# coding=utf-8 # Copyright 2022 The Salesforce Team Authors and The HuggingFace Team. All rights reserved. # # Licensed under the BSD-3-clause license (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # https://opensource.org/licenses/BSD-3-Clause # # 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. import math from typing import Optional, Union import torch from torch import Tensor, device, nn from torch.nn import CrossEntropyLoss from ...activations import ACT2FN from ...cache_utils import Cache, DynamicCache, EncoderDecoderCache from ...generation import GenerationMixin from ...modeling_layers import GradientCheckpointingLayer from ...modeling_outputs import ( BaseModelOutputWithPastAndCrossAttentions, BaseModelOutputWithPoolingAndCrossAttentions, CausalLMOutputWithCrossAttentions, ) from ...modeling_utils import PreTrainedModel from ...pytorch_utils import apply_chunking_to_forward, find_pruneable_heads_and_indices, prune_linear_layer from ...utils import logging from ...utils.deprecation import deprecate_kwarg from .configuration_blip import BlipTextConfig logger = logging.get_logger(__name__) # Adapted from https://github.com/salesforce/BLIP/blob/main/models/med.py#L52 class BlipTextEmbeddings(nn.Module): """Construct the embeddings from word and position embeddings.""" def __init__(self, config): super().__init__() self.word_embeddings = nn.Embedding(config.vocab_size, config.hidden_size, padding_idx=config.pad_token_id) self.position_embeddings = nn.Embedding(config.max_position_embeddings, config.hidden_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.hidden_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)), persistent=False ) self.position_embedding_type = getattr(config, "position_embedding_type", "absolute") self.config = config def forward( self, input_ids: Optional[torch.LongTensor] = None, position_ids: Optional[torch.LongTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, past_key_values_length: int = 0, ) -> torch.Tensor: 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] if inputs_embeds is None: inputs_embeds = self.word_embeddings(input_ids) embeddings = inputs_embeds 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 # Adapted from https://github.com/salesforce/BLIP/blob/main/models/med.py#L97 class BlipTextSelfAttention(nn.Module): def __init__(self, config, is_cross_attention, layer_idx=None): super().__init__() self.config = config if config.hidden_size % config.num_attention_heads != 0 and not hasattr(config, "embedding_size"): raise ValueError( "The hidden size (%d) is not a multiple of the number of attention heads (%d)" % (config.hidden_size, 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.layer_idx = layer_idx self.query = nn.Linear(config.hidden_size, self.all_head_size) if is_cross_attention: self.key = nn.Linear(config.encoder_hidden_size, self.all_head_size) self.value = nn.Linear(config.encoder_hidden_size, self.all_head_size) else: 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) def save_attn_gradients(self, attn_gradients): self.attn_gradients = attn_gradients def get_attn_gradients(self): return self.attn_gradients def save_attention_map(self, attention_map): self.attention_map = attention_map def get_attention_map(self): return self.attention_map @deprecate_kwarg("past_key_value", new_name="past_key_values", version="4.58") def forward( self, hidden_states: torch.Tensor, attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_values: Optional[Cache] = None, output_attentions: Optional[bool] = False, cache_position: Optional[torch.Tensor] = None, ) -> tuple[torch.Tensor]: batch_size, seq_length, _ = hidden_states.shape query_layer = ( self.query(hidden_states) .view(batch_size, -1, self.num_attention_heads, self.attention_head_size) .transpose(1, 2) ) # 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 attention_mask = encoder_attention_mask if is_cross_attention else attention_mask is_updated = False if past_key_values is not None: if isinstance(past_key_values, EncoderDecoderCache): is_updated = past_key_values.is_updated.get(self.layer_idx) if is_cross_attention: # after the first generated id, we can subsequently re-use all key/value_layer from cache curr_past_key_value = past_key_values.cross_attention_cache else: curr_past_key_value = past_key_values.self_attention_cache else: curr_past_key_value = past_key_values current_states = encoder_hidden_states if is_cross_attention else hidden_states if is_cross_attention and past_key_values is not None and is_updated: # reuse k,v, cross_attentions key_layer = curr_past_key_value.layers[self.layer_idx].keys value_layer = curr_past_key_value.layers[self.layer_idx].values else: key_layer = ( self.key(current_states) .view(batch_size, -1, self.num_attention_heads, self.attention_head_size) .transpose(1, 2) ) value_layer = ( self.value(current_states) .view(batch_size, -1, self.num_attention_heads, self.attention_head_size) .transpose(1, 2) ) if past_key_values is not None: # save all key/value_layer to cache to be re-used for fast auto-regressive generation cache_position = cache_position if not is_cross_attention else None key_layer, value_layer = curr_past_key_value.update( key_layer, value_layer, self.layer_idx, {"cache_position": cache_position} ) # set flag that curr layer for cross-attn is already updated so we can re-use in subsequent calls if is_cross_attention and isinstance(past_key_values, EncoderDecoderCache): past_key_values.is_updated[self.layer_idx] = True # 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 BlipTextModel forward() function) attention_scores = attention_scores + attention_mask.to(attention_scores.device) # 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_dropped = self.dropout(attention_probs) # Mask heads if we want to if head_mask is not None: attention_probs_dropped = attention_probs_dropped * head_mask context_layer = torch.matmul(attention_probs_dropped, 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) return context_layer, attention_probs # Copied from transformers.models.bert.modeling_bert.BertSelfOutput with Bert -> BlipText class BlipTextSelfOutput(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: torch.Tensor, input_tensor: torch.Tensor) -> torch.Tensor: hidden_states = self.dense(hidden_states) hidden_states = self.dropout(hidden_states) hidden_states = self.LayerNorm(hidden_states + input_tensor) return hidden_states # Adapted from https://github.com/salesforce/BLIP/blob/main/models/med.py#242 class BlipTextAttention(nn.Module): def __init__(self, config, is_cross_attention=False, layer_idx=None): super().__init__() self.self = BlipTextSelfAttention(config, is_cross_attention, layer_idx=layer_idx) self.output = BlipTextSelfOutput(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) @deprecate_kwarg("past_key_value", new_name="past_key_values", version="4.58") def forward( self, hidden_states: torch.Tensor, attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, encoder_hidden_states: Optional[torch.FloatTensor] = None, past_key_values: Optional[Cache] = None, output_attentions: Optional[bool] = False, cache_position: Optional[torch.Tensor] = None, ) -> tuple[torch.Tensor]: self_outputs = self.self( hidden_states, attention_mask=attention_mask, head_mask=head_mask, encoder_hidden_states=encoder_hidden_states, past_key_values=past_key_values, output_attentions=output_attentions, cache_position=cache_position, ) 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 with Bert -> BlipText class BlipTextIntermediate(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: torch.Tensor) -> torch.Tensor: 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 with Bert -> BlipText class BlipTextOutput(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: torch.Tensor, input_tensor: torch.Tensor) -> torch.Tensor: hidden_states = self.dense(hidden_states) hidden_states = self.dropout(hidden_states) hidden_states = self.LayerNorm(hidden_states + input_tensor) return hidden_states class BlipTextLayer(GradientCheckpointingLayer): def __init__(self, config, layer_num): super().__init__() self.config = config self.chunk_size_feed_forward = config.chunk_size_feed_forward self.seq_len_dim = 1 self.attention = BlipTextAttention(config, layer_idx=layer_num) self.layer_num = layer_num if self.config.is_decoder: self.crossattention = BlipTextAttention( config, is_cross_attention=self.config.is_decoder, layer_idx=layer_num ) self.intermediate = BlipTextIntermediate(config) self.output = BlipTextOutput(config) @deprecate_kwarg("past_key_value", new_name="past_key_values", version="4.58") def forward( self, hidden_states: torch.Tensor, attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_values: Optional[Cache] = None, output_attentions: Optional[bool] = False, cache_position: Optional[torch.Tensor] = None, ) -> tuple[torch.Tensor]: self_attention_outputs = self.attention( hidden_states, attention_mask=attention_mask, head_mask=head_mask, output_attentions=output_attentions, past_key_values=past_key_values, cache_position=cache_position, ) attention_output = self_attention_outputs[0] outputs = self_attention_outputs[1:] if encoder_hidden_states is not None: cross_attention_outputs = self.crossattention( attention_output, attention_mask=encoder_attention_mask, head_mask=head_mask, encoder_hidden_states=encoder_hidden_states, past_key_values=past_key_values, output_attentions=output_attentions, cache_position=cache_position, ) attention_output = cross_attention_outputs[0] outputs = outputs + cross_attention_outputs[1:] # add cross attentions if we output attention weights layer_output = apply_chunking_to_forward( self.feed_forward_chunk, self.chunk_size_feed_forward, self.seq_len_dim, attention_output ) return (layer_output,) + 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 # Adapted from https://github.com/salesforce/BLIP/blob/main/models/med.py#L386 class BlipTextEncoder(nn.Module): def __init__(self, config): super().