruntime.model import ORTModel
 
 
@@ -276,7 +276,6 @@ def main():
         )
 
     os.makedirs(training_args.output_dir, exist_ok=True)
-    model_path = os.path.join(training_args.output_dir, "model.onnx")
     optimized_model_path = os.path.join(training_args.output_dir, "model_optimized.onnx")
 
     tokenizer = AutoTokenizer.from_pretrained(model_args.tokenizer_name or model_args.model_name_or_path)
@@ -357,7 +356,7 @@ def main():
                 eval_dataset = eval_dataset.align_labels_with_mapping(
                     label2id=model.config.label2id, label_column=label_column_name
                 )
-            except Exception as e:
+            except Exception:
                 logger.warning(
                     f"\nModel label mapping: {model.config.label2id}"
                     f"\nDataset label features: {eval_dataset.features[label_column_name]}"
@@ -489,7 +488,7 @@ def compute_metrics(p):
         outputs = ort_model.evaluation_loop(eval_dataset)
 
         # Save evaluation metrics
-        with open(os.path.join(training_args.output_dir, f"eval_results.json"), "w") as f:
+        with open(os.path.join(training_args.output_dir, "eval_results.json"), "w") as f:
             json.dump(outputs.metrics, f, indent=4, sort_keys=True)
 
     # Prediction
@@ -525,7 +524,7 @@ def compute_metrics(p):
         ]
 
         # Save test metrics
-        with open(os.path.join(training_args.output_dir, f"predict_results.json"), "w") as f:
+        with open(os.path.join(training_args.output_dir, "predict_results.json"), "w") as f:
             json.dump(outputs.metrics, f, indent=4, sort_keys=True)
 
         # Save predictions

examples/onnxruntime/quantization/image-classification/run_image_classification.py

@@ -30,13 +30,12 @@
 import torch
 import transformers
 from datasets import load_dataset
+from evaluate import load
+from onnxruntime.quantization import QuantFormat, QuantizationMode, QuantType
 from torchvision.transforms import CenterCrop, Compose, Normalize, Resize, ToTensor
 from transformers import AutoFeatureExtractor, EvalPrediction, HfArgumentParser, TrainingArguments
-from transformers.onnx import FeaturesManager
 from transformers.utils.versions import require_version
 
-from evaluate import load
-from onnxruntime.quantization import QuantFormat, QuantizationMode, QuantType
 from optimum.onnxruntime import ORTQuantizer
 from optimum.onnxruntime.configuration import AutoCalibrationConfig, QuantizationConfig
 from optimum.onnxruntime.model import ORTModel
@@ -399,7 +398,7 @@ def compute_metrics(p: EvalPrediction):
             eval_dataset = eval_dataset.align_labels_with_mapping(
                 label2id=model.config.label2id, label_column=labels_column
             )
-        except Exception as e:
+        except Exception:
             logger.warning(
                 f"\nModel label mapping: {model.config.label2id}"
                 f"\nDataset label features: {eval_dataset.features[labels_column]}"
@@ -418,7 +417,7 @@ def compute_metrics(p: EvalPrediction):
         )
         outputs = ort_model.evaluation_loop(eval_dataset)
         # Save metrics
-        with open(os.path.join(training_args.output_dir, f"eval_results.json"), "w") as f:
+        with open(os.path.join(training_args.output_dir, "eval_results.json"), "w") as f:
             json.dump(outputs.metrics, f, indent=4, sort_keys=True)
 
 

examples/onnxruntime/quantization/multiple-choice/run_swag.py

@@ -25,19 +25,17 @@
 from dataclasses import dataclass, field
 from functools import partial
 from itertools import chain
-from typing import Optional, Union
+from typing import Optional
 
 import datasets
 import numpy as np
-import torch
 import transformers
 from datasets import load_dataset
+from onnxruntime.quantization import QuantFormat, QuantizationMode, QuantType
 from transformers import AutoTokenizer, HfArgumentParser, TrainingArguments
 from transformers.tokenization_utils_base import PreTrainedTokenizerBase
-from transformers.trainer_utils import get_last_checkpoint
 from transformers.utils import check_min_version
 
