Refactor: Removed un-necessary `object` base class (#32230)

* Refactored to remove un-necessary object base class.

* small fix.

examples/research_projects/bertabs/modeling_bertabs.py

@@ -557,7 +557,7 @@ class MultiHeadedAttention(nn.Module):
             return context
 
 
-class DecoderState(object):
+class DecoderState:
     """Interface for grouping together the current state of a recurrent
     decoder. In the simplest case just represents the hidden state of
     the model.  But can also be used for implementing various forms of
@@ -694,7 +694,7 @@ def build_predictor(args, tokenizer, symbols, model, logger=None):
     return translator
 
 
-class GNMTGlobalScorer(object):
+class GNMTGlobalScorer:
     """
     NMT re-ranking score from
     "Google's Neural Machine Translation System" :cite:`wu2016google`
@@ -717,7 +717,7 @@ class GNMTGlobalScorer(object):
         return normalized_probs
 
 
-class PenaltyBuilder(object):
+class PenaltyBuilder:
     """
     Returns the Length and Coverage Penalty function for Beam Search.
 
@@ -763,7 +763,7 @@ class PenaltyBuilder(object):
         return logprobs
 
 
-class Translator(object):
+class Translator:
     """
     Uses a model to translate a batch of sentences.
 
@@ -1002,7 +1002,7 @@ def tile(x, count, dim=0):
 #
 
 
-class BertSumOptimizer(object):
+class BertSumOptimizer:
     """Specific optimizer for BertSum.
 
     As described in [1], the authors fine-tune BertSum for abstractive

examples/research_projects/fsner/src/fsner/tokenizer_utils.py

@@ -3,7 +3,7 @@ import torch
 from transformers import AutoTokenizer
 
 
-class FSNERTokenizerUtils(object):
+class FSNERTokenizerUtils:
     def __init__(self, pretrained_model_name_or_path):
         self.tokenizer = AutoTokenizer.from_pretrained(pretrained_model_name_or_path)
 

examples/research_projects/lxmert/modeling_frcnn.py

@@ -417,7 +417,7 @@ class ShapeSpec(namedtuple("_ShapeSpec", ["channels", "height", "width", "stride
         return super().__new__(cls, channels, height, width, stride)
 
 
-class Box2BoxTransform(object):
+class Box2BoxTransform:
     """
     This R-CNN transformation scales the box's width and height
     by exp(dw), exp(dh) and shifts a box's center by the offset
@@ -519,7 +519,7 @@ class Box2BoxTransform(object):
         return pred_boxes
 
 
-class Matcher(object):
+class Matcher:
     """
     This class assigns to each predicted "element" (e.g., a box) a ground-truth
     element. Each predicted element will have exactly zero or one matches; each
@@ -622,7 +622,7 @@ class Matcher(object):
         match_labels[pred_inds_with_highest_quality] = 1
 
 
-class RPNOutputs(object):
+class RPNOutputs:
     def __init__(
         self,
         box2box_transform,
@@ -1132,7 +1132,7 @@ class ROIPooler(nn.Module):
         return output
 
 
-class ROIOutputs(object):
+class ROIOutputs:
     def __init__(self, cfg, training=False):
         self.smooth_l1_beta = cfg.ROI_BOX_HEAD.SMOOTH_L1_BETA
         self.box2box_transform = Box2BoxTransform(weights=cfg.ROI_BOX_HEAD.BBOX_REG_WEIGHTS)

examples/research_projects/movement-pruning/emmental/modules/binarizer.py

@@ -108,7 +108,7 @@ class TopKBinarizer(autograd.Function):
         return gradOutput, None
 
 
-class MagnitudeBinarizer(object):
+class MagnitudeBinarizer:
     """
     Magnitude Binarizer.
     Computes a binary mask M from a real value matrix S such that `M_{i,j} = 1` if and only if `S_{i,j}`

examples/research_projects/performer/modeling_flax_performer_utils.py

@@ -284,7 +284,7 @@ def make_fast_generalized_attention(
     return attention_fn
 
 
-class RandomMatrix(object):
+class RandomMatrix:
     r"""
     Abstract class providing a method for constructing 2D random arrays. Class is responsible for constructing 2D
     random arrays.
@@ -348,7 +348,7 @@ class GaussianOrthogonalRandomMatrix(RandomMatrix):
         return jnp.matmul(jnp.diag(multiplier), final_matrix)
 
 
-class FastAttention(object):
+class FastAttention:
     r"""
     Abstract class providing a method for fast attention. Class is responsible for providing a method
     <dot_product_attention> for fast approximate attention.

