llava/model/llava_arch.py

#    Copyright 2023 Haotian Liu
#
#    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.


from abc import ABC, abstractmethod

import torch
import torch.nn as nn
import transformers

from llava.model.utils import get_w

from .multimodal_encoder.builder import build_vision_tower

from llava.constants import (
    GROUND_TOKEN,
    IGNORE_INDEX,
    IMAGE_TOKEN_INDEX,
    DEFAULT_IMAGE_PATCH_TOKEN,
    DEFAULT_IM_START_TOKEN,
    DEFAULT_IM_END_TOKEN,
    PROFILE_RUNTIME,
)

import time
from transformers.utils import logging

logger = logging.get_logger("transformers")


class LlavaMetaModel:
    def __init__(self, config):
        super(LlavaMetaModel, self).__init__(config)

        if hasattr(config, "mm_vision_tower"):
            self.vision_tower = build_vision_tower(config, delay_load=True)
            if self.vision_tower is not None:
                self.mm_projector = nn.Linear(
                    self.vision_tower.hidden_size, config.hidden_size
                )  # placeholder, this will be re-initialized later in initialize_vision_modules()

    def get_vision_tower(self):
        vision_tower = getattr(self, "vision_tower", None)
        if type(vision_tower) is list:
            vision_tower = vision_tower[0]
        return vision_tower

    def initialize_vision_modules(self, model_args, fsdp=None):
        vision_tower = model_args.vision_tower
        mm_vision_select_layer = model_args.mm_vision_select_layer
        mm_vision_select_feature = model_args.mm_vision_select_feature
        pretrain_mm_mlp_adapter = model_args.pretrain_mm_mlp_adapter
        pretrain_vision_tower = model_args.pretrain_vision_tower

        self.config.mm_vision_tower = vision_tower

        if hasattr(self, "vision_tower"):
            del self.vision_tower
            torch.cuda.empty_cache()
        vision_tower = build_vision_tower(model_args)

        if vision_tower is None:
            return

        if fsdp is not None and len(fsdp) > 0:
            self.vision_tower = [vision_tower]
        else:
            self.vision_tower = vision_tower

        # add these model args to HF config so that they can be saved (used for loading checkpoint)
        self.config.use_mm_proj = True
        self.config.mm_hidden_size = vision_tower.hidden_size
        self.config.mm_vision_select_layer = mm_vision_select_layer
        self.config.mm_vision_select_feature = mm_vision_select_feature

        self.config.num_points = model_args.num_points
        self.config.feature_dim = model_args.feature_dim
        self.config.num_latents = model_args.num_latents
        self.config.d_latents = model_args.d_latents
        self.config.num_cross_attention_heads = model_args.num_cross_attention_heads
        self.config.position_encoding_type = model_args.position_encoding_type

        if hasattr(self, "mm_projector"):
            del self.mm_projector
            torch.cuda.empty_cache()
        self.mm_projector = nn.Linear(self.config.mm_hidden_size, self.config.hidden_size)

        if pretrain_mm_mlp_adapter is not None:
            mm_projector_weights = torch.load(pretrain_mm_mlp_adapter, map_location="cpu")

            self.mm_projector.load_state_dict(get_w(mm_projector_weights, "mm_projector"))

        torch.cuda.empty_cache()


class LlavaMetaForCausalLM(ABC):
    @abstractmethod
    def get_model(self):
        pass

    def get_vision_tower(self):
        return self.get_model().get_vision_tower()

    def encode_images(self, images):
        vision_features_before_mm_projection = self.get_model().get_vision_tower()(
            images
        )  # for minkowski, the output of this step will be float32
        vision_features_before_mm_projection = vision_features_before_mm_projection.to(
            dtype=self.dtype
        )  # convert back to the dtype of the LLM (bfloat16 in most cases), no-op if the dtype is already the same
        vision_features = self.get_model().mm_projector(
            vision_features_before_mm_projection
        )  # vision_features and mm_projector are both float32

