# coding=utf-8 # Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved. # # This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX # and OPT implementations in this library. It has been modified from its # original forms to accommodate minor architectural differences compared # to GPT-NeoX and OPT used by the Meta AI team that trained the model. # # 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 Idefics model.""" from dataclasses import dataclass from typing import Any, Dict, List, Optional, Tuple, Union import torch import torch.nn.functional as F import torch.utils.checkpoint from torch import nn from torch.nn import CrossEntropyLoss from ...activations import ACT2FN from ...cache_utils import Cache, DynamicCache, StaticCache from ...generation import GenerationMixin from ...modeling_attn_mask_utils import AttentionMaskConverter from ...modeling_outputs import ModelOutput from ...modeling_utils import PretrainedConfig, PreTrainedModel from ...pytorch_utils import ALL_LAYERNORM_LAYERS from ...utils import ( add_start_docstrings, add_start_docstrings_to_model_forward, is_torch_flex_attn_available, logging, replace_return_docstrings, ) from .configuration_idefics import IdeficsConfig from .perceiver import IdeficsPerceiverResampler from .vision import IdeficsVisionTransformer if is_torch_flex_attn_available(): from torch.nn.attention.flex_attention import BlockMask from ...integrations.flex_attention import make_flex_block_causal_mask logger = logging.get_logger(__name__) _CONFIG_FOR_DOC = "IdeficsConfig" @dataclass class IdeficsBaseModelOutputWithPast(ModelOutput): """ Base class for Idefics model's outputs that may also contain a past key/values (to speed up sequential decoding). Args: last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`): Sequence of hidden-states at the output of the last layer of the model. If `past_key_values` is used only the last hidden-state of the sequences of shape `(batch_size, 1, hidden_size)` is output. past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`): Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and optionally if `config.is_encoder_decoder=True` 2 additional tensors of shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`. Contains pre-computed hidden-states (key and values in the self-attention blocks and optionally if `config.is_encoder_decoder=True` in the cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding. hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`): Tuple of `torch.FloatTensor` (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer plus the optional initial embedding outputs. attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`): Tuple of `torch.FloatTensor` (one for each layer) of shape `(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. image_hidden_states (`tuple(torch.FloatTensor)`, *optional*): Tuple of `torch.FloatTensor` (one for the output of the image embeddings, `(batch_size, num_images, sequence_length, hidden_size)`. image_hidden_states of the model produced by the vision encoder, and optionally by the perceiver """ last_hidden_state: Optional[torch.FloatTensor] = None past_key_values: Optional[Tuple[Tuple[torch.FloatTensor]]] = None hidden_states: Optional[Tuple[torch.FloatTensor]] = None attentions: Optional[Tuple[torch.FloatTensor]] = None image_hidden_states: Optional[Tuple[torch.FloatTensor]] = None @dataclass class IdeficsCausalLMOutputWithPast(ModelOutput): """ Base class for Idefics causal language model (or autoregressive) outputs. Args: loss (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided): Language modeling loss (for next-token prediction). logits (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`): Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax). past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`): Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) Contains pre-computed hidden-states (key and values in the self-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding. hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`): Tuple of `torch.FloatTensor` (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer plus the optional initial embedding outputs. attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`): Tuple of `torch.FloatTensor` (one for each layer) of shape `(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. image_hidden_states (`tuple(torch.FloatTensor)`, *optional*): Tuple of `torch.FloatTensor` (one for the output of the image embeddings, `(batch_size, num_images, sequence_length, hidden_size)`. image_hidden_states of the model produced by the vision encoder, and optionally by the perceiver """ loss: Optional[torch.FloatTensor] = None logits: Optional[torch.FloatTensor] = None past_key_values: Optional[List[torch.FloatTensor]] = None hidden_states: Optional[Tuple[torch.FloatTensor]] = None attentions: Optional[Tuple[torch.FloatTensor]] = None image_hidden_states: Optional[Tuple[torch.FloatTensor]] = None def expand_inputs_for_generation( input_ids, expand_size=1, is_encoder_decoder=False, attention_mask=None, encoder_outputs=None, **model_kwargs, ): expanded_return_idx = ( torch.arange(input_ids.shape[0]).view(-1, 1).repeat(1, expand_size).view(-1).to(input_ids.device) ) input_ids = input_ids.index_select(0, expanded_return_idx) model_kwargs["pixel_values"] = model_kwargs.get("pixel_values", None) model_kwargs["image_encoder_embeddings"] = model_kwargs.get("image_encoder_embeddings", None) model_kwargs["perceiver_embeddings"] = model_kwargs.get("perceiver_embeddings", None) model_kwargs["image_attention_mask"] = model_kwargs.get("image_attention_mask", None) if "token_type_ids" in model_kwargs: token_type_ids = model_kwargs["token_type_ids"] model_kwargs["token_type_ids"] = token_type_ids.index_select(0, expanded_return_idx) if attention_mask is not None: model_kwargs["attention_mask"] = attention_mask.index_select(0, expanded_return_idx) if model_kwargs["image_attention_mask"] is not None: model_kwargs["image_attention_mask"] = model_kwargs["image_attention_mask"].index_select( 0, expanded_return_idx ) if model_kwargs["pixel_values"] is not None: model_kwargs["pixel_values"] = model_kwargs["pixel_values"].index_select(0, expanded_return_idx) elif model_kwargs["image_encoder_embeddings"] is not None: model_kwargs["image_encoder_embeddings"] = model_kwargs["image_encoder_embeddings"].index_select( 0, expanded_return_idx ) elif model_kwargs["perceiver_embeddings"] is not None: model_kwargs["perceiver_embeddings"] = model_kwargs["perceiver_embeddings"].index_select( 0, expanded_return_idx ) return input_ids, model_kwargs def freeze_model(model, module_exceptions=[]): mapping = { "LayerNorm": nn.LayerNorm, "Linear": nn.Linear, "Embedding": nn.Embedding, } module_exceptions_mapped = [mapping[m] for m in module_exceptions] for module in model.modules(): if module_exceptions and any(isinstance(module, t) for t in module_exceptions_mapped): module.requires_grad_(True) # Explicitely setting it to true to avoid any mistakes else: module.requires_grad_(False) return model class IdeficsDecoupledEmbedding(nn.Embedding): # Derived from https://pytorch.org/docs/stable/_modules/torch/nn/modules/sparse.html#Embedding """ Implements a decoupling of parameters to allow freezing (or not) a subset of the embeddings. In practise, the regular `weight` can