examples/research_projects/visual_bert/modeling_frcnn.py

@@ -417,7 +417,7 @@ class ShapeSpec(namedtuple("_ShapeSpec", ["channels", "height", "width", "stride
         return super().__new__(cls, channels, height, width, stride)
 
 
-class Box2BoxTransform(object):
+class Box2BoxTransform:
     """
     This R-CNN transformation scales the box's width and height
     by exp(dw), exp(dh) and shifts a box's center by the offset
@@ -519,7 +519,7 @@ class Box2BoxTransform(object):
         return pred_boxes
 
 
-class Matcher(object):
+class Matcher:
     """
     This class assigns to each predicted "element" (e.g., a box) a ground-truth
     element. Each predicted element will have exactly zero or one matches; each
@@ -622,7 +622,7 @@ class Matcher(object):
         match_labels[pred_inds_with_highest_quality] = 1
 
 
-class RPNOutputs(object):
+class RPNOutputs:
     def __init__(
         self,
         box2box_transform,
@@ -1132,7 +1132,7 @@ class ROIPooler(nn.Module):
         return output
 
 
-class ROIOutputs(object):
+class ROIOutputs:
     def __init__(self, cfg, training=False):
         self.smooth_l1_beta = cfg.ROI_BOX_HEAD.SMOOTH_L1_BETA
         self.box2box_transform = Box2BoxTransform(weights=cfg.ROI_BOX_HEAD.BBOX_REG_WEIGHTS)

src/transformers/models/bert/tokenization_bert.py

@@ -281,7 +281,7 @@ class BertTokenizer(PreTrainedTokenizer):
         return (vocab_file,)
 
 
-class BasicTokenizer(object):
+class BasicTokenizer:
     """
     Constructs a BasicTokenizer that will run basic tokenization (punctuation splitting, lower casing, etc.).
 
@@ -442,7 +442,7 @@ class BasicTokenizer(object):
         return "".join(output)
 
 
-class WordpieceTokenizer(object):
+class WordpieceTokenizer:
     """Runs WordPiece tokenization."""
 
     def __init__(self, vocab, unk_token, max_input_chars_per_word=100):

src/transformers/models/bert_japanese/tokenization_bert_japanese.py

@@ -691,7 +691,7 @@ class CharacterTokenizer:
 
 
 # Copied from transformers.models.bert.tokenization_bert.BasicTokenizer
-class BasicTokenizer(object):
+class BasicTokenizer:
     """
     Constructs a BasicTokenizer that will run basic tokenization (punctuation splitting, lower casing, etc.).
 
@@ -853,7 +853,7 @@ class BasicTokenizer(object):
 
 
 # Copied from transformers.models.bert.tokenization_bert.WordpieceTokenizer
-class WordpieceTokenizer(object):
+class WordpieceTokenizer:
     """Runs WordPiece tokenization."""
 
     def __init__(self, vocab, unk_token, max_input_chars_per_word=100):
@@ -910,7 +910,7 @@ class WordpieceTokenizer(object):
         return output_tokens
 
 
-class SentencepieceTokenizer(object):
+class SentencepieceTokenizer:
     """
     Runs sentencepiece tokenization. Based on transformers.models.albert.tokenization_albert.AlbertTokenizer.
     """

src/transformers/models/clip/tokenization_clip.py

@@ -90,7 +90,7 @@ def whitespace_tokenize(text):
 
 
 # Copied from transformers.models.bert.tokenization_bert.BasicTokenizer
-class BasicTokenizer(object):
+class BasicTokenizer:
     """
     Constructs a BasicTokenizer that will run basic tokenization (punctuation splitting, lower casing, etc.).
 

src/transformers/models/conditional_detr/modeling_conditional_detr.py

@@ -2596,7 +2596,7 @@ def _max_by_axis(the_list):
 
 
 # Copied from transformers.models.detr.modeling_detr.NestedTensor
-class NestedTensor(object):
+class NestedTensor:
     def __init__(self, tensors, mask: Optional[Tensor]):
         self.tensors = tensors
         self.mask = mask

src/transformers/models/convbert/tokenization_convbert.py

@@ -285,7 +285,7 @@ class ConvBertTokenizer(PreTrainedTokenizer):
 
 
 # Copied from transformers.models.bert.tokenization_bert.BasicTokenizer
-class BasicTokenizer(object):
+class BasicTokenizer:
     """
     Constructs a BasicTokenizer that will run basic tokenization (punctuation splitting, lower casing, etc.).
 
