# coding=utf-8 # Copyright 2019 The Google AI Language Team Authors and The HuggingFace Inc. team. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. """PyTorch ELECTRA model. """ import math import os from dataclasses import dataclass from typing import Optional, Tuple import torch import torch.utils.checkpoint from packaging import version from torch import nn from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss from ...activations import ACT2FN, get_activation from ...file_utils import ( ModelOutput, add_code_sample_docstrings, add_start_docstrings, add_start_docstrings_to_model_forward, replace_return_docstrings, ) from ...modeling_outputs import ( BaseModelOutputWithCrossAttentions, BaseModelOutputWithPastAndCrossAttentions, MaskedLMOutput, MultipleChoiceModelOutput, QuestionAnsweringModelOutput, SequenceClassifierOutput, TokenClassifierOutput, ) from ...modeling_utils import ( PreTrainedModel, SequenceSummary, apply_chunking_to_forward, find_pruneable_heads_and_indices, prune_linear_layer, ) from ...utils import logging from .configuration_electra import ElectraConfig logger = logging.get_logger(__name__) _CHECKPOINT_FOR_DOC = "google/electra-small-discriminator" _CONFIG_FOR_DOC = "ElectraConfig" _TOKENIZER_FOR_DOC = "ElectraTokenizer" ELECTRA_PRETRAINED_MODEL_ARCHIVE_LIST = [ "google/electra-small-generator", "google/electra-base-generator", "google/electra-large-generator", "google/electra-small-discriminator", "google/electra-base-discriminator", "google/electra-large-discriminator", # See all ELECTRA models at https://huggingface.co/models?filter=electra ] def load_tf_weights_in_electra(model, config, tf_checkpoint_path, discriminator_or_generator="discriminator"): """Load tf checkpoints in a pytorch model.""" try: import re import numpy as np import tensorflow as tf except ImportError: logger.error( "Loading a TensorFlow model in PyTorch, requires TensorFlow to be installed. Please see " "https://www.tensorflow.org/install/ for installation instructions." ) raise tf_path = os.path.abspath(tf_checkpoint_path) logger.info(f"Converting TensorFlow checkpoint from {tf_path}") # Load weights from TF model init_vars = tf.train.list_variables(tf_path) names = [] arrays = [] for name, shape in init_vars: logger.info(f"Loading TF weight {name} with shape {shape}") array = tf.train.load_variable(tf_path, name) names.append(name) arrays.append(array) for name, array in zip(names, arrays): original_name: str = name try: if isinstance(model, ElectraForMaskedLM): name = name.replace("electra/embeddings/", "generator/embeddings/") if discriminator_or_generator == "generator": name = name.replace("electra/", "discriminator/") name = name.replace("generator/", "electra/") name = name.replace("dense_1", "dense_prediction") name = name.replace("generator_predictions/output_bias", "generator_lm_head/bias") name = name.split("/") # print(original_name, name) # adam_v and adam_m are variables used in AdamWeightDecayOptimizer to calculated m and v # which are not required for using pretrained model if any(n in ["global_step", "temperature"] for n in name): logger.info(f"Skipping {original_name}") continue pointer = model for m_name in name: if re.fullmatch(r"[A-Za-z]+_\d+", m_name): scope_names = re.split(r"_(\d+)", m_name) else: scope_names = [m_name] if scope_names[0] == "kernel" or scope_names[0] == "gamma": pointer = getattr(pointer, "weight") elif scope_names[0] == "output_bias" or scope_names[0] == "beta": pointer = getattr(pointer, "bias") elif scope_names[0] == "output_weights": pointer = getattr(pointer, "weight") elif scope_names[0] == "squad": pointer = getattr(pointer, "classifier") else: pointer = getattr(pointer, scope_names[0]) if len(scope_names) >= 2: num = int(scope_names[1]) pointer = pointer[num] if m_name.endswith("_embeddings"): pointer = getattr(pointer, "weight") elif m_name == "kernel": array = np.transpose(array) try: assert ( pointer.shape == array.shape ), f"Pointer shape {pointer.shape} and array shape {array.shape} mismatched" except AssertionError as e: e.args += (pointer.shape, array.shape) raise print(f"Initialize PyTorch weight {name}", original_name) pointer.data = torch.from_numpy(array) except AttributeError as e: print(f"Skipping {original_name}", name, e) continue return model class ElectraEmbeddings(nn.Module): """Construct the embeddings from word, position and token_type embeddings.""" def __init__(self, config): super().__init__() self.word_embeddings = nn.Embedding(config.vocab_size, config.embedding_size, padding_idx=config.pad_token_id) self.position_embeddings = nn.Embedding(config.max_position_embeddings, config.embedding_size) self.token_type_embeddings = nn.Embedding(config.type_vocab_size, config.embedding_size) # self.LayerNorm is not snake-cased to stick with TensorFlow model variable name and be able to load # any TensorFlow checkpoint file self.LayerNorm = nn.LayerNorm(config.embedding_size, eps=config.layer_norm_eps) self.dropout = nn.Dropout(config.hidden_dropout_prob) # position_ids (1, len position emb) is contiguous in memory and exported when serialized self.register_buffer("position_ids", torch.arange(config.max_position_embeddings).expand((1, -1))) self.position_embedding_type = getattr(config, "position_embedding_type", "absolute") if version.parse(torch.