be trained or frozen (i.e. `partially_freeze=True`), and if `num_additional_embeddings` > 0, then it will create `num_additional_embeddings` additional parameters that are always trained. If `num_additional_embeddings=0`, then the module defaults back to the regular behavior of `nn.Embedding`. """ def __init__( self, num_embeddings, num_additional_embeddings, embedding_dim, partially_freeze: Optional[bool] = False, device=None, dtype=None, padding_idx=None, **kwargs, ) -> None: """ Args: num_embeddings (`int`): Size of the dictionary of embeddings num_additional_embeddings (`int`): Number of additional embeddings. Only useful when you `partially_freeze=True`. embedding_dim (`int`): The size of each embedding vector partially_freeze: (`bool`, *optional*, defaults to `False`): If `True`, the regular `weight` will be frozen. `additional_weight` is never frozen. padding_idx (`int`, *optional*): The padding index (needs to be less than num_embeddings) Note: there are a lot of other parameters to initialize a standard `nn.Embedding` such as `padding_idx`, `max_norm` or `norm_type`. We are not supporting these. """ if padding_idx is not None and padding_idx > num_embeddings: raise ValueError(f"padding_idx must be within num_embeddings. Got {padding_idx} and {num_embeddings}") super().__init__( num_embeddings=num_embeddings, embedding_dim=embedding_dim, device=device, dtype=dtype, padding_idx=padding_idx, **kwargs, ) self.num_embeddings = num_embeddings self.padding_idx = padding_idx self.num_additional_embeddings = num_additional_embeddings self.partially_freeze = partially_freeze if partially_freeze: self.weight.requires_grad_(False) if self.num_additional_embeddings > 0: self.additional_embedding = nn.Embedding( num_embeddings=self.num_additional_embeddings, embedding_dim=embedding_dim, device=device, dtype=dtype, ) def forward(self, input_ids): """ we have 2 embeddings, with different indices - one pretrained self.weight and another self.additional_embedding.weight that is being trained. in order to make a lookup of the input ids, we: 1. find out the indices of the entries belonging to the 2nd embedding 2. extract those values while subtracting the size of the first embedding (num_embeddings), since the 2nd embedding starts from 0 and not num_embeddings 3. perform the 2nd embedding lookup 4. now we handle the 1st embedding, we overwrite indices belonging to the 2nd embedding with a padding index 5. perform the 1st embedding lookup 6. now we overwrite the values in the 1st embedding lookup with the values of the 2nd embedding lookup note: for the 1st embedding lookup we could have looked up only the low indices and not do the padding, but then we have to create a new tensor and populate it with 2 tensors that are spread out across various indices - i.e. not a simple concat - I haven't benchmarked the complex case if it's any faster, given that seqlens are usually relatively short it's probably not faster or if faster not by much - but might be a good idea to measure. """ if self.num_additional_embeddings == 0: return F.embedding(input_ids, self.weight) # Clone so that we don't modify the original input_ids later on input_ids = input_ids.clone() additional_vocab_indices = torch.where(input_ids >= self.num_embeddings) input_ids_additional_vocab = input_ids[additional_vocab_indices] additional_embeddings = self.additional_embedding(input_ids_additional_vocab - self.num_embeddings) # for successful lookup replace input_ids with 0, the results of these will be discarded anyway input_ids[additional_vocab_indices] = 0 full_vector = F.embedding(input_ids, self.weight) # overwrite the records with high indices full_vector[additional_vocab_indices] = additional_embeddings return full_vector def extra_repr(self) -> str: return "num_embeddings={}, num_additional_embeddings={}, embedding_dim={}, partially_freeze={}".format( self.num_embeddings, self.num_additional_embeddings, self.embedding_dim, self.partially_freeze, ) class IdeficsDecoupledLinear(nn.Linear): # Derived from https://pytorch.org/docs/stable/_modules/torch/nn/modules/linear.html#Linear """ Implements a decoupling of parameters to allow freezing (or not) a subset of the parameters. In practise, the regular `weight` can be trained or frozen (i.e. `partially_freeze=True`), and if `out_additional_features` > 0, then it will create `out_additional_features * in_features` additional parameters that are always trained. If `out_additional_features=0`, then the module defaults back to the regular behavior of `nn.Linear`. """ def __init__( self, in_features: int, out_features: int, out_additional_features: int = 0, bias: bool = True, partially_freeze: bool = True, device=None, dtype=None, ) -> None: """ out_additional_features: int. Number of additional trainable dimensions. Only makes sense when `partially_freeze=True`. partially_freeze: bool. If True, the regular `weight` will be frozen and extra parameters (if any) will be trainable. If False, default to the regular behavior of nn.Linear. """ super().__init__(in_features, out_features, bias, device, dtype) self.out_additional_features = out_additional_features self.partially_freeze = partially_freeze self.in_features = in_features self.out_features = out_features if partially_freeze: self.weight.requires_grad_(False) if bias: self.bias.requires_grad_(False) if out_additional_features > 0: self.additional_fc = nn.Linear( in_features=in_features, out_features=out_additional_features, bias=bias, device=device, dtype=dtype, ) def forward(self, input: torch.Tensor) -> torch.Tensor: output = F.linear(input, self.weight, self.bias) if self.out_additional_features > 0: additional_features = self.additional_fc(input) output = torch.cat((output, additional_features), -1) return output def extra_repr(self) -> str: """Overwriting `nn.Linear.extra_repr` to include new parameters.""" return "in_features={}, out_features={}, out_additional_features={}, bias={}, partially_freeze={}".format( self.in_features, self.out_features, self.out_additional_features, self.bias is not None, self.partially_freeze, ) # this was adapted from LlamaRMSNorm class IdeficsRMSNorm(nn.Module): def __init__(self, hidden_size, eps=1e-6): """ IdeficsRMSNorm is equivalent to T5LayerNorm """ super().__init__() self.weight = nn.Parameter(torch.ones(hidden_size)) self.variance_epsilon = eps def forward(self, hidden_states): variance = hidden_states.to(torch.float32).pow(2).mean(-1, keepdim=True) hidden_states = hidden_states * torch.rsqrt(variance + self.variance_epsilon) # convert into half-precision if necessary if self.weight.dtype in [torch.float16, torch.bfloat16]: hidden_states = hidden_states.to(self.weight.dtype) return self.weight * hidden_states def extra_repr(self): return f"{tuple(self.weight.shape)}, eps={self.variance_epsilon}" ALL_LAYERNORM_LAYERS.append(IdeficsRMSNorm) # this was adapted from LlamaRotaryEmbedding class IdeficsEmbedding(torch.nn.Module): def __init__(self, dim, max_position_embeddings=2048, base=10000, device=None): super().