modeling_360vl.py

from typing import List, Optional, Tuple, Union

import torch
import torch.nn as nn

from torch.nn import CrossEntropyLoss

from transformers import AutoConfig, AutoModelForCausalLM, \
                         LlamaConfig, LlamaModel, LlamaForCausalLM

from transformers.modeling_outputs import CausalLMOutputWithPast

from PIL import Image

from abc import ABC, abstractmethod
import os

import math
from transformers import CLIPVisionModel, CLIPImageProcessor, CLIPVisionConfig
from functools import partial
from transformers.configuration_utils import PretrainedConfig

from timm.models.layers import LayerNorm, LayerNorm2d
from timm.models.regnet import RegStage
from torch.nn import functional as F
import math
from einops import rearrange



CONTROLLER_HEART_BEAT_EXPIRATION = 30
WORKER_HEART_BEAT_INTERVAL = 15

LOGDIR = "."

# Model Constants
IGNORE_INDEX = -100
IMAGE_TOKEN_INDEX = -200
DEFAULT_IMAGE_TOKEN = "<image>"
DEFAULT_IMAGE_PATCH_TOKEN = "<im_patch>"
DEFAULT_IM_START_TOKEN = "<im_start>"
DEFAULT_IM_END_TOKEN = "<im_end>"





class CLIPVisionTower(nn.Module):
    def __init__(self, vision_tower, args, delay_load=False):
        super().__init__()

        self.is_loaded = False

        self.vision_tower_name = vision_tower
        self.select_layer = args.mm_vision_select_layer
        self.select_feature = getattr(args, 'mm_vision_select_feature', 'patch')

        if not delay_load:
            self.load_model()
        else:
            self.cfg_only = CLIPVisionConfig.from_pretrained(self.vision_tower_name)

    def load_model(self):
        self.image_processor = CLIPImageProcessor.from_pretrained(self.vision_tower_name)
        self.vision_tower = CLIPVisionModel.from_pretrained(self.vision_tower_name)
        self.vision_tower.requires_grad_(False)

        self.is_loaded = True

    def feature_select(self, image_forward_outs):
        image_features = image_forward_outs.hidden_states[self.select_layer]
        if self.select_feature == 'patch':
            image_features = image_features[:, 1:]
        elif self.select_feature == 'cls_patch':
            image_features = image_features
        else:
            raise ValueError(f'Unexpected select feature: {self.select_feature}')
        return image_features

    @torch.no_grad()
    def forward(self, images):
        if type(images) is list:
            image_features = []
            for image in images:
                image_forward_out = self.vision_tower(image.to(device=self.device, dtype=self.dtype).unsqueeze(0), output_hidden_states=True)
                image_feature = self.feature_select(image_forward_out).to(image.dtype)
                image_features.append(image_feature)
        else:
            image_forward_outs = self.vision_tower(images.to(device=self.device, dtype=self.dtype), output_hidden_states=True)
            image_features = self.feature_select(image_forward_outs).to(images.dtype)

        return image_features

    @property
    def dummy_feature(self):
        return torch.zeros(1, self.hidden_size, device=self.device, dtype=self.dtype)

    @property
    def dtype(self):
        return self.vision_tower.dtype

    @property
    def device(self):
        return self.vision_tower.device

    @property
    def config(self):
        if self.is_loaded:
            return self.vision_tower.config
        else:
            return self.cfg_only

    @property
    def hidden_size(self):
        return self.config.hidden_size

    @property
    def num_patches(self):
        return (self.config.image_size // self.config.patch_size) ** 2


def build_vision_tower(vision_tower_cfg, **kwargs):
    vision_tower = getattr(vision_tower_cfg, 'mm_vision_tower', getattr(vision_tower_cfg, 'vision_tower', None))
    is_absolute_path_exists = os.path.exists(vision_tower)
    
    if is_absolute_path_exists or vision_tower.startswith("openai") or vision_tower.startswith("laion"):
        return CLIPVisionTower(vision_tower, args=vision_tower_cfg, **kwargs)

    raise ValueError(f'Unknown vision tower: {vision_tower}')





class HoneybeeVisualProjectorConfig(PretrainedConfig):
    model_type = "mllm_visual_projector"