__version__) > version.parse("1.6.0"): self.register_buffer( "token_type_ids", torch.zeros(self.position_ids.size(), dtype=torch.long, device=self.position_ids.device), persistent=False, ) # Copied from transformers.models.bert.modeling_bert.BertEmbeddings.forward def forward( self, input_ids=None, token_type_ids=None, position_ids=None, inputs_embeds=None, past_key_values_length=0 ): if input_ids is not None: input_shape = input_ids.size() else: input_shape = inputs_embeds.size()[:-1] seq_length = input_shape[1] if position_ids is None: position_ids = self.position_ids[:, past_key_values_length : seq_length + past_key_values_length] # Setting the token_type_ids to the registered buffer in constructor where it is all zeros, which usually occurs # when its auto-generated, registered buffer helps users when tracing the model without passing token_type_ids, solves # issue #5664 if token_type_ids is None: if hasattr(self, "token_type_ids"): buffered_token_type_ids = self.token_type_ids[:, :seq_length] buffered_token_type_ids_expanded = buffered_token_type_ids.expand(input_shape[0], seq_length) token_type_ids = buffered_token_type_ids_expanded else: token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=self.position_ids.device) if inputs_embeds is None: inputs_embeds = self.word_embeddings(input_ids) token_type_embeddings = self.token_type_embeddings(token_type_ids) embeddings = inputs_embeds + token_type_embeddings if self.position_embedding_type == "absolute": position_embeddings = self.position_embeddings(position_ids) embeddings += position_embeddings embeddings = self.LayerNorm(embeddings) embeddings = self.dropout(embeddings) return embeddings # Copied from transformers.models.bert.modeling_bert.BertSelfAttention with Bert->Electra class ElectraSelfAttention(nn.Module): def __init__(self, config): super().__init__() if config.hidden_size % config.num_attention_heads != 0 and not hasattr(config, "embedding_size"): raise ValueError( f"The hidden size ({config.hidden_size}) is not a multiple of the number of attention " f"heads ({config.num_attention_heads})" ) self.num_attention_heads = config.num_attention_heads self.attention_head_size = int(config.hidden_size / config.num_attention_heads) self.all_head_size = self.num_attention_heads * self.attention_head_size self.query = nn.Linear(config.hidden_size, self.all_head_size) self.key = nn.Linear(config.hidden_size, self.all_head_size) self.value = nn.Linear(config.hidden_size, self.all_head_size) self.dropout = nn.Dropout(config.attention_probs_dropout_prob) self.position_embedding_type = getattr(config, "position_embedding_type", "absolute") if self.position_embedding_type == "relative_key" or self.position_embedding_type == "relative_key_query": self.max_position_embeddings = config.max_position_embeddings self.distance_embedding = nn.Embedding(2 * config.max_position_embeddings - 1, self.attention_head_size) self.is_decoder = config.is_decoder def transpose_for_scores(self, x): new_x_shape = x.size()[:-1] + (self.num_attention_heads, self.attention_head_size) x = x.view(*new_x_shape) return x.permute(0, 2, 1, 3) def forward( self, hidden_states, attention_mask=None, head_mask=None, encoder_hidden_states=None, encoder_attention_mask=None, past_key_value=None, output_attentions=False, ): mixed_query_layer = self.query(hidden_states) # If this is instantiated as a cross-attention module, the keys # and values come from an encoder; the attention mask needs to be # such that the encoder's padding tokens are not attended to. is_cross_attention = encoder_hidden_states is not None if is_cross_attention and past_key_value is not None: # reuse k,v, cross_attentions key_layer = past_key_value[0] value_layer = past_key_value[1] attention_mask = encoder_attention_mask elif is_cross_attention: key_layer = self.transpose_for_scores(self.key(encoder_hidden_states)) value_layer = self.transpose_for_scores(self.value(encoder_hidden_states)) attention_mask = encoder_attention_mask elif past_key_value is not None: key_layer = self.transpose_for_scores(self.key(hidden_states)) value_layer = self.transpose_for_scores(self.value(hidden_states)) key_layer = torch.cat([past_key_value[0], key_layer], dim=2) value_layer = torch.cat([past_key_value[1], value_layer], dim=2) else: key_layer = self.transpose_for_scores(self.key(hidden_states)) value_layer = self.transpose_for_scores(self.value(hidden_states)) query_layer = self.transpose_for_scores(mixed_query_layer) if self.is_decoder: # if cross_attention save Tuple(torch.Tensor, torch.Tensor) of all cross attention key/value_states. # Further calls to cross_attention layer can then reuse all cross-attention # key/value_states (first "if" case) # if uni-directional self-attention (decoder) save Tuple(torch.Tensor, torch.Tensor) of # all previous decoder key/value_states. Further calls to uni-directional self-attention # can concat previous decoder key/value_states to current projected key/value_states (third "elif" case) # if encoder bi-directional self-attention `past_key_value` is always `None` past_key_value = (key_layer, value_layer) # Take the dot product between "query" and "key" to get the raw attention scores. attention_scores = torch.matmul(query_layer, key_layer.transpose(-1, -2)) if self.position_embedding_type == "relative_key" or self.position_embedding_type == "relative_key_query": seq_length = hidden_states.size()[1] position_ids_l = torch.arange(seq_length, dtype=torch.long, device=hidden_states.device).view(-1, 1) position_ids_r = torch.arange(seq_length, dtype=torch.long, device=hidden_states.device).view(1, -1) distance = position_ids_l - position_ids_r positional_embedding = self.distance_embedding(distance + self.max_position_embeddings - 1) positional_embedding = positional_embedding.to(dtype=query_layer.dtype) # fp16 compatibility if self.position_embedding_type == "relative_key": relative_position_scores = torch.einsum("bhld,lrd->bhlr", query_layer, positional_embedding) attention_scores = attention_scores + relative_position_scores elif self.position_embedding_type == "relative_key_query": relative_position_scores_query = torch.einsum("bhld,lrd->bhlr", query_layer, positional_embedding) relative_position_scores_key = torch.einsum("bhrd,lrd->bhlr", key_layer, positional_embedding) attention_scores = attention_scores + relative_position_scores_query + relative_position_scores_key attention_scores = attention_scores / math.sqrt(self.attention_head_size) if attention_mask is not None: # Apply the attention mask is (precomputed for all layers in ElectraModel forward() function) attention_scores = attention_scores + attention_mask # Normalize the attention scores to probabilities. attention_probs = nn.Softmax(dim=-1)(attention_scores) # This is actually dropping out entire tokens to attend to, which might # seem a bit unusual, but is taken from the original Transformer paper. attention_probs = self.dropout(attention_probs) # Mask heads if we want to if head_mask is not None: attention_probs = attention_probs * head_mask context_layer = torch.matmul(attention_probs, value_layer) context_layer = context_layer.permute(0, 2, 1, 3).contiguous() new_context_layer_shape = context_layer.size()[:-2] + (self.all_head_size,) context_layer = context_layer.view(*new_context_layer_shape) outputs = (context_layer, attention_probs) if output_attentions else (context_layer,) if self.is_decoder: outputs = outputs + (past_key_value,) return outputs # Copied from transformers.models.bert.modeling_bert.BertSelfOutput class ElectraSelfOutput(nn.Module): def __init__(self, config): super().