__init__() self.dim = dim self.max_position_embeddings = max_position_embeddings self.base = base inv_freq = 1.0 / ( self.base ** (torch.arange(0, self.dim, 2, dtype=torch.int64).to(device=device, dtype=torch.float) / self.dim) ) self.register_buffer("inv_freq", inv_freq, persistent=False) # Build here to make `torch.jit.trace` work. self._set_cos_sin_cache( seq_len=max_position_embeddings, device=self.inv_freq.device, dtype=torch.get_default_dtype() ) def _set_cos_sin_cache(self, seq_len, device, dtype): self.max_seq_len_cached = seq_len t = torch.arange(self.max_seq_len_cached, device=device, dtype=torch.int64).type_as(self.inv_freq) freqs = torch.einsum("i,j->ij", t, self.inv_freq) # Different from paper, but it uses a different permutation in order to obtain the same calculation emb = torch.cat((freqs, freqs), dim=-1) self.register_buffer("cos_cached", emb.cos().to(dtype), persistent=False) self.register_buffer("sin_cached", emb.sin().to(dtype), persistent=False) def forward(self, x, seq_len=None): # x: [bs, num_attention_heads, seq_len, head_size] if seq_len > self.max_seq_len_cached: self._set_cos_sin_cache(seq_len=seq_len, device=x.device, dtype=x.dtype) return ( self.cos_cached[:seq_len].to(dtype=x.dtype), self.sin_cached[:seq_len].to(dtype=x.dtype), ) def rotate_half(x): """Rotates half the hidden dims of the input.""" x1 = x[..., : x.shape[-1] // 2] x2 = x[..., x.shape[-1] // 2 :] return torch.cat((-x2, x1), dim=-1) def apply_rotary_pos_emb(q, k, cos, sin, position_ids, unsqueeze_dim=1): """Applies Rotary Position Embedding to the query and key tensors. Args: q (`torch.Tensor`): The query tensor. k (`torch.Tensor`): The key tensor. cos (`torch.Tensor`): The cosine part of the rotary embedding. sin (`torch.Tensor`): The sine part of the rotary embedding. position_ids (`torch.Tensor`): The position indices of the tokens corresponding to the query and key tensors. For example, this can be used to pass offsetted position ids when working with a KV-cache. unsqueeze_dim (`int`, *optional*, defaults to 1): The 'unsqueeze_dim' argument specifies the dimension along which to unsqueeze cos[position_ids] and sin[position_ids] so that they can be properly broadcasted to the dimensions of q and k. For example, note that cos[position_ids] and sin[position_ids] have the shape [batch_size, seq_len, head_dim]. Then, if q and k have the shape [batch_size, heads, seq_len, head_dim], then setting unsqueeze_dim=1 makes cos[position_ids] and sin[position_ids] broadcastable to the shapes of q and k. Similarly, if q and k have the shape [batch_size, seq_len, heads, head_dim], then set unsqueeze_dim=2. Returns: `tuple(torch.Tensor)` comprising of the query and key tensors rotated using the Rotary Position Embedding. """ cos = cos[position_ids].unsqueeze(unsqueeze_dim) sin = sin[position_ids].unsqueeze(unsqueeze_dim) q_embed = (q * cos) + (rotate_half(q) * sin) k_embed = (k * cos) + (rotate_half(k) * sin) return q_embed, k_embed # this was adapted from LlamaMLP class IdeficsMLP(nn.Module): def __init__( self, hidden_size: int, intermediate_size: int, hidden_act: str, ): super().__init__() self.gate_proj = nn.Linear(hidden_size, intermediate_size, bias=False) self.down_proj = nn.Linear(intermediate_size, hidden_size, bias=False) self.up_proj = nn.Linear(hidden_size, intermediate_size, bias=False) self.act_fn = ACT2FN[hidden_act] def forward(self, x): return self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x)) # this was adapted from LlamaAttention class IdeficsAttention(nn.Module): """Multi-headed attention from 'Attention Is All You Need' paper""" def __init__( self, hidden_size: int, num_heads: int, dropout: float = 0.0, is_cross_attention: bool = False, config: PretrainedConfig = None, qk_layer_norms: bool = False, layer_idx: Optional[int] = None, ): super().__init__() self.hidden_size = hidden_size self.num_heads = num_heads self.head_dim = hidden_size // num_heads self.dropout = dropout self.is_causal = True self.layer_idx = layer_idx if layer_idx is None: logger.warning_once( f"Instantiating {self.__class__.__name__} without passing a `layer_idx` is not recommended and will " "lead to errors during the forward call if caching is used. Please make sure to provide a `layer_idx` " "when creating this class." ) if (self.head_dim * num_heads) != self.hidden_size: raise ValueError( f"hidden_size must be divisible by num_heads (got `hidden_size`: {self.hidden_size}" f" and `num_heads`: {num_heads})." ) self.is_cross_attention = is_cross_attention if not hasattr(nn.functional, "scaled_dot_product_attention"): raise ValueError("this model requires pytorch 2.0 or higher") if self.is_cross_attention: kv_input_dim = ( self.hidden_size if not hasattr(config.vision_config, "embed_dim") else config.vision_config.embed_dim ) self.q_proj = nn.Linear( self.hidden_size, num_heads * self.head_dim, bias=False, ) self.k_proj = nn.Linear(kv_input_dim, num_heads * self.head_dim, bias=False) self.v_proj = nn.Linear( kv_input_dim, num_heads * self.head_dim, bias=False, ) else: self.q_proj = nn.Linear( self.hidden_size, num_heads * self.head_dim, bias=False, ) self.k_proj = nn.Linear( self.hidden_size, num_heads * self.head_dim, bias=False, ) self.v_proj = nn.Linear( self.hidden_size, num_heads * self.head_dim, bias=False, ) self.o_proj = nn.Linear( num_heads * self.head_dim, hidden_size, bias=False, ) self.rotary_emb = IdeficsEmbedding(self.head_dim) self.qk_layer_norms = qk_layer_norms if self.qk_layer_norms: self.q_layer_norm = IdeficsRMSNorm(self.head_dim, eps=config.rms_norm_eps) self.k_layer_norm = IdeficsRMSNorm(self.head_dim, eps=config.rms_norm_eps) def _shape(self, tensor: torch.Tensor, seq_len: int, bsz: int): return tensor.view(bsz, seq_len, self.num_heads, self.head_dim).transpose(1, 2).contiguous() def forward( self, hidden_states: torch.Tensor, key_value_states: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Tuple[torch.Tensor]] = None, output_attentions: bool = False, use_cache: bool = False, cache_position: Optional[torch.LongTensor] = None, ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]: # if key_value_states are provided this layer is used as a cross-attention layer is_cross_attention = self.is_cross_attention or key_value_states is not None bsz, q_len, _ = hidden_states.size() query_states = self.q_proj(hidden_states).view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2) if not is_cross_attention: key_states = self.k_proj(hidden_states).view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2) value_states = self.v_proj(hidden_states).view(bsz, q_len, self.num_heads, self.head_dim).transpose(1, 2) else: _, kv_len, _ = key_value_states.size() # Note that, in this case, `kv_len` == `kv_seq_len` key_states = self.k_proj(key_value_states).view(bsz, kv_len, self.num_heads, self.head_dim).transpose(1, 2) value_states = ( self.v_proj(key_value_states).view(bsz, kv_len, self.num_heads, self.head_dim).transpose(1, 2) ) kv_seq_len = key_states.shape[-2] if past_key_value is not None: kv_seq_len += cache_position[0] if not is_cross_attention: cos, sin = self.rotary_emb(value_states, seq_len=max(kv_seq_len, q_len)) query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids) # [bsz, nh, t, hd] if past_key_value is not None: # sin and cos are specific to RoPE models; cache_position needed for the static cache cache_kwargs = {"cache_position": cache_position} key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs) if self.qk_layer_norms: query_states = self.q_layer_norm(query_states) key_states = self.k_layer_norm(key_states) causal_mask = attention_mask if attention_mask is not None: causal_mask = causal_mask[:, :, :, : key_states.shape[-2]] # SDPA