__init__() self.config = config self.layer = nn.ModuleList([BlipTextLayer(config, i) for i in range(config.num_hidden_layers)]) self.gradient_checkpointing = False def forward( self, hidden_states: torch.Tensor, attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_values: Optional[Cache] = None, use_cache: Optional[bool] = None, output_attentions: Optional[bool] = False, output_hidden_states: Optional[bool] = False, return_dict: Optional[bool] = True, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPastAndCrossAttentions]: if self.gradient_checkpointing and self.training: if use_cache: logger.warning( "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..." ) use_cache = False if use_cache: if isinstance(past_key_values, tuple): logger.warning_once( "Passing a tuple of `past_key_values` is deprecated and will be removed in Transformers v4.58.0. " "You should pass an instance of `EncoderDecoderCache` instead, e.g. " "`past_key_values=EncoderDecoderCache.from_legacy_cache(past_key_values)`." ) past_key_values = EncoderDecoderCache.from_legacy_cache(past_key_values) # The model acts as encoder decoder but is not an encoder decoder. So we cast all cache objects to # `EncoderDecoderCache` type assuming that the incoming cache is from `self_attention` elif isinstance(past_key_values, DynamicCache): past_key_values = EncoderDecoderCache(past_key_values, DynamicCache(config=self.config)) elif past_key_values is None: past_key_values = EncoderDecoderCache( DynamicCache(config=self.config), DynamicCache(config=self.config) ) all_hidden_states = () if output_hidden_states else None all_self_attentions = () if output_attentions else None all_cross_attentions = () if output_attentions and encoder_hidden_states is not None else None for i in range(self.config.num_hidden_layers): layer_module = self.layer[i] 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 layer_outputs = layer_module( hidden_states, attention_mask, layer_head_mask, encoder_hidden_states, encoder_attention_mask, past_key_values, output_attentions, cache_position, ) hidden_states = layer_outputs[0] if output_attentions: all_self_attentions = all_self_attentions + (layer_outputs[1],) if encoder_hidden_states is not None: 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, past_key_values, all_hidden_states, all_self_attentions, all_cross_attentions, ] if v is not None ) return BaseModelOutputWithPastAndCrossAttentions( last_hidden_state=hidden_states, past_key_values=past_key_values, hidden_states=all_hidden_states, attentions=all_self_attentions, cross_attentions=all_cross_attentions, ) # Copied from transformers.models.bert.modeling_bert.BertPooler with Bert->BlipText class BlipTextPooler(nn.Module): def __init__(self, config): super().__init__() self.dense = nn.Linear(config.hidden_size, config.hidden_size) self.activation = nn.Tanh() def forward(self, hidden_states: torch.Tensor) -> torch.Tensor: # We "pool" the model by simply taking the hidden state corresponding # to the first token. first_token_tensor = hidden_states[:, 0] pooled_output = self.dense(first_token_tensor) pooled_output = self.activation(pooled_output) return pooled_output # Copied from transformers.models.bert.modeling_bert.BertPredictionHeadTransform with Bert->BlipText class BlipTextPredictionHeadTransform(nn.Module): def __init__(self, config): super().__init__() self.dense = nn.Linear(config.hidden_size, config.hidden_size) if isinstance(config.hidden_act, str): self.transform_act_fn = ACT2FN[config.hidden_act] else: self.transform_act_fn = config.hidden_act self.LayerNorm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps) def forward(self, hidden_states: torch.Tensor) -> torch.Tensor: hidden_states = self.dense(hidden_states) hidden_states = self.transform_act_fn(hidden_states) hidden_states = self.LayerNorm(hidden_states) return hidden_states # Copied from transformers.models.bert.modeling_bert.BertLMPredictionHead with Bert->BlipText class BlipTextLMPredictionHead(nn.Module): def __init__(self, config): super().