    def __init__(
        self,
        projector_type: str = "resampler",
        hidden_size: int = 1024,  #
        num_hidden_layers: int = 6,  #
        num_attention_heads: int = 16,  #
        intermediate_size: int = 4096,  #
        attention_probs_dropout_prob: float = 0.1,  #
        initializer_range: float = 0.02,
        layer_norm_eps: float = 1e-6,  #
        encoder_hidden_size: int = 1024,  # This will be overwritten by vision_model's hidden_size
        pos_emb=False,
        feature_layer_index=-1,  # vision feature layer index; -1: last layer
        num_eos_tokens=1,
        use_cls=True,
        prenorm=False,
        **kwargs,
    ):
        super().__init__(**kwargs)
        self.projector_type = projector_type
        self.hidden_size = hidden_size
        self.num_hidden_layers = num_hidden_layers
        self.num_attention_heads = num_attention_heads
        self.intermediate_size = intermediate_size
        self.attention_probs_dropout_prob = attention_probs_dropout_prob
        self.initializer_range = initializer_range
        self.layer_norm_eps = layer_norm_eps
        self.encoder_hidden_size = encoder_hidden_size

        self.pos_emb = pos_emb
        self.feature_layer_index = feature_layer_index
        self.num_eos_tokens = num_eos_tokens
        self.use_cls = use_cls
        self.prenorm = prenorm

    @classmethod
    def from_pretrained(
        cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs
    ) -> "PretrainedConfig":
        config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)

        # get the visual_projector config dict if we are loading from HoneybeeConfig
        if config_dict.get("model_type") == "mllm":
            config_dict = config_dict["visual_projector_config"]

        if (
            "model_type" in config_dict
            and hasattr(cls, "model_type")
            and config_dict["model_type"] != cls.model_type
        ):
            logger.warning(
                f"You are using a model of type {config_dict['model_type']} to instantiate a model of type "
                f"{cls.model_type}. This is not supported for all configurations of models and can yield errors."
            )

        return cls.from_dict(config_dict, **kwargs)

def build_pos_embeds(
    config: HoneybeeVisualProjectorConfig, num_input_tokens: int, vision_hidden_size: int
):
    # pos emb
    # true
    if config.pos_emb:
        pos_emb = torch.nn.Parameter(torch.zeros(1, num_input_tokens, vision_hidden_size))
        nn.init.trunc_normal_(pos_emb, mean=0.0, std=0.02)
    else:
        pos_emb = None

    return pos_emb


def build_eos_tokens(config: HoneybeeVisualProjectorConfig, output_hidden_size: int):
    # think tokens
    num_eos_tokens = config.num_eos_tokens
    # 0
    if num_eos_tokens:
        eos_tokens = torch.nn.Parameter(torch.randn(1, num_eos_tokens, output_hidden_size))
        nn.init.trunc_normal_(eos_tokens, mean=0.0, std=config.initializer_range)
    else:
        eos_tokens = None

    return eos_tokens


def build_prenorm(config: HoneybeeVisualProjectorConfig):
    # false
    if config.prenorm:
        prenorm = LayerNorm(config.encoder_hidden_size)
    else:
        prenorm = None
    return prenorm


def build_mlp(depth, hidden_size, output_hidden_size):
    layers = [nn.Linear(hidden_size, output_hidden_size)]
    for _ in range(1, depth):
        layers.append(nn.SiLU())
        layers.append(nn.Linear(output_hidden_size, output_hidden_size))
    return nn.Sequential(*layers)

def get_abs_pos(abs_pos, tgt_size):
    # abs_pos: L, C
    # tgt_size: M
    # return: M, C
    # 16,24
    src_size = int(math.sqrt(abs_pos.size(1)))
    # 32,48
    tgt_size = int(math.sqrt(tgt_size))
    dtype = abs_pos.dtype

    if src_size != tgt_size:
        return F.interpolate(
            abs_pos.float().reshape(1, src_size, src_size, -1).permute(0, 3, 1, 2),
            size=(tgt_size, tgt_size),
            mode="bicubic",
            align_corners=False,
        ).permute(0, 2, 3, 1).flatten(0, 2).to(dtype=dtype)
    else:
        return abs_pos


class Projector(nn.Module):
    """Base projector class"""

    def __init__(
        self,
        config: HoneybeeVisualProjectorConfig,
        num_input_tokens: int,
        output_hidden_size: int,
    ):
        super().__init__()
        self.config = config
        self.num_input_tokens = num_input_tokens
        self.output_hidden_size = output_hidden_size