with memory-efficient backend is currently (torch==2.1.2) bugged with non-contiguous inputs with custom attn_mask, # Reference: https://github.com/pytorch/pytorch/issues/112577. if query_states.device.type == "cuda" and attention_mask is not None: query_states = query_states.contiguous() key_states = key_states.contiguous() value_states = value_states.contiguous() # We dispatch to SDPA's Flash Attention or Efficient kernels via this `is_causal` if statement instead of an inline conditional assignment # in SDPA to support both torch.compile's dynamic shapes and full graph options. An inline conditional prevents dynamic shapes from compiling. # The q_len > 1 is necessary to match with AttentionMaskConverter.to_causal_4d that does not create a causal mask in case q_len == 1. is_causal = True if self.is_causal and causal_mask is None and q_len > 1 else False attn_output = torch.nn.functional.scaled_dot_product_attention( query_states, key_states, value_states, attn_mask=causal_mask, dropout_p=self.dropout if self.training else 0.0, is_causal=is_causal, ) if attn_output.size() != (bsz, self.num_heads, q_len, self.head_dim): raise ValueError( f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.head_dim)}, but is" f" {attn_output.size()}" ) attn_output = attn_output.transpose(1, 2) attn_output = attn_output.reshape(bsz, q_len, self.hidden_size) attn_output = self.o_proj(attn_output) attn_weights = None if output_attentions: logger.warning_once( "attn_weights are not extracted in scaled_dot_product_attention. The model returns None instead" ) return attn_output, attn_weights, past_key_value # this was adapted from LlamaDecoderLayer class IdeficsDecoderLayer(nn.Module): def __init__(self, config: IdeficsConfig, layer_idx: Optional[int] = None): super().__init__() self.hidden_size = config.hidden_size self.self_attn = IdeficsAttention( hidden_size=self.hidden_size, num_heads=config.num_attention_heads, dropout=config.dropout, config=config, layer_idx=layer_idx, ) self.mlp = IdeficsMLP( hidden_size=self.hidden_size, intermediate_size=config.intermediate_size, hidden_act=config.hidden_act, ) self.input_layernorm = IdeficsRMSNorm(config.hidden_size, eps=config.rms_norm_eps) self.post_attention_layernorm = IdeficsRMSNorm(config.hidden_size, eps=config.rms_norm_eps) self.dropout = config.dropout def forward( self, hidden_states: torch.Tensor, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Tuple[torch.Tensor]] = None, output_attentions: Optional[bool] = False, use_cache: Optional[bool] = False, cache_position: Optional[torch.LongTensor] = None, ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]: """ Args: hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)` attention_mask (`torch.FloatTensor`, *optional*): attention mask of size `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values. output_attentions (`bool`, *optional*): Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned tensors for more detail. 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`). past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states """ residual = hidden_states hidden_states = self.input_layernorm(hidden_states) # Self Attention hidden_states, self_attn_weights, present_key_value = self.self_attn( hidden_states=hidden_states, attention_mask=attention_mask, position_ids=position_ids, past_key_value=past_key_value, output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, ) hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training) hidden_states = residual + hidden_states # Fully Connected residual = hidden_states hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training) hidden_states = residual + hidden_states outputs = (hidden_states,) if output_attentions: outputs += (self_attn_weights,) if use_cache: outputs += (present_key_value,) return outputs class IdeficsGatedCrossAttentionLayer(nn.Module): def __init__(self, config: IdeficsConfig, layer_idx: Optional[int] = None): super().__init__() self.hidden_size = config.hidden_size self.cross_attn = IdeficsAttention( hidden_size=self.hidden_size, num_heads=config.num_attention_heads, is_cross_attention=True, dropout=config.dropout, config=config, qk_layer_norms=config.qk_layer_norms, layer_idx=layer_idx, ) self.mlp = IdeficsMLP( hidden_size=self.hidden_size, intermediate_size=config.intermediate_size, hidden_act=config.hidden_act, ) self.input_layernorm = IdeficsRMSNorm(config.hidden_size, eps=config.rms_norm_eps) self.post_attention_layernorm = IdeficsRMSNorm(config.hidden_size, eps=config.rms_norm_eps) self.config = config.dropout self.act_cross_attn = nn.Tanh() self.act_dense = nn.Tanh() if config.alpha_initializer == "zeros": if config.alpha_type == "vector": self.alpha_cross_attn = nn.Parameter(torch.zeros(1, 1, self.hidden_size)) self.alpha_dense = nn.Parameter(torch.zeros(1, 1, self.hidden_size)) elif config.alpha_type == "float": self.alpha_cross_attn = nn.Parameter(torch.zeros(1)) self.alpha_dense = nn.Parameter(torch.zeros(1)) else: raise ValueError(f"Unknown value for `alpha_type` ({config.alpha_type})") elif config.alpha_initializer == "ones": if config.alpha_type == "vector": self.alpha_cross_attn = nn.Parameter(torch.ones(1, 1, self.hidden_size)) self.alpha_dense = nn.Parameter(torch.ones(1, 1, self.hidden_size)) elif config.alpha_type == "float": self.alpha_cross_attn = nn.Parameter(torch.ones(1)) self.alpha_dense = nn.Parameter(torch.ones(1)) else: raise ValueError(f"Unknown value for `alpha_type` ({config.alpha_type})") elif config.alpha_initializer in {"normal", "gaussian", "random"}: if config.alpha_type == "vector": self.alpha_cross_attn = nn.Parameter( torch.normal(mean=0.0, std=config.alphas_initializer_range, size=(1, 1, self.hidden_size)) ) self.alpha_dense = nn.Parameter( torch.normal(mean=0.0, std=config.alphas_initializer_range, size=(1, 1, self.hidden_size)) ) elif config.alpha_type == "float": self.alpha_cross_attn = nn.Parameter( torch.normal(mean=0.0, std=config.alphas_initializer_range, size=(1)) ) self.alpha_dense = nn.Parameter(torch.normal(mean=0.0, std=config.alphas_initializer_range, size=(1))) else: raise ValueError(f"Unknown value for `alpha_type` ({config.alpha_type})") else: raise NotImplementedError(f"Alpha initialization scheme {config.alpha_initializer} not yet implemented!") if not (hasattr(self, "alpha_cross_attn") and hasattr(self, "alpha_dense")): raise ValueError("Alpha parameters not initialized correctly!") def forward( self, hidden_states: torch.Tensor, attention_mask: Optional[torch.Tensor] = None, image_hidden_states: Optional[torch.Tensor] = None, image_attention_mask: Optional[torch.Tensor] = None, cross_attention_gate: Optional[torch.Tensor] = None, output_attentions: Optional[bool] = False, use_cache: Optional[bool] = False, past_key_value: Optional[Tuple[torch.Tensor]] = None, ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]: """ Args: hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)` attention_mask (`torch.FloatTensor`, *optional*): attention mask of size `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values. image_attention_mask (`torch.FloatTensor`, *optional*): image attention mask of size `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values. cross_attention_gate (`torch.FloatTensor`, *optional*): gate of size `(batch, seq_len)` used to zero-out cross-attention output for tokens attending no images. output_attentions (`bool`, *optional*): Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned tensors for more detail. 