Update quality tooling for formatting (#21480)

* Result of black 23.1

* Update target to Python 3.7

* Switch flake8 to ruff

* Configure isort

* Configure isort

* Apply isort with line limit

* Put the right black version

* adapt black in check copies

* Fix copies

tests/trainer/test_trainer.py

@@ -29,10 +29,10 @@ from pathlib import Path
 from unittest.mock import Mock, patch
 
 import numpy as np
-
 from huggingface_hub import HfFolder, Repository, delete_repo, set_access_token
 from parameterized import parameterized
 from requests.exceptions import HTTPError
+
 from transformers import (
     AutoTokenizer,
     IntervalStrategy,
@@ -565,7 +565,6 @@ class TrainerIntegrationPrerunTest(TestCasePlus, TrainerIntegrationCommon):
     @require_torch_gpu
     @require_torch_bf16_gpu
     def test_mixed_bf16(self):
-
         # very basic test
         trainer = get_regression_trainer(learning_rate=0.1, bf16=True)
         trainer.train()
@@ -580,7 +579,6 @@ class TrainerIntegrationPrerunTest(TestCasePlus, TrainerIntegrationCommon):
     @require_torch_gpu
     @require_torch_tf32
     def test_tf32(self):
-
         # very basic test
         trainer = get_regression_trainer(learning_rate=0.1, tf32=True)
         trainer.train()
@@ -1289,7 +1287,6 @@ class TrainerIntegrationTest(TestCasePlus, TrainerIntegrationCommon):
     @require_accelerate
     @require_torch_non_multi_gpu
     def test_auto_batch_size_finder(self):
-
         if torch.cuda.is_available():
             torch.backends.cudnn.deterministic = True
 
@@ -1736,7 +1733,6 @@ class TrainerIntegrationTest(TestCasePlus, TrainerIntegrationCommon):
             check_func("test_mem_gpu_alloc_delta", metrics)
 
     def test_mem_metrics(self):
-
         # with mem metrics enabled
         trainer = get_regression_trainer(skip_memory_metrics=False)
         self.check_mem_metrics(trainer, self.assertIn)
@@ -1747,7 +1743,6 @@ class TrainerIntegrationTest(TestCasePlus, TrainerIntegrationCommon):
 
     @require_torch_gpu
     def test_fp16_full_eval(self):
-
         # this is a sensitive test so let's keep debugging printouts in place for quick diagnosis.
         # it's using pretty large safety margins, but small enough to detect broken functionality.
         debug = 0
@@ -2467,7 +2462,6 @@ class TrainerHyperParameterWandbIntegrationTest(unittest.TestCase):
             DEFAULTS = {"a": 0, "b": 0}
 
         def hp_space(trial):
-
             return {
                 "method": "random",
                 "metric": {},

tests/trainer/test_trainer_distributed.py

@@ -66,7 +66,6 @@ if is_torch_available():
 class TestTrainerDistributedNeuronCore(TestCasePlus):
     @require_torch_neuroncore
     def test_trainer(self):
-
         distributed_args = f"""
             -m torch.distributed.launch
             --nproc_per_node=2
@@ -83,7 +82,6 @@ class TestTrainerDistributedNeuronCore(TestCasePlus):
 class TestTrainerDistributed(TestCasePlus):
     @require_torch_multi_gpu
     def test_trainer(self):
-
         distributed_args = f"""
             -m torch.distributed.launch
             --nproc_per_node={torch.cuda.device_count()}