        return vision_features, vision_features_before_mm_projection

    def prepare_inputs_labels_for_multimodal(
        self, input_ids, attention_mask, past_key_values, labels, images
    ):
        vision_tower = self.get_vision_tower()
        if vision_tower is None or images is None or input_ids.shape[1] == 1:
            if (
                past_key_values is not None
                and vision_tower is not None
                and images is not None
                and input_ids.shape[1] == 1
            ):
                attention_mask = torch.ones(
                    (attention_mask.shape[0], past_key_values[-1][-1].shape[-2] + 1),
                    dtype=attention_mask.dtype,
                    device=attention_mask.device,
                )
            vision_features_before_mm_projection = images
            return (
                input_ids,
                attention_mask,
                past_key_values,
                None,
                labels,
                vision_features_before_mm_projection,
            )

        start_time_encode_images = time.time()
        if not isinstance(images, SparseTensor) and (type(images) is list or images.ndim == 5):
            concat_images = torch.cat([image for image in images], dim=0)
            vision_features, vision_features_before_mm_projection = self.encode_images(
                concat_images
            )
            split_sizes = [image.shape[0] for image in images]
            vision_features = torch.split(vision_features, split_sizes, dim=0)
            vision_features = [x.flatten(0, 1) for x in vision_features]
        else:
            vision_features, vision_features_before_mm_projection = self.encode_images(images)
        if PROFILE_RUNTIME:
            logger.info(f"Time to encode images: {time.time() - start_time_encode_images}")

        start_time_for_loop = time.time()
        new_input_embeds = []
        new_labels = [] if labels is not None else None
        cur_image_idx = 0
        for batch_idx, cur_input_ids in enumerate(input_ids):
            if (cur_input_ids == IMAGE_TOKEN_INDEX).sum() == 0:
                # multimodal LLM, but the current sample is not multimodal
                cur_input_embeds = self.get_model().embed_tokens(cur_input_ids)
                cur_input_embeds = (
                    cur_input_embeds
                    + (0.0 * self.get_model().mm_projector(vision_tower.dummy_feature)).sum()
                )
                new_input_embeds.append(cur_input_embeds)
                if labels is not None:
                    new_labels.append(labels[batch_idx])
                cur_image_idx += 1
                continue
            image_token_indices = torch.where(cur_input_ids == IMAGE_TOKEN_INDEX)[0]
            cur_new_input_embeds = []
            if labels is not None:
                cur_labels = labels[batch_idx]
                cur_new_labels = []
                assert cur_labels.shape == cur_input_ids.shape

            # The following while loop looks for all image tokens in the current sentence
            # and replace them with the corresponding image features.
            while image_token_indices.numel() > 0:
                cur_vision_features = vision_features[cur_image_idx]
                image_token_start = image_token_indices[0]
                if getattr(self.config, "tune_mm_mlp_adapter", False) and getattr(
                    self.config, "mm_use_im_start_end", False
                ):
                    cur_new_input_embeds.append(
                        self.get_model()
                        .embed_tokens(cur_input_ids[: image_token_start - 1])
                        .detach()
                    )
                    cur_new_input_embeds.append(
                        self.get_model().embed_tokens(
                            cur_input_ids[image_token_start - 1 : image_token_start]
                        )
                    )
                    cur_new_input_embeds.append(cur_vision_features)
                    cur_new_input_embeds.append(
                        self.get_model().embed_tokens(
                            cur_input_ids[image_token_start + 1 : image_token_start + 2]
                        )
                    )
                    if labels is not None:
                        cur_new_labels.append(cur_labels[:image_token_start])
                        cur_new_labels.append(
                            torch.full(
                                (cur_vision_features.shape[0],),
                                IGNORE_INDEX,
                                device=labels.device,
                                dtype=labels.dtype,
                            )
                        )
                        cur_new_labels.append(cur_labels[image_token_start : image_token_start + 1])
                        cur_labels = cur_labels[image_token_start + 2 :]
                else:
                    cur_new_input_embeds.append(
                        self.get_model().embed_tokens(cur_input_ids[:image_token_start])
                    )
                    cur_new_input_embeds.append(cur_vision_features)
                    if labels is not None:
                        cur_new_labels.append(cur_labels[:image_token_start])
                        cur_new_labels.append(
                            torch.full(
                                (cur_vision_features.shape[0],),
                                IGNORE_INDEX,
                                device=labels.device,
                                dtype=labels.dtype,
                            )
                        )
                        cur_labels = cur_labels[image_token_start + 1 :]
                cur_image_idx += 1
                if getattr(self.config, "tune_mm_mlp_adapter", False) and getattr(
                    self.config, "mm_use_im_start_end", False
                ):
                    cur_input_ids = cur_input_ids[image_token_start + 2 :]
                else:
                    cur_input_ids = cur_input_ids[image_token_start + 1 :]
                image_token_indices = torch.where(cur_input_ids == IMAGE_TOKEN_INDEX)[0]
            if cur_input_ids.numel() > 0:
                if getattr(self.config, "tune_mm_mlp_adapter", False) and getattr(
                    self.config, "mm_use_im_start_end", False
                ):
                    cur_new_input_embeds.append(
                        self.get_model().embed_tokens(cur_input_ids).detach()
                    )
                else:
                    cur_new_input_embeds.append(self.get_model().embed_tokens(cur_input_ids))
                if labels is not None:
                    cur_new_labels.append(cur_labels)
            cur_new_input_embeds = [x.to(device=self.device) for x in cur_new_input_embeds]
            cur_new_input_embeds = torch.cat(cur_new_input_embeds, dim=0)
            new_input_embeds.append(cur_new_input_embeds)
            if labels is not None:
                cur_new_labels = torch.cat(cur_new_labels, dim=0)
                new_labels.append(cur_new_labels)
        if PROFILE_RUNTIME:
            logger.info(f"Time for loop: {time.time() - start_time_for_loop}")