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`). past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states """ if image_hidden_states is None: raise ValueError( "`image_hidden_states` is required for Idefics cross attention module which are visual features to be" " conditioned on." ) if cross_attention_gate is None: raise ValueError( "`cross_attention_gate` is required for Idefics cross attention module to zero-out the cross-attention hidden_states attending to no images." ) if past_key_value is not None: raise NotImplementedError("Past key value states are not implemented for Idefics cross attention module.") residual = hidden_states hidden_states = self.input_layernorm(hidden_states) # Self Attention hidden_states, self_attn_weights, present_key_value = self.cross_attn( hidden_states=hidden_states, key_value_states=image_hidden_states, attention_mask=image_attention_mask, output_attentions=output_attentions, ) hidden_states = nn.functional.dropout(hidden_states, p=self.config, training=self.training) # Fill in zeros for cross_attention hidden_states of tokens attending to no images hidden_states = hidden_states.masked_fill((cross_attention_gate == 0)[:, :, None], 0.0) hidden_states = residual + self.act_cross_attn(self.alpha_cross_attn) * hidden_states # Fully Connected residual = hidden_states hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = nn.functional.dropout(hidden_states, p=self.config, training=self.training) hidden_states = residual + self.act_dense(self.alpha_dense) * hidden_states outputs = (hidden_states,) if output_attentions: outputs += (self_attn_weights,) if use_cache: outputs += (present_key_value,) return outputs LLAMA_START_DOCSTRING = r""" This model inherits from [`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](https://pytorch.org/docs/stable/nn.html#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 ([`IdeficsConfig`]): 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 [`~PreTrainedModel.from_pretrained`] method to load the model weights. """ @add_start_docstrings( "The bare LLaMA Model outputting raw hidden-states without any specific head on top.", LLAMA_START_DOCSTRING, ) class IdeficsPreTrainedModel(PreTrainedModel): config_class = IdeficsConfig base_model_prefix = "model" supports_gradient_checkpointing = True _no_split_modules = ["IdeficsDecoderLayer", "IdeficsGatedCrossAttentionLayer"] _supports_sdpa = True _supports_cache_class = True _supports_static_cache = False # IDEFICS cannot compile due to dynamic control flow when checking inputs def _init_weights(self, module): # important: this ported version of Idefics isn't meant for training from scratch - only # inference and fine-tuning - so the proper init weights code has been removed - the m4 code # base should be used for training from scratch and it contains the correct code. std = self.config.initializer_range if isinstance(module, nn.Linear): module.weight.data.normal_(mean=0.0, std=std) if module.bias is not None: module.bias.data.zero_() elif isinstance(module, nn.Embedding): module.weight.data.normal_(mean=0.0, std=std) if module.padding_idx is not None: module.weight.data[module.padding_idx].zero_() LLAMA_INPUTS_DOCSTRING = r""" Args: input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`): Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you provide it. Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for details. [What are input IDs?](../glossary#input-ids) attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *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#attention-mask) Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for details. If `past_key_values` is used, optionally only the last `decoder_input_ids` have to be input (see `past_key_values`). If you want to change padding behavior, you should read [`modeling_opt._prepare_decoder_attention_mask`] and modify to your needs. See diagram 1 in [the paper](https://arxiv.org/abs/1910.13461) for more information on the default strategy. - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. position_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.n_positions - 1]`. [What are position IDs?](../glossary#position-ids) past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`): Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`. Contains pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential 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)`. inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*): Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model's internal embedding lookup matrix. 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 (`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 (`bool`, *optional*): Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for more detail. return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*): Indices depicting the position of the input sequence tokens in the sequence. Contrarily to `position_ids`, this tensor is not affected by padding. It is used to update the cache in the correct position and to infer the complete sequence length. """ @add_start_docstrings( "The bare LLaMA Model outputting raw hidden-states without any specific head on top.", LLAMA_START_DOCSTRING, ) class IdeficsModel(IdeficsPreTrainedModel): """ Transformer decoder consisting of `config.num_hidden_layers` layers. Each layer is a [`IdeficsDecoderLayer`] Args: config: IdeficsConfig """ def __init__(self, config: IdeficsConfig): super().__init__(config) self.config = config self.padding_idx = config.pad_token_id self.vocab_size = config.vocab_size self.embed_tokens = IdeficsDecoupledEmbedding( num_embeddings=config.vocab_size, num_additional_embeddings=config.additional_vocab_size, embedding_dim=config.hidden_size, partially_freeze=config.freeze_text_layers, padding_idx=self.padding_idx, ) self.image_size = config.vision_config.image_size self.vision_config = config.vision_config self.vision_model = IdeficsVisionTransformer(config.vision_config) # Perceiver Resampler if config.use_resampler: perceiver_config = config.perceiver_config self.perceiver_resampler = IdeficsPerceiverResampler( config, config.vision_config.embed_dim, perceiver_config.resampler_depth, perceiver_config.resampler_n_heads, perceiver_config.resampler_head_dim, perceiver_config.resampler_n_latents, ) self.layers = nn.ModuleList( [IdeficsDecoderLayer(config, layer_idx=i) for i in range(config.num_hidden_layers)] ) self.cross_layer_interval = config.cross_layer_interval num_cross_layers = config.num_hidden_layers // self.cross_layer_interval self.gated_cross_attn_layers = nn.ModuleList( [IdeficsGatedCrossAttentionLayer(config, layer_idx=i) for i in range(num_cross_layers)] ) self.gradient_checkpointing = False self.norm = IdeficsRMSNorm(config.hidden_size, eps=config.rms_norm_eps) # Initialize weights and apply final processing self.post_init() self.freeze_relevant_params(config) def freeze_relevant_params(self, config=None): if config is None: config = self.config if config.freeze_text_layers: self.freeze_text_layers(config.freeze_text_module_exceptions) if config.freeze_vision_layers: freeze_model(self.vision_model, module_exceptions=config.freeze_vision_module_exceptions) def