__init__() self.transform = BlipTextPredictionHeadTransform(config) # The output weights are the same as the input embeddings, but there is # an output-only bias for each token. self.decoder = nn.Linear(config.hidden_size, config.vocab_size, bias=False) self.bias = nn.Parameter(torch.zeros(config.vocab_size)) # Need a link between the two variables so that the bias is correctly resized with `resize_token_embeddings` self.decoder.bias = self.bias def _tie_weights(self): self.decoder.bias = self.bias def forward(self, hidden_states): hidden_states = self.transform(hidden_states) hidden_states = self.decoder(hidden_states) return hidden_states # Copied from transformers.models.bert.modeling_bert.BertOnlyMLMHead with Bert->BlipText class BlipTextOnlyMLMHead(nn.Module): def __init__(self, config): super().__init__() self.predictions = BlipTextLMPredictionHead(config) def forward(self, sequence_output: torch.Tensor) -> torch.Tensor: prediction_scores = self.predictions(sequence_output) return prediction_scores # Adapted from https://github.com/salesforce/BLIP/blob/main/models/med.py#L548 class BlipTextPreTrainedModel(PreTrainedModel): """ An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained models. """ config: BlipTextConfig base_model_prefix = "bert" _no_split_modules = [] def _init_weights(self, module): """Initialize the weights""" if isinstance(module, (nn.Linear, nn.Embedding)): # 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) elif isinstance(module, nn.LayerNorm): module.bias.data.zero_() module.weight.data.fill_(1.0) if isinstance(module, nn.Linear) and module.bias is not None: module.bias.data.zero_() # Adapted from https://github.com/salesforce/BLIP/blob/3a29b7410476bf5f2ba0955827390eb6ea1f4f9d/models/med.py#L571 class BlipTextModel(BlipTextPreTrainedModel): """ The model can behave as an encoder (with only self-attention) as well as a decoder, in which case a layer of cross-attention is added between the self-attention layers, following the architecture described in [Attention is all you need](https://huggingface.co/papers/1706.03762) by Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser and Illia Polosukhin. argument and `is_decoder` set to `True`; an `encoder_hidden_states` is then expected as an input to the forward pass. """ def __init__(self, config, add_pooling_layer=True): super().__init__(config) self.config = config self.embeddings = BlipTextEmbeddings(config) self.encoder = BlipTextEncoder(config) self.pooler = BlipTextPooler(config) if add_pooling_layer else None self.post_init() def get_input_embeddings(self): return self.embeddings.word_embeddings def set_input_embeddings(self, value): self.embeddings.word_embeddings = value # Copied from transformers.models.bert.modeling_bert.BertModel._prune_heads 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) def get_extended_attention_mask( self, attention_mask: Tensor, input_shape: tuple[int], device: device, is_decoder: bool ) -> Tensor: """ Makes broadcastable attention and causal masks so that future and masked tokens are ignored. Arguments: attention_mask (`torch.Tensor`): Mask with ones indicating tokens to attend to, zeros for tokens to ignore. input_shape (`tuple[int]`): The shape of the input to the model. device (`torch.device`): The device of the input to the model. Returns: `torch.Tensor` The extended attention mask, with a the same dtype as `attention_mask.dtype`. """ # We can provide a self-attention mask of dimensions [batch_size, from_seq_length, to_seq_length] # ourselves in which case we just need to make it broadcastable to all heads. if attention_mask.dim() == 3: extended_attention_mask = attention_mask[:, None, :, :] elif attention_mask.dim() == 2: # Provided a padding mask of dimensions [batch_size, seq_length] # - if the model is a decoder, apply a causal mask in addition to the padding mask # - if the model is an encoder, make the mask broadcastable to [batch_size, num_heads, seq_length, seq_length] if is_decoder: batch_size, seq_length = input_shape seq_ids = torch.arange(seq_length, device=device) causal_mask = seq_ids[None, None, :].repeat(batch_size, seq_length, 1) <= seq_ids[None, :, None] # in case past_key_values are used we need to add a prefix ones mask to the causal mask causal_mask = causal_mask.to(attention_mask.dtype) if causal_mask.shape[1] < attention_mask.shape[1]: prefix_seq_len = attention_mask.shape[1] - causal_mask.shape[1] causal_mask = torch.cat( [ torch.ones( (batch_size, seq_length, prefix_seq_len), device=device, dtype=causal_mask.dtype ), causal_mask, ], axis=-1, ) extended_attention_mask = causal_mask[:, None, :, :] * attention_mask[:, None, None, :] else: extended_attention_mask = attention_mask[:, None, None, :] else: raise ValueError( f"Wrong shape for input_ids (shape {input_shape}) or attention_mask (shape {attention_mask.shape})" ) # Since attention_mask is 1.0 for positions we want to attend and 0.0 for # masked