__init__() self.dense = nn.Linear(config.hidden_size, config.hidden_size) self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps) self.dropout = nn.Dropout(config.hidden_dropout_prob) def forward(self, hidden_states, input_tensor): hidden_states = self.dense(hidden_states) hidden_states = self.dropout(hidden_states) hidden_states = self.LayerNorm(hidden_states + input_tensor) return hidden_states # Copied from transformers.models.bert.modeling_bert.BertAttention with Bert->Electra class ElectraAttention(nn.Module): def __init__(self, config): super().__init__() self.self = ElectraSelfAttention(config) self.output = ElectraSelfOutput(config) self.pruned_heads = set() def prune_heads(self, heads): if len(heads) == 0: return heads, index = find_pruneable_heads_and_indices( heads, self.self.num_attention_heads, self.self.attention_head_size, self.pruned_heads ) # Prune linear layers self.self.query = prune_linear_layer(self.self.query, index) self.self.key = prune_linear_layer(self.self.key, index) self.self.value = prune_linear_layer(self.self.value, index) self.output.dense = prune_linear_layer(self.output.dense, index, dim=1) # Update hyper params and store pruned heads self.self.num_attention_heads = self.self.num_attention_heads - len(heads) self.self.all_head_size = self.self.attention_head_size * self.self.num_attention_heads self.pruned_heads = self.pruned_heads.union(heads) def forward( self, hidden_states, attention_mask=None, head_mask=None, encoder_hidden_states=None, encoder_attention_mask=None, past_key_value=None, output_attentions=False, ): self_outputs = self.self( hidden_states, attention_mask, head_mask, encoder_hidden_states, encoder_attention_mask, past_key_value, output_attentions, ) attention_output = self.output(self_outputs[0], hidden_states) outputs = (attention_output,) + self_outputs[1:] # add attentions if we output them return outputs # Copied from transformers.models.bert.modeling_bert.BertIntermediate class ElectraIntermediate(nn.Module): def __init__(self, config): super().__init__() self.dense = nn.Linear(config.hidden_size, config.intermediate_size) if isinstance(config.hidden_act, str): self.intermediate_act_fn = ACT2FN[config.hidden_act] else: self.intermediate_act_fn = config.hidden_act def forward(self, hidden_states): hidden_states = self.dense(hidden_states) hidden_states = self.intermediate_act_fn(hidden_states) return hidden_states # Copied from transformers.models.bert.modeling_bert.BertOutput class ElectraOutput(nn.Module): def __init__(self, config): super().__init__() self.dense = nn.Linear(config.intermediate_size, config.hidden_size) self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps) self.dropout = nn.Dropout(config.hidden_dropout_prob) def forward(self, hidden_states, input_tensor): hidden_states = self.dense(hidden_states) hidden_states = self.dropout(hidden_states) hidden_states = self.LayerNorm(hidden_states + input_tensor) return hidden_states # Copied from transformers.models.bert.modeling_bert.BertLayer with Bert->Electra class ElectraLayer(nn.Module): def __init__(self, config): super().__init__() self.chunk_size_feed_forward = config.chunk_size_feed_forward self.seq_len_dim = 1 self.attention = ElectraAttention(config) self.is_decoder = config.is_decoder self.add_cross_attention = config.add_cross_attention if self.add_cross_attention: assert self.is_decoder, f"{self} should be used as a decoder model if cross attention is added" self.crossattention = ElectraAttention(config) self.intermediate = ElectraIntermediate(config) self.output = ElectraOutput(config) def forward( self, hidden_states, attention_mask=None, head_mask=None, encoder_hidden_states=None, encoder_attention_mask=None, past_key_value=None, output_attentions=False, ): # decoder uni-directional self-attention cached key/values tuple is at positions 1,2 self_attn_past_key_value = past_key_value[:2] if past_key_value is not None else None self_attention_outputs = self.attention( hidden_states, attention_mask, head_mask, output_attentions=output_attentions, past_key_value=self_attn_past_key_value, ) attention_output = self_attention_outputs[0] # if decoder, the last output is tuple of self-attn cache if self.is_decoder: outputs = self_attention_outputs[1:-1] present_key_value = self_attention_outputs[-1] else: outputs = self_attention_outputs[1:] # add self attentions if we output attention weights cross_attn_present_key_value = None if self.is_decoder and encoder_hidden_states is not None: assert hasattr( self, "crossattention" ), f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers by setting `config.add_cross_attention=True`" # cross_attn cached key/values tuple is at positions 3,4 of past_key_value tuple cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None cross_attention_outputs = self.crossattention( attention_output, attention_mask, head_mask, encoder_hidden_states, encoder_attention_mask, cross_attn_past_key_value, output_attentions, ) attention_output = cross_attention_outputs[0] outputs = outputs + cross_attention_outputs[1:-1] # add cross attentions if we output attention weights # add cross-attn cache to positions 3,4 of present_key_value tuple cross_attn_present_key_value = cross_attention_outputs[-1] present_key_value = present_key_value + cross_attn_present_key_value layer_output = apply_chunking_to_forward( self.feed_forward_chunk, self.chunk_size_feed_forward, self.seq_len_dim, attention_output ) outputs = (layer_output,) + outputs # if decoder, return the attn key/values as the last output if self.is_decoder: outputs = outputs + (present_key_value,) return outputs def feed_forward_chunk(self, attention_output): intermediate_output = self.intermediate(attention_output) layer_output = self.output(intermediate_output, attention_output) return layer_output # Copied from transformers.models.bert.modeling_bert.BertEncoder with Bert->Electra class ElectraEncoder(nn.Module): def __init__(self, config): super().