freeze_text_layers(self, module_exceptions=[]): for module in [self.layers, self.norm]: freeze_model(module, module_exceptions=module_exceptions) def freeze_vision_layers(self, module_exceptions=[]): freeze_model(self.vision_model, module_exceptions=module_exceptions) def get_input_embeddings(self): return self.embed_tokens def set_input_embeddings(self, value): self.embed_tokens = value @add_start_docstrings_to_model_forward(LLAMA_INPUTS_DOCSTRING) def forward( self, input_ids: Optional[torch.LongTensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[List[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, pixel_values: Optional[torch.FloatTensor] = None, image_encoder_embeddings: Optional[torch.FloatTensor] = None, perceiver_embeddings: Optional[torch.FloatTensor] = None, image_attention_mask: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, interpolate_pos_encoding: Optional[bool] = False, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[Tuple, IdeficsBaseModelOutputWithPast]: device = input_ids.device if input_ids is not None else inputs_embeds.device 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 ) use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") if self.gradient_checkpointing and self.training and use_cache: logger.warning_once( "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." ) use_cache = False if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) # kept for BC (non `Cache` `past_key_values` inputs) return_legacy_cache = False if use_cache and not isinstance(past_key_values, Cache): return_legacy_cache = True if past_key_values is None: past_key_values = DynamicCache() else: past_key_values = DynamicCache.from_legacy_cache(past_key_values) logger.warning_once( "We detected that you are passing `past_key_values` as a tuple of tuples. This is deprecated and " "will be removed in v4.47. Please convert your cache or use an appropriate `Cache` class " "(https://huggingface.co/docs/transformers/kv_cache#legacy-cache-format)" ) batch_size, seq_length, _ = inputs_embeds.shape past_key_values_length = past_key_values.get_seq_length() if past_key_values is not None else 0 seq_length_with_past = seq_length + past_key_values_length if cache_position is None: cache_position = torch.arange( past_key_values_length, past_key_values_length + inputs_embeds.shape[1], device=inputs_embeds.device ) if attention_mask is not None and position_ids is None: # create position_ids on the fly for batch generation position_ids = attention_mask.long().cumsum(-1) - 1 position_ids.masked_fill_(attention_mask == 0, 1) position_ids = position_ids[:, -seq_length:] elif position_ids is None: position_ids = cache_position.unsqueeze(0) if sum([x is None for x in [pixel_values, image_encoder_embeddings, perceiver_embeddings]]) != 2: raise ValueError( "Exactly 1 of pixel_values, image_encoder_embeddings or perceiver_embeddings has to be not-None." ) elif pixel_values is not None: pixel_values = pixel_values.to(dtype=self.dtype, device=device) # fp16 compatibility batch_size, num_images = pixel_values.shape[:2] pixel_values = pixel_values.contiguous().view(batch_size * num_images, *pixel_values.shape[2:]) # Get sequence from the vision encoder image_hidden_states = self.vision_model( pixel_values=pixel_values, interpolate_pos_encoding=interpolate_pos_encoding ).last_hidden_state elif image_encoder_embeddings is not None: batch_size, num_images, image_seq_len, image_hidden_size = image_encoder_embeddings.size() image_hidden_states = image_encoder_embeddings.to(dtype=self.dtype, device=device) image_hidden_states = image_hidden_states.view(batch_size * num_images, image_seq_len, image_hidden_size) if self.config.use_resampler: if perceiver_embeddings is None: perceiver_embeddings = self.perceiver_resampler(image_hidden_states) image_seq_len, image_hidden_size = perceiver_embeddings.size(1), perceiver_embeddings.size(2) else: batch_size, num_images, image_seq_len, image_hidden_size = perceiver_embeddings.size() image_hidden_states = perceiver_embeddings elif perceiver_embeddings is None: image_seq_len, image_hidden_size = image_hidden_states.size(1), image_hidden_states.size(2) else: raise ValueError("If `perceiver_embeddings` are passed, use_resampler should be True") image_hidden_states = image_hidden_states.view(batch_size, num_images * image_seq_len, image_hidden_size) # # Hack to use the model in full language modeling mode # image_attention_mask = torch.zeros(batch_size, seq_length, 1, dtype=torch.long, device=image_hidden_states.device) # Make image_attention_mask compatible with hidden states text_seq_len = image_attention_mask.size(1) image_attention_mask = image_attention_mask.unsqueeze(-1) image_attention_mask = image_attention_mask.repeat(1, 1, 1, image_seq_len) image_attention_mask = image_attention_mask.view(batch_size, text_seq_len, num_images * image_seq_len) if image_hidden_states is not None: image_batch_size, image_sequence_length, _ = image_hidden_states.size() image_hidden_shape = (image_batch_size, image_sequence_length) if image_attention_mask is None: image_attention_mask = torch.ones(image_hidden_shape, device=device) image_attention_mask = self.invert_attention_mask(image_attention_mask) else: image_attention_mask = None # cross_attention_gate: # For any tokens attending to no images, the hidden_states comming out of the cross-attention should be zeroed-out. # `image_attention_mask` has shape [bsz, 1, num_images, hidden_size] with elements equal to either 0.0 or a very negative number. # If any of the elements are 0.0, then the token is attending to at least one image and the gate value is 1. Otherwise the gate value is 0. # `cross_attention_gate` has shape [bsz, seq_len] with elements equal to either 0.0 or 1.0. cross_attention_gate = ((((image_attention_mask == 0.0).any(dim=-1)).to(dtype=self.dtype)).squeeze(dim=1)).to( device ) # embed positions if attention_mask is None: attention_mask = torch.ones( (batch_size, seq_length_with_past), dtype=torch.bool, device=inputs_embeds.device ) attention_mask = self._update_causal_mask( attention_mask, inputs_embeds, cache_position, past_key_values, output_attentions ) hidden_states = inputs_embeds # decoder layers all_hidden_states = () if output_hidden_states else None all_self_attns = () if output_attentions else None next_decoder_cache = None for idx, decoder_layer in enumerate(self.layers): if output_hidden_states: all_hidden_states += (hidden_states,) def vblock( main_block, hidden_states, attention_mask, position_ids, past_key_value, image_hidden_states, image_attention_mask, cross_attention_gate, output_attentions, use_cache, layer_idx, cross_layer_interval, gated_cross_attn_layers, cache_position, ): # TODO(ls): Add cross attention values to respective lists if layer_idx % cross_layer_interval == 0: xblock = gated_cross_attn_layers[layer_idx // cross_layer_interval] outputs = xblock( hidden_states, attention_mask=attention_mask, image_hidden_states=image_hidden_states, image_attention_mask=image_attention_mask, cross_attention_gate=cross_attention_gate, output_attentions=output_attentions, use_cache=use_cache, past_key_value=None, # not implemented ) hidden_states = outputs[0] layer_outputs = main_block( hidden_states, attention_mask=attention_mask, position_ids=position_ids, past_key_value=past_key_value, output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, ) return