modeling_fuyu.py

# coding=utf-8
# Copyright 2023 HuggingFace Inc. team. All rights reserved.
#
# 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 Fuyu model."""
from typing import List, Optional, Tuple, Union

import torch
import torch.utils.checkpoint
from torch import nn

from ...modeling_outputs import BaseModelOutputWithPast
from ...modeling_utils import PreTrainedModel
from ...models.auto.modeling_auto import AutoModelForCausalLM
from ...utils import add_start_docstrings, add_start_docstrings_to_model_forward, logging
from .configuration_fuyu import FuyuConfig


logger = logging.get_logger(__name__)

_CONFIG_FOR_DOC = "FuyuConfig"


FUYU_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 ([`FuyuConfig`]):
            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 Fuyu Model outputting raw hidden-states without any specific head on top.",
    FUYU_START_DOCSTRING,
)
class FuyuPreTrainedModel(PreTrainedModel):
    config_class = FuyuConfig
    base_model_prefix = "fuyu"
    supports_gradient_checkpointing = True
    _no_split_modules = []
    _skip_keys_device_placement = "past_key_values"

    def _init_weights(self, module):
        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_()


FUYU_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.
"""


@add_start_docstrings(
    "The bare Fuyu Model outputting raw hidden-states without any specific head on top.",
    FUYU_START_DOCSTRING,
)
class FuyuForCausalLM(FuyuPreTrainedModel):
    """
    Transformer decoder consisting of *config.num_hidden_layers* layers. Each layer is a [`FuyuDecoderLayer`]

    Args:
        config: FuyuConfig
    """

    def __init__(self, config: FuyuConfig):
        super().__init__(config)
        self.padding_idx = config.pad_token_id
        self.vocab_size = config.vocab_size
        self.language_model = AutoModelForCausalLM.from_config(config.text_config)

        self.vision_embed_tokens = nn.Linear(
            config.patch_size * config.patch_size * config.num_channels, config.hidden_size
        )

        self.gradient_checkpointing = False
        # Initialize weights and apply final processing
        self.post_init()

    def get_input_embeddings(self):
        return self.language_model.get_input_embeddings()

    def set_input_embeddings(self, value):
        self.language_model.set_input_embeddings(value)

    def gather_continuous_embeddings(
        self,
        word_embeddings: torch.Tensor,
        continuous_embeddings: List[torch.Tensor],
        image_patch_input_indices: torch.Tensor,
    ) -> torch.Tensor:
        """This function places the continuous_embeddings into the word_embeddings at the locations
        indicated by image_patch_input_indices. Different batch elements can have different numbers of continuous
        embeddings.

        Args:
            word_embeddings: Tensor of word embeddings. Shape: [b, s, h]
            continuous_embeddings:
                Tensor of continuous embeddings. The length of the list is the batch size. Each entry is
            shape [num_image_embeddings, hidden], and num_image_embeddings needs to match the number of non-negative
            indices in image_patch_input_indices for that batch element.
            image_patch_input_indices: Tensor of indices of the image patches in the input_ids tensor. Shape: [b, s]
        """
        if not (word_embeddings.shape[0] == len(continuous_embeddings)):
            raise ValueError(
                f"Batch sizes must match! Got {len(continuous_embeddings)=} and {word_embeddings.shape[0]=}"
            )

        output_embeddings = word_embeddings.clone()
        for batch_idx in range(word_embeddings.shape[0]):
            # First, find the positions of all the non-negative values in image_patch_input_indices, those are the
            # positions in word_embeddings that we want to replace with content from continuous_embeddings.
            dst_indices = torch.nonzero(image_patch_input_indices[batch_idx] >= 0, as_tuple=True)[0]
            # Next look up those indices in image_patch_input_indices to find the indices in continuous_embeddings that we
            # want to use to replace the values in word_embeddings.
            src_indices = image_patch_input_indices[batch_idx][dst_indices]
            # Check if we have more indices than embeddings. Note that we could have fewer indices if images got truncated.
            if src_indices.shape[0] > continuous_embeddings[batch_idx].shape[0]:
                raise ValueError(
                    f"Number of continuous embeddings {continuous_embeddings[batch_idx].shape=} does not match "
                    f"number of continuous token ids {src_indices.shape=} in batch element {batch_idx}."
                )
            output_embeddings[batch_idx, dst_indices] = continuous_embeddings[batch_idx][src_indices]
        return output_embeddings

