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

examples/research_projects/onnx/summarization/bart_onnx/generation_onnx.py

@@ -264,7 +264,6 @@ class BARTGenerator(torch.nn.Module, GenerationMixin):
 
         past: List[torch.Tensor] = []
         while cur_len < max_length:
-
             logits, past = self._decoder_forward(input_ids, encoder_output, attention_mask, past)
             next_token_logits = logits[:, -1, :]
 
@@ -303,7 +302,6 @@ class BARTGenerator(torch.nn.Module, GenerationMixin):
         decoder_start_token_id,
         bos_token_id: Optional[int] = None,
     ) -> torch.LongTensor:
-
         decoder_input_ids = (
             torch.ones((input_ids.shape[0], 1), dtype=input_ids.dtype, device=input_ids.device)
             * decoder_start_token_id
@@ -633,7 +631,6 @@ class BARTBeamSearchGenerator(BARTGenerator):
     def beam_search(
         self, input_ids, encoder_output, attention_mask, num_beams, max_length, pad_token_id: int, eos_token_id: int
     ):
-
         batch_size = self.beam_scorer.batch_size
 
         num_beams = self.beam_scorer.num_beams

examples/research_projects/onnx/summarization/bart_onnx/reduce_onnx_size.py

@@ -5,7 +5,6 @@ Code to remove duplicate initializers to reduce ONNX model size.
 import os
 
 import numpy
-
 import onnx
 
 

examples/research_projects/onnx/summarization/run_onnx_exporter.py

@@ -22,12 +22,12 @@ import os
 import sys
 
 import numpy as np
-import torch
-
 import onnxruntime
-import transformers
+import torch
 from bart_onnx.generation_onnx import BARTBeamSearchGenerator
 from bart_onnx.reduce_onnx_size import remove_dup_initializers
+
+import transformers
 from transformers import BartForConditionalGeneration, BartTokenizer