layer_outputs if self.gradient_checkpointing and self.training: past_key_values = None if use_cache: logger.warning_once( "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..." ) use_cache = False layer_outputs = self._gradient_checkpointing_func( vblock, decoder_layer, hidden_states, attention_mask, position_ids, past_key_values, image_hidden_states, image_attention_mask, cross_attention_gate, output_attentions, use_cache, idx, self.cross_layer_interval, self.gated_cross_attn_layers, cache_position, ) else: layer_outputs = vblock( decoder_layer, hidden_states, attention_mask=attention_mask, position_ids=position_ids, past_key_value=past_key_values, image_hidden_states=image_hidden_states, image_attention_mask=image_attention_mask, cross_attention_gate=cross_attention_gate, output_attentions=output_attentions, use_cache=use_cache, layer_idx=idx, cross_layer_interval=self.cross_layer_interval, gated_cross_attn_layers=self.gated_cross_attn_layers, cache_position=cache_position, ) hidden_states = layer_outputs[0] if use_cache: next_decoder_cache = layer_outputs[2 if output_attentions else 1] if output_attentions: all_self_attns += (layer_outputs[1],) hidden_states = self.norm(hidden_states) # add hidden states from the last decoder layer if output_hidden_states: all_hidden_states += (hidden_states,) next_cache = next_decoder_cache if use_cache else None if return_legacy_cache: next_cache = next_cache.to_legacy_cache() image_hidden_states = image_hidden_states.view(batch_size, num_images, image_seq_len, image_hidden_size) if not return_dict: return tuple( v for v in [hidden_states, next_cache, all_hidden_states, all_self_attns, image_hidden_states] if v is not None ) return IdeficsBaseModelOutputWithPast( last_hidden_state=hidden_states, past_key_values=next_cache, hidden_states=all_hidden_states, attentions=all_self_attns, image_hidden_states=image_hidden_states, ) # Copied from transformers.models.llama.modeling_llama.LlamaModel._update_causal_mask def _update_causal_mask( self, attention_mask: torch.Tensor, input_tensor: torch.Tensor, cache_position: torch.Tensor, past_key_values: Cache, output_attentions: bool = False, ): if self.config._attn_implementation == "flash_attention_2": if attention_mask is not None and (attention_mask == 0.0).any(): return attention_mask return None if self.config._attn_implementation == "flex_attention": if isinstance(attention_mask, torch.Tensor): attention_mask = make_flex_block_causal_mask(attention_mask) if isinstance(attention_mask, BlockMask): return attention_mask # For SDPA, when possible, we will rely on its `is_causal` argument instead of its `attn_mask` argument, in # order to dispatch on Flash Attention 2. This feature is not compatible with static cache, as SDPA will fail # to infer the attention mask. past_seen_tokens = past_key_values.get_seq_length() if past_key_values is not None else 0 using_static_cache = isinstance(past_key_values, StaticCache) # When output attentions is True, sdpa implementation's forward method calls the eager implementation's forward if self.config._attn_implementation == "sdpa" and not using_static_cache and not output_attentions: if AttentionMaskConverter._ignore_causal_mask_sdpa( attention_mask, inputs_embeds=input_tensor, past_key_values_length=past_seen_tokens, is_training=self.training, ): return None dtype, device = input_tensor.dtype, input_tensor.device sequence_length = input_tensor.shape[1] if using_static_cache: target_length = past_key_values.get_max_cache_shape() else: target_length = ( attention_mask.shape[-1] if isinstance(attention_mask, torch.Tensor) else past_seen_tokens + sequence_length + 1 ) # In case the provided `attention` mask is 2D, we generate a causal mask here (4D). causal_mask = self._prepare_4d_causal_attention_mask_with_cache_position( attention_mask, sequence_length=sequence_length, target_length=target_length, dtype=dtype, device=device, cache_position=cache_position, batch_size=input_tensor.shape[0], ) if ( self.config._attn_implementation == "sdpa" and attention_mask is not None and attention_mask.device.type in ["cuda", "xpu"] and not output_attentions ): # Attend to all tokens in fully masked rows in the causal_mask, for example the relevant first rows when # using left padding. This is required by F.scaled_dot_product_attention memory-efficient attention path. # Details: https://github.com/pytorch/pytorch/issues/110213 min_dtype = torch.finfo(dtype).min causal_mask = AttentionMaskConverter._unmask_unattended(causal_mask, min_dtype) return causal_mask @staticmethod # Copied from transformers.models.llama.modeling_llama.LlamaModel._prepare_4d_causal_attention_mask_with_cache_position def _prepare_4d_causal_attention_mask_with_cache_position( attention_mask: torch.Tensor, sequence_length: int, target_length: int, dtype: torch.dtype, device: torch.device, cache_position: torch.Tensor, batch_size: int, **kwargs, ): """ Creates a causal 4D mask of shape `(batch_size, 1, query_length, key_value_length)` from a 2D mask of shape `(batch_size, key_value_length)`, or if the input `attention_mask` is already 4D, do nothing. Args: attention_mask (`torch.Tensor`): A 2D attention mask of shape `(batch_size, key_value_length)` or a 4D attention mask of shape `(batch_size, 1, query_length, key_value_length)`. sequence_length (`int`): The sequence length being processed. target_length (`int`): The target length: when generating with static cache, the mask should be as long as the static cache, to account for the 0 padding, the part of the cache that is not filled yet. dtype (`torch.dtype`): The dtype to use for the 4D attention mask. device (`torch.device`): The device to place the 4D attention mask on. cache_position (`torch.Tensor`): Indices depicting the position of the input sequence tokens in the sequence. batch_size (`torch.Tensor`): Batch size. """ if attention_mask is not None and attention_mask.dim() == 4: # In this case we assume that the mask comes already in inverted form and requires no inversion or slicing. causal_mask = attention_mask else: min_dtype = torch.finfo(dtype).min causal_mask = torch.full( (sequence_length, target_length), fill_value=min_dtype, dtype=dtype, device=device ) if sequence_length != 1: causal_mask = torch.triu(causal_mask, diagonal=1) causal_mask *= torch.arange(target_length, device=device) > cache_position.reshape(-1, 1) causal_mask = causal_mask[None, None, :, :].expand(batch_size, 1, -1, -1) if attention_mask is not None: causal_mask = causal_mask.clone() # copy to contiguous memory for in-place edit mask_length = attention_mask.shape[-1] padding_mask = causal_mask[:, :, :, :mask_length] + attention_mask[:, None, None, :].to( causal_mask.device ) padding_mask = padding_mask == 0 causal_mask[:, :, :, :mask_length] = causal_mask[:, :, :, :mask_length].masked_fill( padding_mask, min_dtype ) return causal_mask class IdeficsForVisionText2Text(IdeficsPreTrainedModel, GenerationMixin): _keys_to_ignore_on_load_missing = [r"lm_head.weight"] _tied_weights_keys = ["model.embed_tokens.weight", "lm_head.weight"] def __init__(self, config, vision_model=None): super().