        # think tokens
        self.eos_tokens = build_eos_tokens(config, output_hidden_size)

        # pos emb
        self.pos_emb = build_pos_embeds(config, num_input_tokens, config.encoder_hidden_size)

        self.prenorm = build_prenorm(config)

        self.build_net()

    def build_net(self):
        raise NotImplementedError()

    def _forward(self, x):
        raise NotImplementedError()

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        """
        Args:
            x: (B, L, encoder_hidden_size) tensor from the visual backbone (CLIP visual encoder), including cls token.
        """
        if self.prenorm is not None:
            x = self.prenorm(x)

        if self.pos_emb is not None:
            # self.pos_emb = self.pos_emb[:,1:]
            pos_emb = get_abs_pos(self.pos_emb[:,1:], x.size(1))
            pos_emb = pos_emb.to(device=x.device)
            x += pos_emb

        x = self._forward(x)  # (B, L, output_hidden_size)

        B = x.size(0)
        if self.eos_tokens is not None:
            x = torch.cat([x, self.eos_tokens.expand(B, -1, -1)], dim=1)
        return x


class ConvProjector(Projector):
    def _forward(self, x):
        # x: [B, L, dim]
        # x = x[:, 1:]  # drop cls token and 2d forward

        hw = int(x.size(1) ** 0.5)
        x = rearrange(x, "b (h w) d -> b d h w", h=hw, w=hw)
        x = self.net(x)
        x = rearrange(x, "b d h w -> b (h w) d")
        x = self.readout(x)

        return x


class CAbstractor(ConvProjector):
    """C-Abstractor"""
    def build_net(self):
        encoder_hidden_size = self.config.encoder_hidden_size
        hidden_size = self.config.hidden_size
        output_hidden_size = self.output_hidden_size
        depth = self.config.depth
        mlp_depth = self.config.mlp_depth

        n_queries = self.config.num_queries
        assert (n_queries ** 0.5).is_integer(), "n_queries must be square number"
        hw = int(n_queries ** 0.5)

        # RegBlock = ResBlock + SE
        RegBlock = partial(
            RegStage,
            stride=1,
            dilation=1,
            act_layer=nn.SiLU,
            norm_layer=LayerNorm2d,
        )

        s1 = RegBlock(
            depth,
            encoder_hidden_size,
            hidden_size,
        )
        sampler = nn.AdaptiveAvgPool2d((hw, hw))
        s2 = RegBlock(
            depth,
            hidden_size,
            hidden_size,
        )

        self.net = nn.Sequential(s1, sampler, s2)
        self.readout = build_mlp(mlp_depth, hidden_size, output_hidden_size)

class IdentityMap(nn.Module):
    def __init__(self):
        super().__init__()

    def forward(self, x, *args, **kwargs):
        return x

    @property
    def config(self):
        return {"mm_projector_type": 'identity'}


class SimpleResBlock(nn.Module):
    def __init__(self, channels):
        super().__init__()
        self.pre_norm = nn.LayerNorm(channels)

        self.proj = nn.Sequential(
            nn.Linear(channels, channels),
            nn.GELU(),
            nn.Linear(channels, channels)
        )
    def forward(self, x):
        x = self.pre_norm(x)
        return x + self.proj(x)


def build_honeybee_projector(config, projector_type, num_tokens,lm_hidden_size):
    """Build projector (abstractor) and query_tokens (optionally for resampler)"""
    proj_config = config
    proj_type = projector_type
    num_tokens = num_tokens
    output_hidden_size = lm_hidden_size  # LM hidden size

    abstractor = {
        "c-abs": CAbstractor,
    }[
        proj_type
    ](proj_config, num_tokens, output_hidden_size)
    return abstractor


def build_vision_projector(config, delay_load=False, **kwargs):
    projector_type = getattr(config, 'mm_projector_type', 'linear')

    if projector_type == 'linear':
        return nn.Linear(config.mm_hidden_size, config.hidden_size)

    if projector_type == 'c-abs':

