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

from .multimodal_encoder.builder import build_image_tower, build_video_tower
from .multimodal_projector.builder import build_vision_projector

from llava.constants import IGNORE_INDEX, X_TOKEN_INDEX, DEFAULT_X_PATCH_TOKEN, DEFAULT_X_START_TOKEN, DEFAULT_X_END_TOKEN


class LlavaMetaModel:

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

        if hasattr(config, "mm_image_tower"):
            self.image_tower = build_image_tower(config, delay_load=True)
            self.mm_projector = build_vision_projector(config)
        if hasattr(config, "mm_video_tower"):
            self.video_tower = build_video_tower(config, delay_load=True)
            self.mm_projector = build_vision_projector(config)

    def get_image_tower(self):
        image_tower = getattr(self, 'image_tower', None)
        if type(image_tower) is list:
            image_tower = image_tower[0]
        return image_tower

    def get_video_tower(self):
        video_tower = getattr(self, 'video_tower', None)
        if type(video_tower) is list:
            video_tower = video_tower[0]
        return video_tower

    def initialize_image_modules(self, model_args, fsdp=None):
        image_tower = model_args.image_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

        self.config.mm_image_tower = image_tower

        image_tower = build_image_tower(model_args)

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

        self.config.use_mm_proj = True
        self.config.mm_projector_type = getattr(model_args, 'mm_projector_type', 'linear')
        self.config.mm_hidden_size = image_tower.hidden_size
        self.config.mm_vision_select_layer = mm_vision_select_layer
        self.config.mm_vision_select_feature = mm_vision_select_feature

        self.mm_projector = build_vision_projector(self.config)

        if pretrain_mm_mlp_adapter is not None:
            mm_projector_weights = torch.load(pretrain_mm_mlp_adapter, map_location='cpu')
            def get_w(weights, keyword):
                return {k.split(keyword + '.')[1]: v for k, v in weights.items() if keyword in k}

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

    def initialize_video_modules(self, model_args, fsdp=None):
        video_tower = model_args.video_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

        self.config.mm_video_tower = video_tower

        video_tower = build_video_tower(model_args)

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

        self.config.use_mm_proj = True
        self.config.mm_projector_type = getattr(model_args, 'mm_projector_type', 'linear')
        self.config.mm_hidden_size = video_tower.hidden_size
        self.config.mm_vision_select_layer = mm_vision_select_layer
        self.config.mm_vision_select_feature = mm_vision_select_feature

        self.mm_projector = build_vision_projector(self.config)

        if pretrain_mm_mlp_adapter is not None:
            mm_projector_weights = torch.load(pretrain_mm_mlp_adapter, map_location='cpu')
            def get_w(weights, keyword):
                return {k.split(keyword + '.')[1]: v for k, v in weights.items() if keyword in k}

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

class LlavaMetaForCausalLM(ABC):

    @abstractmethod
    def get_model(self):
        pass

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

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

    def get_all_tower(self, keys):
        tower = {key: getattr(self, f'get_{key}_tower') for key in keys}
        return tower

    def encode_images(self, images):
        image_features = self.get_model().get_image_tower()(images)
        image_features = self.get_model().mm_projector(image_features)
        return image_features