__init__(config) self.model = IdeficsModel(config) self.lm_head = IdeficsDecoupledLinear( in_features=config.hidden_size, out_features=config.vocab_size, out_additional_features=config.additional_vocab_size, bias=False, partially_freeze=config.freeze_lm_head, ) # Initialize weights and apply final processing self.post_init() def get_input_embeddings(self): return self.model.embed_tokens def set_input_embeddings(self, value): self.model.embed_tokens = value def get_output_embeddings(self): return self.lm_head def set_output_embeddings(self, new_embeddings): self.lm_head = new_embeddings def set_decoder(self, decoder): self.model = decoder def get_decoder(self): return self.model def tie_weights(self): """ Overwrite `transformers.modeling_utils.PreTrainedModel.tie_weights` to handle the case of IdeficsDecoupledLinear and IdeficsDecoupledEmbedding. """ output_embeddings = self.get_output_embeddings() input_embeddings = self.get_input_embeddings() if getattr(self.config, "tie_word_embeddings", True): output_embeddings.weight = input_embeddings.weight if input_embeddings.num_additional_embeddings > 0: assert output_embeddings.out_additional_features == input_embeddings.num_additional_embeddings output_embeddings.additional_fc.weight = input_embeddings.additional_embedding.weight if hasattr(output_embeddings, "out_features") and hasattr(input_embeddings, "num_embeddings"): output_embeddings.out_features = input_embeddings.num_embeddings if hasattr(output_embeddings, "out_additional_features") and hasattr( input_embeddings, "num_additional_embeddings" ): output_embeddings.out_additional_features = input_embeddings.num_additional_embeddings @add_start_docstrings_to_model_forward(LLAMA_INPUTS_DOCSTRING) @replace_return_docstrings(output_type=IdeficsCausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC) def forward( self, input_ids: Optional[torch.LongTensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[List[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, pixel_values: Optional[torch.FloatTensor] = None, image_encoder_embeddings: Optional[torch.FloatTensor] = None, perceiver_embeddings: Optional[torch.FloatTensor] = None, image_attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, interpolate_pos_encoding: Optional[bool] = False, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[Tuple, IdeficsCausalLMOutputWithPast]: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): Labels for computing the masked language modeling loss. Indices should either be in `[0, ..., config.vocab_size]` or -100 (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]`. Returns: Example: ```python >>> from transformers import AutoProcessor, IdeficsForVisionText2Text >>> model = IdeficsForVisionText2Text.from_pretrained("HuggingFaceM4/idefics-9b") >>> processor = AutoProcessor.from_pretrained("HuggingFaceM4/idefics-9b") >>> dogs_image_url_1 = "https://huggingface.co/datasets/hf-internal-testing/fixtures_nlvr2/raw/main/image1.jpeg" >>> dogs_image_url_2 = "https://huggingface.co/datasets/hf-internal-testing/fixtures_nlvr2/raw/main/image2.jpeg" >>> prompts = [ ... [ ... "User:", ... dogs_image_url_1, ... "Describe this image.\nAssistant: An image of two dogs.\n", ... "User:", ... dogs_image_url_2, ... "Describe this image.\nAssistant:", ... ] ... ] >>> inputs = processor(prompts, return_tensors="pt") >>> generate_ids = model.generate(**inputs, max_new_tokens=6) >>> processor.batch_decode(generate_ids, skip_special_tokens=True) ```""" 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 # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs = self.model( input_ids=input_ids, attention_mask=attention_mask, position_ids=position_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, pixel_values=pixel_values, image_encoder_embeddings=image_encoder_embeddings, perceiver_embeddings=perceiver_embeddings, image_attention_mask=image_attention_mask, use_cache=use_cache, output_attentions=output_attentions, output_hidden_states=output_hidden_states, interpolate_pos_encoding=interpolate_pos_encoding, return_dict=return_dict, cache_position=cache_position, ) hidden_states = outputs[0] logits = self.lm_head(hidden_states) loss = None if labels is not None: labels = labels.to(logits.device) # Shift so that tokens < n predict n if attention_mask is not None: # we use the input attention mask to shift the logits and labels, because it is 2D. # we also crop attn mask in case it is longer, which happens in PrefixTuning with peft shift_attention_mask = attention_mask[:, -(logits.shape[1] - 1) :].to(logits.device) shift_logits = logits[..., :-1, :][shift_attention_mask != 0].contiguous() shift_labels = labels[..., 1:][shift_attention_mask != 0].contiguous() else: shift_logits = logits[..., :-1, :].contiguous() shift_labels = labels[..., 1:].contiguous() # Flatten the tokens loss_fct = CrossEntropyLoss() loss = loss_fct(shift_logits.view(-1, shift_logits.size(-1)), shift_labels.view(-1)) if not return_dict: output = (logits,) + outputs[1:] return (loss,) + output if loss is not None else output return IdeficsCausalLMOutputWithPast( loss=loss, logits=logits, past_key_values=outputs.past_key_values, hidden_states=outputs.hidden_states, attentions=outputs.attentions, image_hidden_states=outputs.image_hidden_states, ) def prepare_inputs_for_generation( self, input_ids, attention_mask=None, position_ids=None, inputs_embeds=None, past_key_values=None, cache_position=None, pixel_values=None, image_hidden_states=None, image_attention_mask=None, use_cache=None, **kwargs, ): # Overwritten -- custom processing based on `config.use_resampler` images_kwargs = {} if image_hidden_states is not None: if self.config.use_resampler: images_kwargs["perceiver_embeddings"] = image_hidden_states else: images_kwargs["image_encoder_embeddings"] = image_hidden_states else: images_kwargs["pixel_values"] = pixel_values images_kwargs["interpolate_pos_encoding"] = kwargs.pop("interpolate_pos_encoding", False) model_inputs = super().prepare_inputs_for_generation( input_ids, past_key_values=past_key_values, attention_mask=attention_mask, inputs_embeds=inputs_embeds, cache_position=cache_position, position_ids=position_ids, use_cache=use_cache, image_attention_mask=image_attention_mask, **images_kwargs, **kwargs, ) if image_attention_mask is not None and inputs_embeds is None: seq_length = model_inputs["input_ids"].shape[1] model_inputs["image_attention_mask"] = image_attention_mask[:, -seq_length:] return model_inputs def _update_model_kwargs_for_generation( self, outputs: ModelOutput, model_kwargs: Dict[str, Any], is_encoder_decoder: bool = False, **kwargs, ) -> Dict[str, Any]: model_kwargs = super()._update_model_kwargs_for_generation( outputs, model_kwargs, is_encoder_decoder, **kwargs, ) if "image_attention_mask" in model_kwargs: image_attention_mask = model_kwargs["image_attention_mask"] last_mask = image_attention_mask[:, -1, :].unsqueeze(1) if model_kwargs.get("use_cache", True): model_kwargs["image_attention_mask"] = last_mask else: model_kwargs["image_attention_mask"] = torch.cat([image_attention_mask, last_mask], dim=1) # Get the precomputed image_hidden_states model_kwargs["image_hidden_states"] = outputs.image_hidden_states return model_kwargs @staticmethod def _reorder_cache(past, beam_idx): reordered_past = () for layer_past in past: reordered_past += (tuple(past_state.index_select(0, beam_idx) for past_state in layer_past),) return reordered_past __all__ = ["IdeficsForVisionText2Text", "IdeficsModel", "IdeficsPreTrainedModel"]