positions, this operation will create a tensor which is 0.0 for # positions we want to attend and -10000.0 for masked positions. # Since we are adding it to the raw scores before the softmax, this is # effectively the same as removing these entirely. extended_attention_mask = extended_attention_mask.to(dtype=self.dtype) # fp16 compatibility extended_attention_mask = (1.0 - extended_attention_mask) * -10000.0 return extended_attention_mask def forward( self, input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, encoder_embeds: Optional[torch.Tensor] = None, encoder_hidden_states: Optional[torch.Tensor] = None, encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[Cache] = None, use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, is_decoder: Optional[bool] = False, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: r""" encoder_hidden_states (`torch.FloatTensor`, *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 (`torch.FloatTensor`, *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 for tokens that are **not masked**, - 0 for tokens that are **masked**. past_key_values (`Cache`, *optional*): Contains precomputed key and value hidden states of the attention blocks. Can be used to speed up decoding. If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those that don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of all `decoder_input_ids` of shape `(batch_size, sequence_length)`. use_cache (`bool`, *optional*): If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). """ 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 is_decoder: use_cache = use_cache if use_cache is not None else self.config.use_cache else: use_cache = False 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: self.warn_if_padding_and_no_attention_mask(input_ids, attention_mask) input_shape = input_ids.size() batch_size, seq_length = input_shape device = input_ids.device elif inputs_embeds is not None: input_shape = inputs_embeds.size()[:-1] batch_size, seq_length = input_shape device = inputs_embeds.device elif encoder_embeds is not None: input_shape = encoder_embeds.size()[:-1] batch_size, seq_length = input_shape device = encoder_embeds.device else: raise ValueError("You have to specify either input_ids or inputs_embeds or encoder_embeds") past_key_values_length = 0 if past_key_values is not None: past_key_values_length = ( past_key_values[0][0].shape[-2] if not isinstance(past_key_values, Cache) else past_key_values.get_seq_length() ) if attention_mask is None: attention_mask = torch.ones((batch_size, seq_length + past_key_values_length)).to(device) # We can provide a self-attention mask of dimensions [batch_size, from_seq_length, to_seq_length] # ourselves in which case we just need to make it broadcastable to all heads. extended_attention_mask: torch.Tensor = self.get_extended_attention_mask( attention_mask, input_shape, device, is_decoder ) # If a 2D or 3D attention mask is provided for the cross-attention # we need to make broadcastable to [batch_size, num_heads, seq_length, seq_length] if encoder_hidden_states is not None: if isinstance(encoder_hidden_states, list): encoder_batch_size, encoder_sequence_length, _ = encoder_hidden_states[0].size() else: encoder_batch_size, encoder_sequence_length, _ = encoder_hidden_states.size() encoder_hidden_shape = (encoder_batch_size, encoder_sequence_length) if isinstance(encoder_attention_mask, list): encoder_extended_attention_mask = [self.invert_attention_mask(mask) for mask in encoder_attention_mask] elif encoder_attention_mask is None: encoder_attention_mask = torch.ones(encoder_hidden_shape, device=device) encoder_extended_attention_mask = self.invert_attention_mask(encoder_attention_mask) else: encoder_extended_attention_mask = self.invert_attention_mask(encoder_attention_mask) else: encoder_extended_attention_mask = None # Prepare head mask if needed # 1.0 in head_mask indicate we keep the head # attention_probs has shape bsz x n_heads x N x N # input head_mask has shape [num_heads] or [num_hidden_layers x num_heads] # and head_mask is converted to shape [num_hidden_layers x batch x num_heads x seq_length x seq_length] head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers) if encoder_embeds is None: embedding_output = self.embeddings( input_ids=input_ids, position_ids=position_ids, inputs_embeds=inputs_embeds, past_key_values_length=past_key_values_length, ) else: embedding_output = encoder_embeds encoder_outputs = self.encoder( embedding_output, attention_mask=extended_attention_mask, head_mask=head_mask, encoder_hidden_states=encoder_hidden_states, encoder_attention_mask=encoder_extended_attention_mask, past_key_values=past_key_values, use_cache=use_cache, output_attentions=output_attentions, output_hidden_states=output_hidden_states, return_dict=return_dict, cache_position=cache_position, ) sequence_output = encoder_outputs[0] pooled_output = self.pooler(sequence_output) if self.pooler is not None else None if not return_dict: return (sequence_output, pooled_output) + encoder_outputs[1:] return BaseModelOutputWithPoolingAndCrossAttentions( last_hidden_state=sequence_output, pooler_output=pooled_output, past_key_values=encoder_outputs.past_key_values, hidden_states=encoder_outputs.hidden_states, attentions=encoder_outputs.attentions, cross_attentions=encoder_outputs.cross_attentions, ) # Adapted from https://github.com/salesforce/BLIP/blob/main/models/med.py#L811 class BlipTextLMHeadModel(BlipTextPreTrainedModel, GenerationMixin): _tied_weights_keys = ["cls.predictions.decoder.weight", "cls.predictions.decoder.bias"] def __init__(self, config): super().__init__(config) self.bert = BlipTextModel(config, add_pooling_layer=False) self.cls = BlipTextOnlyMLMHead(config) self.label_smoothing = config.label_smoothing def get_input_embeddings(self): return self.bert.get_input_embeddings() def set_input_embeddings(self, new_embeddings): self.bert.set_input_embeddings(new_embeddings) def get_output_embeddings(self): return self.cls.predictions.decoder def set_output_embeddings(self, new_embeddings): self.cls.predictions.decoder = new_embeddings self.cls.predictions.bias = new_embeddings.bias def forward( self, input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, encoder_hidden_states: Optional[torch.Tensor] = None, encoder_attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, past_key_values: Optional[Cache] = None, use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, return_logits: Optional[bool] = False, is_decoder: Optional[bool] = True, reduction: Optional[str] = "mean", cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: r""" encoder_hidden_states (`torch.FloatTensor`, *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 (`torch.FloatTensor`, *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 for tokens that are **not masked**, - 0 for tokens that are **masked**. labels (`torch.LongTensor`, *optional*): Labels for computing the left-to-right language modeling loss (next word prediction). 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 n `[0, ..., config.vocab_size]` past_key_values (`Cache`, *optional*): Contains precomputed key and value hidden states of the attention blocks. Can be used to speed up decoding. If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those that don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of all `decoder_input_ids` of shape `(batch_size, sequence_length)`. use_cache (`bool`, *optional*): If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict if labels is not None: use_cache = False outputs = self.bert( input_ids, attention_mask=attention_mask, position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, encoder_hidden_states=encoder_hidden_states, encoder_attention_mask=encoder_attention_mask, past_key_values=past_key_values, use_cache=use_cache, output_attentions=output_attentions, output_hidden_states=output_hidden_states, return_dict=return_dict, is_decoder=is_decoder, cache_position=cache_position, ) sequence_output = outputs[0] prediction_scores = self.cls(sequence_output) if return_logits: return prediction_scores[:, :-1, :].contiguous() lm_loss = None if labels is not None: # we are doing next-token prediction; shift prediction scores and input ids by one shifted_prediction_scores = prediction_scores[:, :-1, :].contiguous() labels = labels[:, 1:].contiguous().to(shifted_prediction_scores.device) loss_fct = CrossEntropyLoss(reduction=reduction, label_smoothing=self.label_smoothing) lm_loss = loss_fct(shifted_prediction_scores.view(-1, self.config.vocab_size), labels.view(-1)) if reduction == "none": lm_loss = lm_loss.view(prediction_scores.size(0), -1).sum(1) if not return_dict: output = (prediction_scores,) + outputs[2:] return ((lm_loss,) + output) if lm_loss is not None else output return CausalLMOutputWithCrossAttentions( loss=lm_loss, logits=prediction_scores, past_key_values=outputs.past_key_values, hidden_states=outputs.hidden_states, attentions=outputs.attentions, cross_attentions=outputs.cross_attentions, ) def prepare_inputs_for_generation(self, input_ids, past_key_values=None, attention_mask=None, **model_kwargs): # Overwrite -- hardcoded key return (`is_decoder=True`) model_inputs = super().prepare_inputs_for_generation( input_ids, past_key_values=past_key_values, attention_mask=attention_mask, **model_kwargs, ) model_inputs["is_decoder"] = True return model_inputs __all__ = ["BlipTextModel", "BlipTextLMHeadModel", "BlipTextPreTrainedModel"]