Migrate metric to Evaluate in Pytorch examples (#18369)

* Migrate metric to Evaluate in pytorch examples

* Remove unused imports

examples/pytorch/question-answering/run_qa.py

@@ -25,8 +25,9 @@ from dataclasses import dataclass, field
 from typing import Optional
 
 import datasets
-from datasets import load_dataset, load_metric
+from datasets import load_dataset
 
+import evaluate
 import transformers
 from trainer_qa import QuestionAnsweringTrainer
 from transformers import (
@@ -593,7 +594,7 @@ def main():
         references = [{"id": ex["id"], "answers": ex[answer_column_name]} for ex in examples]
         return EvalPrediction(predictions=formatted_predictions, label_ids=references)
 
-    metric = load_metric("squad_v2" if data_args.version_2_with_negative else "squad")
+    metric = evaluate.load("squad_v2" if data_args.version_2_with_negative else "squad")
 
     def compute_metrics(p: EvalPrediction):
         return metric.compute(predictions=p.predictions, references=p.label_ids)

examples/pytorch/question-answering/run_qa_beam_search.py

@@ -25,8 +25,9 @@ from dataclasses import dataclass, field
 from typing import Optional
 
 import datasets
-from datasets import load_dataset, load_metric
+from datasets import load_dataset
 
+import evaluate
 import transformers
 from trainer_qa import QuestionAnsweringTrainer
 from transformers import (
@@ -625,7 +626,7 @@ def main():
         references = [{"id": ex["id"], "answers": ex[answer_column_name]} for ex in examples]
         return EvalPrediction(predictions=formatted_predictions, label_ids=references)
 
-    metric = load_metric("squad_v2" if data_args.version_2_with_negative else "squad")
+    metric = evaluate.load("squad_v2" if data_args.version_2_with_negative else "squad")
 
     def compute_metrics(p: EvalPrediction):
         return metric.compute(predictions=p.predictions, references=p.label_ids)

examples/pytorch/question-answering/run_qa_beam_search_no_trainer.py

@@ -29,10 +29,11 @@ from pathlib import Path
 import datasets
 import numpy as np
 import torch
-from datasets import load_dataset, load_metric
+from datasets import load_dataset
 from torch.utils.data import DataLoader
 from tqdm.auto import tqdm
 
+import evaluate
 import transformers
 from accelerate import Accelerator
 from accelerate.logging import get_logger
@@ -680,7 +681,7 @@ def main():
         references = [{"id": ex["id"], "answers": ex[answer_column_name]} for ex in examples]
         return EvalPrediction(predictions=formatted_predictions, label_ids=references)
 
-    metric = load_metric("squad_v2" if args.version_2_with_negative else "squad")
+    metric = evaluate.load("squad_v2" if args.version_2_with_negative else "squad")
 
     def create_and_fill_np_array(start_or_end_logits, dataset, max_len):
         """

examples/pytorch/question-answering/run_qa_no_trainer.py

@@ -29,10 +29,11 @@ from pathlib import Path
 import datasets
 import numpy as np
 import torch
-from datasets import load_dataset, load_metric
+from datasets import load_dataset
 from torch.utils.data import DataLoader
 from tqdm.auto import tqdm
 
+import evaluate
 import transformers
 from accelerate import Accelerator
 from accelerate.logging import get_logger
@@ -696,7 +697,7 @@ def main():
         references = [{"id": ex["id"], "answers": ex[answer_column_name]} for ex in examples]
         return EvalPrediction(predictions=formatted_predictions, label_ids=references)
 
-    metric = load_metric("squad_v2" if args.version_2_with_negative else "squad")
+    metric = evaluate.load("squad_v2" if args.version_2_with_negative else "squad")
 
     # Create and fill numpy array of size len_of_validation_data * max_length_of_output_tensor
     def create_and_fill_np_array(start_or_end_logits, dataset, max_len):

examples/pytorch/question-answering/run_seq2seq_qa.py

@@ -25,8 +25,9 @@ from dataclasses import dataclass, field
 from typing import List, Optional, Tuple
 
 import datasets
-from datasets import load_dataset, load_metric
+from datasets import load_dataset
 
+import evaluate
 import transformers
 from trainer_seq2seq_qa import QuestionAnsweringSeq2SeqTrainer
 from transformers import (
@@ -581,7 +582,7 @@ def main():
         pad_to_multiple_of=8 if training_args.fp16 else None,
     )
 
-    metric = load_metric("squad_v2" if data_args.version_2_with_negative else "squad")
+    metric = evaluate.load("squad_v2" if data_args.version_2_with_negative else "squad")
 
     def compute_metrics(p: EvalPrediction):
         return metric.compute(predictions=p.predictions, references=p.label_ids)