    def encode_videos(self, videos):
        video_features = self.get_model().get_video_tower()(videos)
        video_features = self.get_model().mm_projector(video_features)
        return video_features
    #
    # 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)
    #         return input_ids, attention_mask, past_key_values, None, labels
    #
    #     if type(images) is list or images.ndim == 5:
    #         concat_images = torch.cat([image for image in images], dim=0)
    #         image_features = self.encode_images(concat_images)
    #         split_sizes = [image.shape[0] for image in images]
    #         image_features = torch.split(image_features, split_sizes, dim=0)
    #         image_features = [x.flatten(0, 1) for x in image_features]
    #     else:
    #         image_features = self.encode_images(images)
    #
    #     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
    #             # FIXME: this is a hacky fix, for deepspeed zero3 to work
    #             half_len = cur_input_ids.shape[0] // 2
    #             cur_image_features = image_features[cur_image_idx]
    #             cur_input_embeds_1 = self.get_model().embed_tokens(cur_input_ids[:half_len])
    #             cur_input_embeds_2 = self.get_model().embed_tokens(cur_input_ids[half_len:])
    #             cur_input_embeds = torch.cat([cur_input_embeds_1, cur_image_features[0:0], cur_input_embeds_2], dim=0)
    #             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]  # 把中间的imgtoken的位置找到
    #         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
    #         while image_token_indices.numel() > 0:
    #             cur_image_features = image_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_image_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_image_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]))  # imgtoken之前的text拿出来，好像都是模板套话
    #                 cur_new_input_embeds.append(cur_image_features)
    #                 if labels is not None:
    #                     cur_new_labels.append(cur_labels[:image_token_start])
    #                     cur_new_labels.append(torch.full((cur_image_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:]   # imgtoken之后的text拿出来，是真的question
    #             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]  # 前面text+图片+后面question
    #         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 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]
    #
    #     return None, attention_mask, past_key_values, new_input_embeds, new_labels
    #
    # 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 = tokenizer.add_tokens([DEFAULT_IM_START_TOKEN, DEFAULT_IM_END_TOKEN], 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
    #
    #         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

    def prepare_inputs_labels_for_multimodal(
        self, input_ids, attention_mask, past_key_values, labels, X_modalities
    ):
        '''
        X_modalities [
        [img_feature, img_feature, video_feature, audio_feature],
        ['image', 'image', 'video', 'audio']
        ]
        '''
        Xs, keys = X_modalities
        all_tower = self.get_all_tower(set(keys)) if len(keys) > 0 else None
        
        # print(2.5)
        if all_tower is None or X_modalities[0][0] is None or input_ids.shape[1] == 1:
            if past_key_values is not None and all_tower is not None and Xs 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)
            return input_ids, attention_mask, past_key_values, None, labels

        # if type(images) is list or images.ndim == 5:
        #     concat_images = torch.cat([image for image in images], dim=0)
        #     image_features = self.encode_images(concat_images)
        #     split_sizes = [image.shape[0] for image in images]
        #     image_features = torch.split(image_features, split_sizes, dim=0)
        #     image_features = [x.flatten(0, 1) for x in image_features]
        # else:
        print(keys)
        X_features = [getattr(self, f'encode_{key}s')(X.unsqueeze(0)) for X, key in zip(Xs, keys)]  # expand to get batchsize
        # X_features = []
        # # print(2.5, *[i.shape for i in Xs], keys)  
        # for X, key in zip(Xs, keys):
        #     temp_X = X.unsqueeze(0)
        #     # print(2.6)
        #     # fn = getattr(self, f'encode_{key}s') 
        #     if key == 'image':
        #         out = self.encode_images(temp_X)
        #         # print(2.65, 'image', out.shape)
        #     elif key == 'video':
        #         out = self.encode_videos(temp_X)
        #         # print(2.65, 'video', out.shape)
        #     else:
        #         raise NameError(f'{key}')
        #     # print(2.8, out.shape)
        #     X_features.append(out)
        X_features = [x.flatten(0, 1) for x in X_features]
        # print([[j, i.shape] for i, j in zip(X_features, keys)])