__init__() self.config = config self.layer = nn.ModuleList([ElectraLayer(config) for _ in range(config.num_hidden_layers)]) def forward( self, hidden_states, attention_mask=None, head_mask=None, encoder_hidden_states=None, encoder_attention_mask=None, past_key_values=None, use_cache=None, output_attentions=False, output_hidden_states=False, return_dict=True, ): all_hidden_states = () if output_hidden_states else None all_self_attentions = () if output_attentions else None all_cross_attentions = () if output_attentions and self.config.add_cross_attention else None next_decoder_cache = () if use_cache else None for i, layer_module in enumerate(self.layer): if output_hidden_states: all_hidden_states = all_hidden_states + (hidden_states,) layer_head_mask = head_mask[i] if head_mask is not None else None past_key_value = past_key_values[i] if past_key_values is not None else None if getattr(self.config, "gradient_checkpointing", False) and self.training: if use_cache: logger.warning( "`use_cache=True` is incompatible with `config.gradient_checkpointing=True`. Setting " "`use_cache=False`..." ) use_cache = False def create_custom_forward(module): def custom_forward(*inputs): return module(*inputs, past_key_value, output_attentions) return custom_forward layer_outputs = torch.utils.checkpoint.checkpoint( create_custom_forward(layer_module), hidden_states, attention_mask, layer_head_mask, encoder_hidden_states, encoder_attention_mask, ) else: layer_outputs = layer_module( hidden_states, attention_mask, layer_head_mask, encoder_hidden_states, encoder_attention_mask, past_key_value, output_attentions, ) hidden_states = layer_outputs[0] if use_cache: next_decoder_cache += (layer_outputs[-1],) if output_attentions: all_self_attentions = all_self_attentions + (layer_outputs[1],) if self.config.add_cross_attention: all_cross_attentions = all_cross_attentions + (layer_outputs[2],) if output_hidden_states: all_hidden_states = all_hidden_states + (hidden_states,) if not return_dict: return tuple( v for v in [ hidden_states, next_decoder_cache, all_hidden_states, all_self_attentions, all_cross_attentions, ] if v is not None ) return BaseModelOutputWithPastAndCrossAttentions( last_hidden_state=hidden_states, past_key_values=next_decoder_cache, hidden_states=all_hidden_states, attentions=all_self_attentions, cross_attentions=all_cross_attentions, ) class ElectraDiscriminatorPredictions(nn.Module): """Prediction module for the discriminator, made up of two dense layers.""" def __init__(self, config): super().__init__() self.dense = nn.Linear(config.hidden_size, config.hidden_size) self.dense_prediction = nn.Linear(config.hidden_size, 1) self.config = config def forward(self, discriminator_hidden_states): hidden_states = self.dense(discriminator_hidden_states) hidden_states = get_activation(self.config.hidden_act)(hidden_states) logits = self.dense_prediction(hidden_states).squeeze(-1) return logits class ElectraGeneratorPredictions(nn.Module): """Prediction module for the generator, made up of two dense layers.""" def __init__(self, config): super().__init__() self.LayerNorm = nn.LayerNorm(config.embedding_size) self.dense = nn.Linear(config.hidden_size, config.embedding_size) def forward(self, generator_hidden_states): hidden_states = self.dense(generator_hidden_states) hidden_states = get_activation("gelu")(hidden_states) hidden_states = self.LayerNorm(hidden_states) return hidden_states class ElectraPreTrainedModel(PreTrainedModel): """ An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained models. """ config_class = ElectraConfig load_tf_weights = load_tf_weights_in_electra base_model_prefix = "electra" _keys_to_ignore_on_load_missing = [r"position_ids"] _keys_to_ignore_on_load_unexpected = [r"electra\.embeddings_project\.weight", r"electra\.embeddings_project\.bias"] # Copied from transformers.models.bert.modeling_bert.BertPreTrainedModel._init_weights def _init_weights(self, module): """Initialize the weights""" if isinstance(module, nn.Linear): # Slightly different from the TF version which uses truncated_normal for initialization # cf https://github.com/pytorch/pytorch/pull/5617 module.weight.data.normal_(mean=0.0, std=self.config.initializer_range) if module.bias is not None: module.bias.data.zero_() elif isinstance(module, nn.Embedding): module.weight.data.normal_(mean=0.0, std=self.config.initializer_range) if module.padding_idx is not None: module.weight.data[module.padding_idx].zero_() elif isinstance(module, nn.LayerNorm): module.bias.data.zero_() module.weight.data.fill_(1.0) @dataclass class ElectraForPreTrainingOutput(ModelOutput): """ Output type of :class:`~transformers.ElectraForPreTraining`. Args: loss (`optional`, returned when ``labels`` is provided, ``torch.FloatTensor`` of shape :obj:`(1,)`): Total loss of the ELECTRA objective. logits (:obj:`torch.FloatTensor` of shape :obj:`(batch_size, sequence_length)`): Prediction scores of the head (scores for each token before SoftMax). hidden_states (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_hidden_states=True`` is passed or when ``config.output_hidden_states=True``): Tuple of :obj:`torch.FloatTensor` (one for the output of the embeddings + one for the output of each layer) of shape :obj:`(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer plus the initial embedding outputs. attentions (:obj:`tuple(torch.FloatTensor)`, `optional`, returned when ``output_attentions=True`` is passed or when ``config.output_attentions=True``): Tuple of :obj:`torch.FloatTensor` (one for each layer) of shape :obj:`(batch_size, num_heads, sequence_length, sequence_length)`. Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads. """ loss: Optional[torch.FloatTensor] = None logits: torch.FloatTensor = None hidden_states: Optional[Tuple[torch.FloatTensor]] = None attentions: Optional[Tuple[torch.FloatTensor]] = None ELECTRA_START_DOCSTRING = r""" This model inherits from :class:`~transformers.PreTrainedModel`. Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.) This model is also a PyTorch `torch.nn.Module `__ subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior. Parameters: config (:class:`~transformers.ElectraConfig`): Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the :meth:`~transformers.PreTrainedModel.from_pretrained` method to load the model weights. """ ELECTRA_INPUTS_DOCSTRING = r""" Args: input_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`): Indices of input sequence tokens in the vocabulary. Indices can be obtained using :class:`~transformers.ElectraTokenizer`. See :meth:`transformers.PreTrainedTokenizer.encode` and :meth:`transformers.PreTrainedTokenizer.__call__` for details. `What are input IDs? <../glossary.html#input-ids>`__ attention_mask (:obj:`torch.FloatTensor` of shape :obj:`({0})`, `optional`): Mask to avoid performing attention on padding token indices. Mask values selected in ``[0, 1]``: - 1 for tokens that are **not masked**, - 0 for tokens that are **masked**. `What are attention masks? <../glossary.html#attention-mask>`__ token_type_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`): Segment token indices to indicate first and second portions of the inputs. Indices are selected in ``[0, 1]``: - 0 corresponds to a `sentence A` token, - 1 corresponds to a `sentence B` token. `What are token type IDs? <../glossary.html#token-type-ids>`_ position_ids (:obj:`torch.LongTensor` of shape :obj:`({0})`, `optional`): Indices of positions of each input sequence tokens in the position embeddings. Selected in the range ``[0, config.max_position_embeddings - 1]``. `What are position IDs? <../glossary.html#position-ids>`_ head_mask (:obj:`torch.FloatTensor` of shape :obj:`(num_heads,)` or :obj:`(num_layers, num_heads)`, `optional`): Mask to nullify selected heads of the self-attention modules. Mask values selected in ``[0, 1]``: - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. inputs_embeds (:obj:`torch.FloatTensor` of shape :obj:`({0}, hidden_size)`, `optional`): Optionally, instead of passing :obj:`input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert :obj:`input_ids` indices into associated vectors than the model's internal embedding lookup matrix. encoder_hidden_states (:obj:`torch.FloatTensor` of shape :obj:`({0}, hidden_size)`, `optional`): Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention if the model is configured as a decoder. encoder_attention_mask (:obj:`torch.FloatTensor` of shape :obj:`({0})`, `optional`): Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used in the cross-attention if the model is configured as a decoder. Mask values selected in ``[0, 1]``: - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. output_attentions (:obj:`bool`, `optional`): Whether or not to return the attentions tensors of all attention layers. See ``attentions`` under returned tensors for more detail. output_hidden_states (:obj:`bool`, `optional`): Whether or not to return the hidden states of all layers. See ``hidden_states`` under returned tensors for more detail. return_dict (:obj:`bool`, `optional`): Whether or not to return a :class:`~transformers.file_utils.ModelOutput` instead of a plain tuple. """ @add_start_docstrings( "The bare Electra Model transformer outputting raw hidden-states without any specific head on top. Identical to " "the BERT model except that it uses an additional linear layer between the embedding layer and the encoder if the " "hidden size and embedding size are different." "" "Both the generator and discriminator checkpoints may be loaded into this model.", ELECTRA_START_DOCSTRING, ) class ElectraModel(ElectraPreTrainedModel): def __init__(self, config): super().__init__(config) self.embeddings = ElectraEmbeddings(config) if config.embedding_size != config.hidden_size: self.embeddings_project = nn.Linear(config.embedding_size, config.hidden_size) self.encoder = ElectraEncoder(config) self.config = config self.init_weights() def get_input_embeddings(self): return self.embeddings.word_embeddings def set_input_embeddings(self, value): self.embeddings.word_embeddings = value def _prune_heads(self, heads_to_prune): """ Prunes heads of the model. heads_to_prune: dict of {layer_num: list of heads to prune in this layer} See base class PreTrainedModel """ for layer, heads in heads_to_prune.items(): self.encoder.layer[layer].attention.prune_heads(heads) @add_start_docstrings_to_model_forward(ELECTRA_INPUTS_DOCSTRING.format("batch_size, sequence_length")) @add_code_sample_docstrings( tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint=_CHECKPOINT_FOR_DOC, output_type=BaseModelOutputWithCrossAttentions, config_class=_CONFIG_FOR_DOC, ) def forward( self, input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, output_attentions=None, output_hidden_states=None, return_dict=None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time") elif input_ids is not None: input_shape = input_ids.size() batch_size, seq_length = input_shape elif inputs_embeds is not None: input_shape = inputs_embeds.size()[:-1] else: raise ValueError("You have to specify either input_ids or inputs_embeds") device = input_ids.device if input_ids is not None else inputs_embeds.device if attention_mask