    @add_start_docstrings_to_model_forward(FUYU_INPUTS_DOCSTRING)
    def forward(
        self,
        input_ids: torch.LongTensor = None,
        image_patches: torch.Tensor = None,  # [batch_size, num_total_patches, patch_size_ x patch_size x num_channels ]
        image_patches_indices: torch.Tensor = 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,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple, BaseModelOutputWithPast]:
        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

        # retrieve input_ids and inputs_embeds
        if input_ids is not None and inputs_embeds is not None:
            raise ValueError("You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time")
        elif input_ids is not None:
            batch_size, seq_length = input_ids.shape
        elif inputs_embeds is not None:
            batch_size, seq_length, _ = inputs_embeds.shape
        else:
            raise ValueError("You have to specify either decoder_input_ids or decoder_inputs_embeds")

        seq_length_with_past = seq_length
        past_key_values_length = 0

        if past_key_values is not None:
            past_key_values_length = past_key_values[0][0].shape[2]
            seq_length_with_past = seq_length_with_past + past_key_values_length

        if position_ids is None:
            device = input_ids.device if input_ids is not None else inputs_embeds.device
            position_ids = torch.arange(
                past_key_values_length, seq_length + past_key_values_length, dtype=torch.long, device=device
            )
            position_ids = position_ids.unsqueeze(0)

        if inputs_embeds is None:
            inputs_embeds = self.language_model.get_input_embeddings()(input_ids)
            if image_patches is not None and past_key_values is None:
                patch_embeddings = [
                    self.vision_embed_tokens(patch.to(self.vision_embed_tokens.weight.dtype)).squeeze(0)
                    for patch in image_patches
                ]
                inputs_embeds = self.gather_continuous_embeddings(
                    word_embeddings=inputs_embeds,
                    continuous_embeddings=patch_embeddings,
                    image_patch_input_indices=image_patches_indices,
                )

        outputs = self.language_model(
            inputs_embeds=inputs_embeds,
            attention_mask=attention_mask,
            position_ids=position_ids,
            past_key_values=past_key_values,
            output_attentions=output_attentions,
            use_cache=use_cache,
        )
        if not return_dict:
            return tuple(v for v in outputs if v is not None)
        return outputs

    def prepare_inputs_for_generation(
        self,
        input_ids,
        past_key_values=None,
        attention_mask=None,
        inputs_embeds=None,
        image_patches=None,
        image_patches_indices=None,
        **kwargs,
    ):
        if past_key_values:
            input_ids = input_ids[:, -1:]

        position_ids = kwargs.get("position_ids", None)
        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)
            if past_key_values:
                position_ids = position_ids[:, -1].unsqueeze(-1)

        # if `inputs_embeds` are passed, we only want to use them in the 1st generation step
        if inputs_embeds is not None and past_key_values is None:
            model_inputs = {"inputs_embeds": inputs_embeds}
        else:
            model_inputs = {"input_ids": input_ids}

        if image_patches_indices is not None:
            model_inputs["image_patches_indices"] = image_patches_indices

        model_inputs.update(
            {
                "position_ids": position_ids,
                "past_key_values": past_key_values,
                "use_cache": kwargs.get("use_cache"),
                "attention_mask": attention_mask,
                "image_patches_indices": image_patches_indices if past_key_values is None else None,
                "image_patches": image_patches if past_key_values is None else None,
            }
        )
        return model_inputs