        new_input_embeds = []
        new_labels = [] if labels is not None else None
        cur_X_idx = 0
        # print(2.9, input_ids.shape)
        for batch_idx, cur_input_ids in enumerate(input_ids):
            # print(333333)
            if (torch.any(torch.stack([cur_input_ids == X_TOKEN_INDEX[key.upper()] for key in keys]), dim=0)).sum() == 0:
                # multimodal LLM, but the current sample is not multimodal
                # FIXME: this is a hacky fix, for deepspeed zero3 to work
                half_len = cur_input_ids.shape[0] // 2
                cur_X_features = X_features[cur_X_idx]
                cur_input_embeds_1 = self.get_model().embed_tokens(cur_input_ids[:half_len])
                cur_input_embeds_2 = self.get_model().embed_tokens(cur_input_ids[half_len:])
                cur_input_embeds = torch.cat([cur_input_embeds_1, cur_X_features[0:0], cur_input_embeds_2], dim=0)
                new_input_embeds.append(cur_input_embeds)
                if labels is not None:
                    new_labels.append(labels[batch_idx])
                cur_X_idx += 1 ############## 注意这里跳过了，如果一个sample是一个modal，那么就跳过1个全zero的modal，如果一个sample对应多个modal，这里的训练逻辑不对！！！
                ###### 但似乎不影响1个sample的inference
                ###### 一个text对应视频和图片，直接走下边了。只有1个text，传入none或者1个/2个全zero都无所谓，反正没有下一个数据了。
                continue
            X_token_indices = torch.where(torch.any(torch.stack([cur_input_ids == X_TOKEN_INDEX[key.upper()] for key in keys]), dim=0))[0]  # 把中间的imgtoken的位置找到
            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
            # print(4444444444)
            while X_token_indices.numel() > 0:
                cur_X_features = X_features[cur_X_idx]
                X_token_start = X_token_indices[0]
                if getattr(self.config, 'tune_mm_mlp_adapter', False) and getattr(self.config, 'mm_use_x_start_end', False):
                    cur_new_input_embeds.append(self.get_model().embed_tokens(cur_input_ids[:X_token_start-1]).detach())
                    cur_new_input_embeds.append(self.get_model().embed_tokens(cur_input_ids[X_token_start-1:X_token_start]))
                    cur_new_input_embeds.append(cur_X_features)
                    cur_new_input_embeds.append(self.get_model().embed_tokens(cur_input_ids[X_token_start+1:X_token_start+2]))
                    if labels is not None:
                        cur_new_labels.append(cur_labels[:X_token_start])
                        cur_new_labels.append(torch.full((cur_X_features.shape[0],), IGNORE_INDEX, device=labels.device, dtype=labels.dtype))
                        cur_new_labels.append(cur_labels[X_token_start:X_token_start+1])
                        cur_labels = cur_labels[X_token_start+2:]
                else:
                    cur_new_input_embeds.append(self.get_model().embed_tokens(cur_input_ids[:X_token_start]))  # imgtoken之前的text拿出来，好像都是模板套话
                    cur_new_input_embeds.append(cur_X_features)
                    if labels is not None:
                        cur_new_labels.append(cur_labels[:X_token_start])
                        cur_new_labels.append(torch.full((cur_X_features.shape[0],), IGNORE_INDEX, device=labels.device, dtype=labels.dtype))
                        cur_labels = cur_labels[X_token_start+1:]
                cur_X_idx += 1
                if getattr(self.config, 'tune_mm_mlp_adapter', False) and getattr(self.config, 'mm_use_x_start_end', False):
                    cur_input_ids = cur_input_ids[X_token_start+2:]
                else:
                    cur_input_ids = cur_input_ids[X_token_start+1:]   # imgtoken之后的text拿出来，是真的question
                X_token_indices = torch.where(torch.any(torch.stack([cur_input_ids == X_TOKEN_INDEX[key.upper()] for key in keys]), dim=0))[0]
            
            # print(55555555555555555)
            if cur_input_ids.numel() > 0:
                if getattr(self.config, 'tune_mm_mlp_adapter', False) and getattr(self.config, 'mm_use_x_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]  # 前面text+图片+后面question
            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 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]

        return None, attention_mask, past_key_values, new_input_embeds, new_labels

    def initialize_X_tokenizer(self, model_args, tokenizer):
        if model_args.mm_use_x_patch_token:
            for x in model_args.X:
                tokenizer.add_tokens([DEFAULT_X_PATCH_TOKEN[x.upper()]], special_tokens=True)
            # tokenizer.add_tokens([DEFAULT_IMAGE_PATCH_TOKEN], special_tokens=True)
            self.resize_token_embeddings(len(tokenizer))

        if model_args.mm_use_x_start_end:
            num_new_tokens = 0
            for x in model_args.X:
                num_new_tokens += tokenizer.add_tokens([DEFAULT_X_START_TOKEN[x.upper()], DEFAULT_X_END_TOKEN[x.upper()]], 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

            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_x_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