is None: attention_mask = torch.ones(input_shape, device=device) if token_type_ids is None: if hasattr(self.embeddings, "token_type_ids"): buffered_token_type_ids = self.embeddings.token_type_ids[:, :seq_length] buffered_token_type_ids_expanded = buffered_token_type_ids.expand(batch_size, seq_length) token_type_ids = buffered_token_type_ids_expanded else: token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=device) extended_attention_mask = self.get_extended_attention_mask(attention_mask, input_shape, device) head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers) hidden_states = self.embeddings( input_ids=input_ids, position_ids=position_ids, token_type_ids=token_type_ids, inputs_embeds=inputs_embeds ) if hasattr(self, "embeddings_project"): hidden_states = self.embeddings_project(hidden_states) hidden_states = self.encoder( hidden_states, attention_mask=extended_attention_mask, head_mask=head_mask, output_attentions=output_attentions, output_hidden_states=output_hidden_states, return_dict=return_dict, ) return hidden_states class ElectraClassificationHead(nn.Module): """Head for sentence-level classification tasks.""" def __init__(self, config): super().__init__() self.dense = nn.Linear(config.hidden_size, config.hidden_size) self.dropout = nn.Dropout(config.hidden_dropout_prob) self.out_proj = nn.Linear(config.hidden_size, config.num_labels) def forward(self, features, **kwargs): x = features[:, 0, :] # take token (equiv. to [CLS]) x = self.dropout(x) x = self.dense(x) x = get_activation("gelu")(x) # although BERT uses tanh here, it seems Electra authors used gelu here x = self.dropout(x) x = self.out_proj(x) return x @add_start_docstrings( """ ELECTRA Model transformer with a sequence classification/regression head on top (a linear layer on top of the pooled output) e.g. for GLUE tasks. """, ELECTRA_START_DOCSTRING, ) class ElectraForSequenceClassification(ElectraPreTrainedModel): def __init__(self, config): super().__init__(config) self.num_labels = config.num_labels self.config = config self.electra = ElectraModel(config) self.classifier = ElectraClassificationHead(config) self.init_weights() @add_start_docstrings_to_model_forward(ELECTRA_INPUTS_DOCSTRING.format("batch_size, sequence_length")) @add_code_sample_docstrings( tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint=_CHECKPOINT_FOR_DOC, output_type=SequenceClassifierOutput, config_class=_CONFIG_FOR_DOC, ) def forward( self, input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, labels=None, output_attentions=None, output_hidden_states=None, return_dict=None, ): r""" labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`): Labels for computing the sequence classification/regression loss. Indices should be in :obj:`[0, ..., config.num_labels - 1]`. If :obj:`config.num_labels == 1` a regression loss is computed (Mean-Square loss), If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy). """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict discriminator_hidden_states = self.electra( input_ids, attention_mask, token_type_ids, position_ids, head_mask, inputs_embeds, output_attentions, output_hidden_states, return_dict, ) sequence_output = discriminator_hidden_states[0] logits = self.classifier(sequence_output) loss = None if labels is not None: if self.config.problem_type is None: if self.num_labels == 1: self.config.problem_type = "regression" elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int): self.config.problem_type = "single_label_classification" else: self.config.problem_type = "multi_label_classification" if self.config.problem_type == "regression": loss_fct = MSELoss() if self.num_labels == 1: loss = loss_fct(logits.squeeze(), labels.squeeze()) else: loss = loss_fct(logits, labels) elif self.config.problem_type == "single_label_classification": loss_fct = CrossEntropyLoss() loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1)) elif self.config.problem_type == "multi_label_classification": loss_fct = BCEWithLogitsLoss() loss = loss_fct(logits, labels) if not return_dict: output = (logits,) + discriminator_hidden_states[1:] return ((loss,) + output) if loss is not None else output return SequenceClassifierOutput( loss=loss, logits=logits, hidden_states=discriminator_hidden_states.hidden_states, attentions=discriminator_hidden_states.attentions, ) @add_start_docstrings( """ Electra model with a binary classification head on top as used during pretraining for identifying generated tokens. It is recommended to load the discriminator checkpoint into that model. """, ELECTRA_START_DOCSTRING, ) class ElectraForPreTraining(ElectraPreTrainedModel): def __init__(self, config): super().__init__(config) self.electra = ElectraModel(config) self.discriminator_predictions = ElectraDiscriminatorPredictions(config) self.init_weights() @add_start_docstrings_to_model_forward(ELECTRA_INPUTS_DOCSTRING.format("batch_size, sequence_length")) @replace_return_docstrings(output_type=ElectraForPreTrainingOutput, config_class=_CONFIG_FOR_DOC) def forward( self, input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, labels=None, output_attentions=None, output_hidden_states=None, return_dict=None, ): r""" labels (``torch.LongTensor`` of shape ``(batch_size, sequence_length)``, `optional`): Labels for computing the ELECTRA loss. Input should be a sequence of tokens (see :obj:`input_ids` docstring) Indices should be in ``[0, 1]``: - 0 indicates the token is an original token, - 1 indicates the token was replaced. Returns: Examples:: >>> from transformers import ElectraTokenizer, ElectraForPreTraining >>> import torch >>> tokenizer = ElectraTokenizer.from_pretrained('google/electra-small-discriminator') >>> model = ElectraForPreTraining.from_pretrained('google/electra-small-discriminator') >>> input_ids = torch.tensor(tokenizer.encode("Hello, my dog is cute", add_special_tokens=True)).unsqueeze(0) # Batch size 1 >>> logits = model(input_ids).logits """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict discriminator_hidden_states = self.electra( input_ids, attention_mask, token_type_ids, position_ids, head_mask, inputs_embeds, output_attentions, output_hidden_states, return_dict, ) discriminator_sequence_output = discriminator_hidden_states[0] logits = self.discriminator_predictions(discriminator_sequence_output) loss = None if labels is not None: loss_fct = nn.BCEWithLogitsLoss() if attention_mask is not None: active_loss = attention_mask.view(-1, discriminator_sequence_output.shape[1]) == 1 active_logits = logits.view(-1, discriminator_sequence_output.shape[1])[active_loss] active_labels = labels[active_loss] loss = loss_fct(active_logits, active_labels.float()) else: loss = loss_fct(logits.view(-1, discriminator_sequence_output.shape[1]), labels.float()) if not return_dict: output = (logits,) + discriminator_hidden_states[1:] return ((loss,) + output) if loss is not None else output return ElectraForPreTrainingOutput( loss=loss, logits=logits, hidden_states=discriminator_hidden_states.hidden_states, attentions=discriminator_hidden_states.attentions, ) @add_start_docstrings( """ Electra model with a language modeling head on top. Even though both the discriminator and generator may be loaded into this model, the generator is the only model of the two to have been trained for the masked language modeling task. """, ELECTRA_START_DOCSTRING, ) class ElectraForMaskedLM(ElectraPreTrainedModel): def __init__(self, config): super().__init__(config) self.electra = ElectraModel(config) self.generator_predictions = ElectraGeneratorPredictions(config) self.generator_lm_head = nn.Linear(config.embedding_size, config.vocab_size) self.init_weights() def get_output_embeddings(self): return self.generator_lm_head def set_output_embeddings(self, word_embeddings): self.generator_lm_head = word_embeddings @add_start_docstrings_to_model_forward(ELECTRA_INPUTS_DOCSTRING.format("batch_size, sequence_length")) @add_code_sample_docstrings( tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint=_CHECKPOINT_FOR_DOC, output_type=MaskedLMOutput, config_class=_CONFIG_FOR_DOC, ) def forward( self, input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, labels=None, output_attentions=None, output_hidden_states=None, return_dict=None, ): r""" labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`): Labels for computing the masked language modeling loss. Indices should be in ``[-100, 0, ..., config.vocab_size]`` (see ``input_ids`` docstring) Tokens with indices set to ``-100`` are ignored (masked), the loss is only computed for the tokens with labels in ``[0, ..., config.vocab_size]`` """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict generator_hidden_states = self.electra( input_ids, attention_mask, token_type_ids, position_ids, head_mask, inputs_embeds, output_attentions, output_hidden_states, return_dict, ) generator_sequence_output = generator_hidden_states[0] prediction_scores = self.generator_predictions(generator_sequence_output) prediction_scores = self.generator_lm_head(prediction_scores) loss = None # Masked language modeling softmax layer if labels is not None: loss_fct = nn.CrossEntropyLoss() # -100 index = padding token loss = loss_fct(prediction_scores.view(-1, self.config.vocab_size), labels.view(-1)) if not return_dict: output = (prediction_scores,) + generator_hidden_states[1:] return ((loss,) + output) if loss is not None else output return MaskedLMOutput( loss=loss, logits=prediction_scores, hidden_states=generator_hidden_states.hidden_states, attentions=generator_hidden_states.attentions, ) @add_start_docstrings( """ Electra model with a token classification head on top. Both the discriminator and generator may be loaded into this model. """, ELECTRA_START_DOCSTRING, ) class ElectraForTokenClassification(ElectraPreTrainedModel): def __init__(self, config): super().__init__(config) self.electra = ElectraModel(config) self.dropout = nn.Dropout(config.hidden_dropout_prob) self.classifier = nn.Linear(config.hidden_size, config.num_labels) self.init_weights() @add_start_docstrings_to_model_forward(ELECTRA_INPUTS_DOCSTRING.format("batch_size, sequence_length")) @add_code_sample_docstrings( tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint=_CHECKPOINT_FOR_DOC, output_type=TokenClassifierOutput, config_class=_CONFIG_FOR_DOC, ) def forward( self, input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, labels=None, output_attentions=None, output_hidden_states=None, return_dict=None, ): r""" labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size, sequence_length)`, `optional`): Labels for computing the token classification loss. Indices should be in ``[0, ..., config.num_labels - 1]``. """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict discriminator_hidden_states = self.electra( input_ids, attention_mask, token_type_ids, position_ids, head_mask, inputs_embeds, output_attentions, output_hidden_states, return_dict, ) discriminator_sequence_output = discriminator_hidden_states[0] discriminator_sequence_output = self.dropout(discriminator_sequence_output) logits = self.classifier(discriminator_sequence_output) loss = None if labels is not None: loss_fct = nn.CrossEntropyLoss() # Only keep active parts of the loss if attention_mask is not None: active_loss = attention_mask.view(-1) == 1 active_logits = logits.view(-1, self.config.num_labels)[active_loss] active_labels = labels.view(-1)[active_loss] loss = loss_fct(active_logits, active_labels) else: loss = loss_fct(logits.view(-1, self.config.num_labels), labels.view(-1)) if not return_dict: output = (logits,) + discriminator_hidden_states[1:] return ((loss,) + output) if loss is not None else output return TokenClassifierOutput( loss=loss, logits=logits, hidden_states=discriminator_hidden_states.hidden_states, attentions=discriminator_hidden_states.attentions, ) @add_start_docstrings( """ ELECTRA Model with a span classification head on top for extractive question-answering tasks like SQuAD (a linear layers on top of the hidden-states output to compute `span start logits` and `span end logits`). """, ELECTRA_START_DOCSTRING, ) class ElectraForQuestionAnswering(ElectraPreTrainedModel): config_class = ElectraConfig base_model_prefix = "electra" def __init__(self, config): super().