-from onnxruntime.quantization import QuantFormat, QuantizationMode, QuantType
 from optimum.onnxruntime import ORTModelForMultipleChoice, ORTQuantizer
 from optimum.onnxruntime.configuration import AutoCalibrationConfig, QuantizationConfig
 from optimum.onnxruntime.model import ORTModel

examples/onnxruntime/quantization/question-answering/run_qa.py

@@ -30,13 +30,13 @@
 import datasets
 import transformers
 from datasets import load_dataset
+from evaluate import load
+from onnxruntime.quantization import QuantFormat, QuantizationMode, QuantType
 from transformers import AutoTokenizer, EvalPrediction, HfArgumentParser, PreTrainedTokenizer, TrainingArguments
-from transformers.onnx import FeaturesManager
 from transformers.utils import check_min_version
 from transformers.utils.versions import require_version
+from utils_qa import postprocess_qa_predictions
 
-from evaluate import load
-from onnxruntime.quantization import QuantFormat, QuantizationMode, QuantType
 from optimum.onnxruntime import ORTQuantizer
 from optimum.onnxruntime.configuration import AutoCalibrationConfig, QuantizationConfig
 from optimum.onnxruntime.model import ORTModel
@@ -48,8 +48,6 @@
     ExcludeNodeAfter,
     ExcludeNodeFollowedBy,
 )
-from trainer_qa import QuestionAnsweringTrainer
-from utils_qa import postprocess_qa_predictions
 
 
 # Will error if the minimal version of Transformers is not installed. Remove at your own risks.
@@ -676,7 +674,7 @@ def compute_metrics(p: EvalPrediction):
         metrics = compute_metrics(predictions)
 
         # Save metrics
-        with open(os.path.join(training_args.output_dir, f"eval_results.json"), "w") as f:
+        with open(os.path.join(training_args.output_dir, "eval_results.json"), "w") as f:
             json.dump(metrics, f, indent=4, sort_keys=True)
 
     # Prediction
@@ -693,7 +691,7 @@ def compute_metrics(p: EvalPrediction):
         metrics = compute_metrics(predictions)
 
         # Save metrics
-        with open(os.path.join(training_args.output_dir, f"predict_results.json"), "w") as f:
+        with open(os.path.join(training_args.output_dir, "predict_results.json"), "w") as f:
             json.dump(metrics, f, indent=4, sort_keys=True)
 
 

examples/onnxruntime/quantization/text-classification/run_glue.py

@@ -30,20 +30,18 @@
 import numpy as np
 import transformers
 from datasets import load_dataset
+from evaluate import load
+from onnxruntime.quantization import QuantFormat, QuantizationMode, QuantType
 from transformers import (
-    AutoConfig,
     AutoTokenizer,
     EvalPrediction,
     HfArgumentParser,
     PreTrainedTokenizer,
     TrainingArguments,
 )
-from transformers.onnx import FeaturesManager
 from transformers.utils import check_min_version
 from transformers.utils.versions import require_version
 
-from evaluate import load
-from onnxruntime.quantization import QuantFormat, QuantizationMode, QuantType
 from optimum.onnxruntime import ORTQuantizer
 from optimum.onnxruntime.configuration import AutoCalibrationConfig, QuantizationConfig
 from optimum.onnxruntime.model import ORTModel
@@ -498,9 +496,9 @@ def compute_metrics(p: EvalPrediction):
 
         try:
             eval_dataset = eval_dataset.align_labels_with_mapping(label2id=model.config.label2id, label_column="label")
-        except Exception as e:
+        except Exception:
             logger.warning(
-                f"\nModel label mapping: {onnx_model.config.label2id}"
+                f"\nModel label mapping: {model.config.label2id}"
                 f"\nDataset label features: {eval_dataset.features['label']}"
                 f"\nCould not guarantee the model label mapping and the dataset labels match."
                 f" Evaluation results may suffer from a wrong matching."
@@ -514,7 +512,7 @@ def compute_metrics(p: EvalPrediction):
         )
         outputs = ort_model.evaluation_loop(eval_dataset)
         # Save metrics
-        with open(os.path.join(training_args.output_dir, f"eval_results.json"), "w") as f:
+        with open(os.path.join(training_args.output_dir, "eval_results.json"), "w") as f:
             json.dump(outputs.metrics, f, indent=4, sort_keys=True)
 
     # Prediction
@@ -536,7 +534,7 @@ def compute_metrics(p: EvalPrediction):
         predictions = np.squeeze(outputs.predictions) if is_regression else np.argmax(outputs.predictions, axis=1)
 
         # Save predictions
-        output_predictions_file = os.path.join(training_args.output_dir, f"prediction.txt")
+        output_predictions_file = os.path.join(training_args.output_dir, "prediction.txt")
         with open(output_predictions_file, "w") as writer:
             logger.info(f"***** Predict results {data_args.task_name} *****")
             writer.write("index\tprediction\n")