        start_time_paddding = time.time()
        # pad all sentences in batch to the same length
        if any(x.shape != new_input_embeds[0].shape for x in new_input_embeds):
            max_len = max(x.shape[0] for x in new_input_embeds)

            new_input_embeds_align = []
            for cur_new_embed in new_input_embeds:
                cur_new_embed = torch.cat(
                    (
                        cur_new_embed,
                        torch.zeros(
                            (max_len - cur_new_embed.shape[0], cur_new_embed.shape[1]),
                            dtype=cur_new_embed.dtype,
                            device=cur_new_embed.device,
                        ),
                    ),
                    dim=0,
                )
                new_input_embeds_align.append(cur_new_embed)
            new_input_embeds = torch.stack(new_input_embeds_align, dim=0)

            if labels is not None:
                new_labels_align = []
                _new_labels = new_labels
                for cur_new_label in new_labels:
                    cur_new_label = torch.cat(
                        (
                            cur_new_label,
                            torch.full(
                                (max_len - cur_new_label.shape[0],),
                                IGNORE_INDEX,
                                dtype=cur_new_label.dtype,
                                device=cur_new_label.device,
                            ),
                        ),
                        dim=0,
                    )
                    new_labels_align.append(cur_new_label)
                new_labels = torch.stack(new_labels_align, dim=0)

            if attention_mask is not None:
                new_attention_mask = []
                for cur_attention_mask, cur_new_labels, cur_new_labels_align in zip(
                    attention_mask, _new_labels, new_labels
                ):
                    new_attn_mask_pad_left = torch.full(
                        (cur_new_labels.shape[0] - labels.shape[1],),
                        True,
                        dtype=attention_mask.dtype,
                        device=attention_mask.device,
                    )
                    new_attn_mask_pad_right = torch.full(
                        (cur_new_labels_align.shape[0] - cur_new_labels.shape[0],),
                        False,
                        dtype=attention_mask.dtype,
                        device=attention_mask.device,
                    )
                    cur_new_attention_mask = torch.cat(
                        (
                            new_attn_mask_pad_left,
                            cur_attention_mask,
                            new_attn_mask_pad_right,
                        ),
                        dim=0,
                    )
                    new_attention_mask.append(cur_new_attention_mask)
                attention_mask = torch.stack(new_attention_mask, dim=0)
                assert attention_mask.shape == new_labels.shape
        else:
            new_input_embeds = torch.stack(new_input_embeds, dim=0)
            if labels is not None:
                new_labels = torch.stack(new_labels, dim=0)

            if attention_mask is not None:
                new_attn_mask_pad_left = torch.full(
                    (
                        attention_mask.shape[0],
                        new_input_embeds.shape[1] - input_ids.shape[1],
                    ),
                    True,
                    dtype=attention_mask.dtype,
                    device=attention_mask.device,
                )
                attention_mask = torch.cat((new_attn_mask_pad_left, attention_mask), dim=1)
                assert attention_mask.shape == new_input_embeds.shape[:2]