__init__(config) self.num_labels = config.num_labels self.electra = ElectraModel(config) self.qa_outputs = nn.Linear(config.hidden_size, config.num_labels) self.init_weights() @add_start_docstrings_to_model_forward(ELECTRA_INPUTS_DOCSTRING.format("batch_size, sequence_length")) @add_code_sample_docstrings( tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint=_CHECKPOINT_FOR_DOC, output_type=QuestionAnsweringModelOutput, config_class=_CONFIG_FOR_DOC, ) def forward( self, input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, start_positions=None, end_positions=None, output_attentions=None, output_hidden_states=None, return_dict=None, ): r""" start_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`): Labels for position (index) of the start of the labelled span for computing the token classification loss. Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the sequence are not taken into account for computing the loss. end_positions (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`): Labels for position (index) of the end of the labelled span for computing the token classification loss. Positions are clamped to the length of the sequence (:obj:`sequence_length`). Position outside of the sequence are not taken into account for computing the loss. """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict discriminator_hidden_states = self.electra( input_ids, attention_mask=attention_mask, token_type_ids=token_type_ids, position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, output_attentions=output_attentions, output_hidden_states=output_hidden_states, ) sequence_output = discriminator_hidden_states[0] logits = self.qa_outputs(sequence_output) start_logits, end_logits = logits.split(1, dim=-1) start_logits = start_logits.squeeze(-1).contiguous() end_logits = end_logits.squeeze(-1).contiguous() total_loss = None if start_positions is not None and end_positions is not None: # If we are on multi-GPU, split add a dimension if len(start_positions.size()) > 1: start_positions = start_positions.squeeze(-1) if len(end_positions.size()) > 1: end_positions = end_positions.squeeze(-1) # sometimes the start/end positions are outside our model inputs, we ignore these terms ignored_index = start_logits.size(1) start_positions = start_positions.clamp(0, ignored_index) end_positions = end_positions.clamp(0, ignored_index) loss_fct = CrossEntropyLoss(ignore_index=ignored_index) start_loss = loss_fct(start_logits, start_positions) end_loss = loss_fct(end_logits, end_positions) total_loss = (start_loss + end_loss) / 2 if not return_dict: output = ( start_logits, end_logits, ) + discriminator_hidden_states[1:] return ((total_loss,) + output) if total_loss is not None else output return QuestionAnsweringModelOutput( loss=total_loss, start_logits=start_logits, end_logits=end_logits, hidden_states=discriminator_hidden_states.hidden_states, attentions=discriminator_hidden_states.attentions, ) @add_start_docstrings( """ ELECTRA Model with a multiple choice classification head on top (a linear layer on top of the pooled output and a softmax) e.g. for RocStories/SWAG tasks. """, ELECTRA_START_DOCSTRING, ) class ElectraForMultipleChoice(ElectraPreTrainedModel): def __init__(self, config): super().__init__(config) self.electra = ElectraModel(config) self.sequence_summary = SequenceSummary(config) self.classifier = nn.Linear(config.hidden_size, 1) self.init_weights() @add_start_docstrings_to_model_forward(ELECTRA_INPUTS_DOCSTRING.format("batch_size, num_choices, sequence_length")) @add_code_sample_docstrings( tokenizer_class=_TOKENIZER_FOR_DOC, checkpoint=_CHECKPOINT_FOR_DOC, output_type=MultipleChoiceModelOutput, config_class=_CONFIG_FOR_DOC, ) def forward( self, input_ids=None, attention_mask=None, token_type_ids=None, position_ids=None, head_mask=None, inputs_embeds=None, labels=None, output_attentions=None, output_hidden_states=None, return_dict=None, ): r""" labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`): Labels for computing the multiple choice classification loss. Indices should be in ``[0, ..., num_choices-1]`` where :obj:`num_choices` is the size of the second dimension of the input tensors. (See :obj:`input_ids` above) """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict num_choices = input_ids.shape[1] if input_ids is not None else inputs_embeds.shape[1] input_ids = input_ids.view(-1, input_ids.size(-1)) if input_ids is not None else None attention_mask = attention_mask.view(-1, attention_mask.size(-1)) if attention_mask is not None else None token_type_ids = token_type_ids.view(-1, token_type_ids.size(-1)) if token_type_ids is not None else None position_ids = position_ids.view(-1, position_ids.size(-1)) if position_ids is not None else None inputs_embeds = ( inputs_embeds.view(-1, inputs_embeds.size(-2), inputs_embeds.size(-1)) if inputs_embeds is not None else None ) discriminator_hidden_states = self.electra( input_ids, attention_mask=attention_mask, token_type_ids=token_type_ids, position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, output_attentions=output_attentions, output_hidden_states=output_hidden_states, return_dict=return_dict, ) sequence_output = discriminator_hidden_states[0] pooled_output = self.sequence_summary(sequence_output) logits = self.classifier(pooled_output) reshaped_logits = logits.view(-1, num_choices) loss = None if labels is not None: loss_fct = CrossEntropyLoss() loss = loss_fct(reshaped_logits, labels) if not return_dict: output = (reshaped_logits,) + discriminator_hidden_states[1:] return ((loss,) + output) if loss is not None else output return MultipleChoiceModelOutput( loss=loss, logits=reshaped_logits, hidden_states=discriminator_hidden_states.hidden_states, attentions=discriminator_hidden_states.attentions, )