@@ -447,7 +447,7 @@ class BasicTokenizer(object):
 
 
 # Copied from transformers.models.bert.tokenization_bert.WordpieceTokenizer
-class WordpieceTokenizer(object):
+class WordpieceTokenizer:
     """Runs WordPiece tokenization."""
 
     def __init__(self, vocab, unk_token, max_input_chars_per_word=100):

src/transformers/models/cpmant/tokenization_cpmant.py

@@ -44,7 +44,7 @@ def load_vocab(vocab_file):
     return vocab
 
 
-class WordpieceTokenizer(object):
+class WordpieceTokenizer:
     def __init__(self, vocab, unk_token="<unk>", max_input_chars_per_word=200):
         self.vocab = vocab
         self.unk_token = unk_token

src/transformers/models/deberta/modeling_deberta.py

@@ -138,7 +138,7 @@ class XSoftmax(torch.autograd.Function):
         return masked_fill(g, output, r_mask, g.op("Constant", value_t=torch.tensor(0, dtype=torch.bool)))
 
 
-class DropoutContext(object):
+class DropoutContext:
     def __init__(self):
         self.dropout = 0
         self.mask = None

src/transformers/models/deberta_v2/modeling_deberta_v2.py

@@ -133,7 +133,7 @@ class XSoftmax(torch.autograd.Function):
 
 
 # Copied from transformers.models.deberta.modeling_deberta.DropoutContext
-class DropoutContext(object):
+class DropoutContext:
     def __init__(self):
         self.dropout = 0
         self.mask = None

src/transformers/models/deformable_detr/modeling_deformable_detr.py

@@ -2491,7 +2491,7 @@ def _max_by_axis(the_list):
 
 
 # Copied from transformers.models.detr.modeling_detr.NestedTensor
-class NestedTensor(object):
+class NestedTensor:
     def __init__(self, tensors, mask: Optional[Tensor]):
         self.tensors = tensors
         self.mask = mask

src/transformers/models/deprecated/deta/modeling_deta.py

@@ -2516,7 +2516,7 @@ def nonzero_tuple(x):
 
 
 # from https://github.com/facebookresearch/detectron2/blob/9921a2caa585d4fa66c4b534b6fab6e74d89b582/detectron2/modeling/matcher.py#L9
-class DetaMatcher(object):
+class DetaMatcher:
     """
     This class assigns to each predicted "element" (e.g., a box) a ground-truth element. Each predicted element will
     have exactly zero or one matches; each ground-truth element may be matched to zero or more predicted elements.

src/transformers/models/deprecated/gptsan_japanese/tokenization_gptsan_japanese.py

@@ -365,7 +365,7 @@ class GPTSanJapaneseTokenizer(PreTrainedTokenizer):
         )
 
 
-class SubWordJapaneseTokenizer(object):
+class SubWordJapaneseTokenizer:
     """
     This tokenizer is based on GPTNeoXJapaneseTokenizer and has the following modifications
     - Decoding byte0~byte255 tokens correctly

src/transformers/models/deprecated/mmbt/configuration_mmbt.py

@@ -21,7 +21,7 @@ from ....utils import logging
 logger = logging.get_logger(__name__)
 
 
-class MMBTConfig(object):
+class MMBTConfig:
     """
     This is the configuration class to store the configuration of a [`MMBTModel`]. It is used to instantiate a MMBT
     model according to the specified arguments, defining the model architecture.

src/transformers/models/deprecated/realm/tokenization_realm.py

@@ -354,7 +354,7 @@ class RealmTokenizer(PreTrainedTokenizer):
         return (vocab_file,)
 
 
-class BasicTokenizer(object):
+class BasicTokenizer:
     """
     Constructs a BasicTokenizer that will run basic tokenization (punctuation splitting, lower casing, etc.).
 
@@ -503,7 +503,7 @@ class BasicTokenizer(object):
         return "".join(output)
 
 
-class WordpieceTokenizer(object):
+class WordpieceTokenizer:
     """Runs WordPiece tokenization."""
 
     def __init__(self, vocab, unk_token, max_input_chars_per_word=100):

src/transformers/models/deprecated/retribert/tokenization_retribert.py

@@ -283,7 +283,7 @@ class RetriBertTokenizer(PreTrainedTokenizer):
         return (vocab_file,)
 
 
-class BasicTokenizer(object):
+class BasicTokenizer:
     """
     Constructs a BasicTokenizer that will run basic tokenization (punctuation splitting, lower casing, etc.).
 
@@ -444,7 +444,7 @@ class BasicTokenizer(object):
         return "".join(output)
 
 
-class WordpieceTokenizer(object):
+class WordpieceTokenizer:
     """Runs WordPiece tokenization."""
 
     def __init__(self, vocab, unk_token, max_input_chars_per_word=100):