        if PROFILE_RUNTIME:
            logger.info(f"Time padding: {time.time() - start_time_paddding}")
        return (
            None,
            attention_mask,
            past_key_values,
            new_input_embeds,
            new_labels,
            vision_features_before_mm_projection,
        )

    def initialize_vision_tokenizer(self, model_args, tokenizer):
        if model_args.mm_use_im_patch_token:
            tokenizer.add_tokens([DEFAULT_IMAGE_PATCH_TOKEN], special_tokens=True)
            self.resize_token_embeddings(len(tokenizer))

        if model_args.mm_use_im_start_end:
            (
                num_new_tokens,
                input_embeddings,
                output_embeddings,
            ) = self.add_special_tokens_and_resize_embeddings(
                special_tokens_list=[DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN],
                tokenizer=tokenizer,
            )

            if model_args.tune_mm_mlp_adapter:
                for p in self.get_input_embeddings().parameters():
                    p.requires_grad = True
                for p in self.get_output_embeddings().parameters():
                    p.requires_grad = False

            if model_args.pretrain_mm_mlp_adapter:
                mm_projector_weights = torch.load(
                    model_args.pretrain_mm_mlp_adapter, map_location="cpu"
                )
                embed_tokens_weight = mm_projector_weights["model.embed_tokens.weight"]
                assert num_new_tokens == 2
                if input_embeddings.shape == embed_tokens_weight.shape:
                    input_embeddings[-num_new_tokens:] = embed_tokens_weight[-num_new_tokens:]
                elif embed_tokens_weight.shape[0] == num_new_tokens:
                    input_embeddings[-num_new_tokens:] = embed_tokens_weight
                else:
                    raise ValueError(
                        f"Unexpected embed_tokens_weight shape. Pretrained: {embed_tokens_weight.shape}. Current: {input_embeddings.shape}. Numer of new tokens: {num_new_tokens}."
                    )
        elif model_args.mm_use_im_patch_token:
            if model_args.tune_mm_mlp_adapter:
                for p in self.get_input_embeddings().parameters():
                    p.requires_grad = False
                for p in self.get_output_embeddings().parameters():
                    p.requires_grad = False

        # add special tokens if bbox_tokenization_type is location_tokens
        if model_args.bbox_tokenization_type == "location_tokens":
            num_special_tokens = model_args.num_voxels_per_axis_for_location_tokens**3
            self.add_special_tokens_and_resize_embeddings(
                special_tokens_list=[f"<loc_{i}>" for i in range(num_special_tokens)],
                tokenizer=tokenizer,
            )
            for p in self.get_input_embeddings().parameters():
                p.requires_grad = True
            for p in self.get_output_embeddings().parameters():
                p.requires_grad = True
        elif model_args.bbox_tokenization_type == "ground_token":
            self.add_special_tokens_and_resize_embeddings(
                special_tokens_list=[GROUND_TOKEN], tokenizer=tokenizer
            )
            for p in self.get_input_embeddings().parameters():
                p.requires_grad = True
            for p in self.get_output_embeddings().parameters():
                p.requires_grad = True

        # add special token to input id mapping
        self.config.added_special_token_to_input_id = tokenizer.get_added_vocab()

    def add_special_tokens_and_resize_embeddings(
        self,
        special_tokens_list: list[str],
        tokenizer: transformers.PreTrainedTokenizer,
    ):
        num_new_tokens = tokenizer.add_tokens(special_tokens_list, special_tokens=True)
        self.resize_token_embeddings(len(tokenizer))

        if num_new_tokens > 0:
            input_embeddings = self.get_input_embeddings().weight.data
            output_embeddings = self.get_output_embeddings().weight.data

            input_embeddings_avg = input_embeddings[:-num_new_tokens].mean(dim=0, keepdim=True)
            output_embeddings_avg = output_embeddings[:-num_new_tokens].mean(dim=0, keepdim=True)

            input_embeddings[-num_new_tokens:] = input_embeddings_avg
            output_embeddings[-num_new_tokens:] = output_embeddings_avg

        return num_new_tokens, input_embeddings, output_embeddings