        local_config_path = config.mm_projector_config
        honeybee_config = HoneybeeVisualProjectorConfig.from_pretrained(local_config_path)

        num_tokens = config.mm_num_tokens

        lm_hidden_size = config.hidden_size

        abstractor = build_honeybee_projector(honeybee_config,projector_type,num_tokens,lm_hidden_size)
        return abstractor

    mlp_gelu_match = re.match(r'^mlp(\d+)x_gelu$', projector_type)
    if mlp_gelu_match:
        mlp_depth = int(mlp_gelu_match.group(1))
        modules = [nn.Linear(config.mm_hidden_size, config.hidden_size)]
        for _ in range(1, mlp_depth):
            modules.append(nn.GELU())
            modules.append(nn.Linear(config.hidden_size, config.hidden_size))
        return nn.Sequential(*modules)

    if projector_type == 'identity':
        return IdentityMap()

    raise ValueError(f'Unknown projector type: {projector_type}')




class QH360_VL_MetaModel:

    def __init__(self, config):
        super(QH360_VL_MetaModel, self).__init__(config)
        if hasattr(config, "mm_vision_tower"):
            self.vision_tower = build_vision_tower(config, delay_load=True)
            self.mm_projector_ctt = build_vision_projector(config)
            self.mm_projector_ori = build_vision_projector(config)



    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


class QH360_VL_MetaForCausalLM(ABC):

    @abstractmethod
    def get_model(self):
        pass

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

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

    def encode_images_noprojector(self, images):
        image_features = self.get_model().get_vision_tower()(images)
        image_features = image_features.detach()
        return image_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:
            image_features = []
            for image in images:
                if image.ndim == 3:
                    image_features.append(self.encode_images(image.unsqueeze(0)).squeeze(0))
                elif image.ndim == 4:
                    #NOTE cc-plan
                    temp_feats = self.encode_images_noprojector(image)
                    src_size = int(math.sqrt(temp_feats.shape[1]))
                    temp_feats = temp_feats.reshape(temp_feats.shape[0]//5,5,-1, temp_feats.shape[-1])
                    x1 = temp_feats[:,4,:,:]
                    x = temp_feats[:,:4,:,:]
                    x = x.reshape(x.shape[0], -1, src_size, src_size, x.shape[-1])
                    x = x.transpose(1,2).reshape(x.shape[0], src_size,2,2, src_size, x.shape[-1])
                    x = x.transpose(1,2).reshape(x.shape[0], -1, x.shape[-1])
                    x1 = self.get_model().mm_projector_ori(x1).squeeze(0)
                    x = self.get_model().mm_projector_ctt(x).squeeze(0)
                    temp_feats_all = torch.cat([x,x1],dim=0)
                    image_features.append(temp_feats_all)
        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]
            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]))
                    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:]
                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 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



class QH360_VLConfig(LlamaConfig):
    model_type = "QH_360VL"


class QH360_VL_LlamaModel(QH360_VL_MetaModel, LlamaModel):
    config_class = QH360_VLConfig

    def __init__(self, config: LlamaConfig):
        super(QH360_VL_LlamaModel, self).__init__(config)


class QH360_VL_LlamaForCausalLM(LlamaForCausalLM, QH360_VL_MetaForCausalLM):
    config_class = QH360_VLConfig

    def __init__(self, config):
        super(LlamaForCausalLM, self).__init__(config)
        config._attn_implementation == "flash_attention_2"
        self.model = QH360_VL_LlamaModel(config)

        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

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

    def get_model(self):
        return self.model

    def forward(
        self,
        input_ids: torch.LongTensor = None,
        attention_mask: Optional[torch.Tensor] = None,
        past_key_values: Optional[List[torch.FloatTensor]] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        labels: Optional[torch.LongTensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        images: Optional[torch.FloatTensor] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple, CausalLMOutputWithPast]:
        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

        input_ids, attention_mask, past_key_values, inputs_embeds, labels = self.prepare_inputs_labels_for_multimodal(input_ids, attention_mask, past_key_values, labels, images)

        # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
        outputs = self.model(
            input_ids=input_ids,
            attention_mask=attention_mask,
            past_key_values=past_key_values,
            inputs_embeds=inputs_embeds,
            use_cache=use_cache,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict
        )

        hidden_states = outputs[0]
        logits = self.lm_head(hidden_states)

        loss = None
        if labels is not None:
            # Shift so that tokens < n predict n
            shift_logits = logits[..., :-1, :].contiguous()
            shift_labels = labels[..., 1:].contiguous()
            # Flatten the tokens
            loss_fct = CrossEntropyLoss()
            shift_logits = shift_logits.view(-1, self.config.vocab_size)
            shift_labels = shift_labels.view(-1)
            # Enable model/pipeline parallelism
            shift_labels = shift_labels.to(shift_logits.device)
            loss = loss_fct(shift_logits, shift_labels)

        if not return_dict:
            output = (logits,) + outputs[1:]
            return (loss,) + output if loss is not None else output

        return CausalLMOutputWithPast(
            loss=loss,
            logits=logits,
            past_key_values=outputs.past_key_values,
            hidden_states=outputs.hidden_states,
            attentions=outputs.attentions,
        )

    def prepare_inputs_for_generation(
        self, input_ids, past_key_values=None, attention_mask=None, inputs_embeds=None, **kwargs
    ):
        if past_key_values:
            input_ids = input_ids[:, -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}

        model_inputs.update(
            {
                "past_key_values": past_key_values,
                "use_cache": kwargs.get("use_cache"),
                "attention_mask": attention_mask,
                "images": kwargs.get("images", None),
            }
        )
        return model_inputs

    def build_conversation_input_ids(
            self,
            tokenizer: "PreTrainedTokenizer",
            query: str,
            image = None,
            image_processor=None,
        ):

        input_msg = [
            {
            "role": "system", 
            "content": "You are a multilingual, helpful, respectful and honest assistant who can respond in the same language, depending on the language of the question. Try to be as helpful as possible while still being safe. Your answer should not contain anything that is false, unhealthy, harmful, immoral, racist, sexist, toxic, dangerous, or illegal, and if the question relates to such content, please decline to answer. Make sure your answer is socially fair and positive. If a question doesn't make any sense, or is inconsistent with the facts, explain why instead of answering the wrong answer. If you don't know the answer to a question, don't share false information."
            },
            {
                "role": "user", 
                "content": "<|reserved_special_token_44|>"+ '\n' + query
            }
        ]

        input_ids = tokenizer.apply_chat_template(
            input_msg,
            add_generation_prompt=True,
            padding="longest",
            return_tensors="pt",
        )
        input_id_list = input_ids[0].tolist()
        input_id_list[input_id_list.index(128049)]=-200
        input_ids = torch.tensor(input_id_list, dtype=input_ids.dtype,device=input_ids.device)
        input_ids = input_ids.unsqueeze(0)
        image_tensor = self.process_images_slid_window(image,image_processor).unsqueeze(0)
        
        return {
            'input_ids': input_ids,
            'image': image_tensor,
        }



    def process_images_slid_window(self, image, image_processor, vit_is=336):

        def get_proper_imgsize(pil_img, vit_is):
            max_w_h = vit_is * 2
            new_pil_img = pil_img.resize((max_w_h, max_w_h)) 
            return new_pil_img

        def tensor_crop(tensor_array, left, upper, right, lower):
            # tensor_array: C * H * W
            return tensor_array[:, upper:lower, left:right]

        def image_slid_window(image, num_slid_window):
            # image: tensor, 3 * 336 * 336 or 3 * 672 * 672
            # image: tensor, 3 * 224 * 224 or 3 * 448 * 448
            if num_slid_window == 5:
                image_x2, image_x1 = image[0], image[1]
                vit_is = image_x1.shape[1]
                h, w  = image_x2.shape[1],image_x2.shape[2]
                image0 = tensor_crop(image_x2, 0, 0, vit_is, vit_is)
                image1 = tensor_crop(image_x2, w-vit_is, 0, w, vit_is)
                image2 = tensor_crop(image_x2, 0, h-vit_is, vit_is, h)
                image3 = tensor_crop(image_x2, w-vit_is, h-vit_is, w, h)
                return torch.stack([image0, image1, image2, image3, image_x1])
            else:
                return image

        def expand2square(pil_img, background_color):
            width, height = pil_img.size
            if width == height:
                return pil_img
            elif width > height:
                result = Image.new(pil_img.mode, (width, width), background_color)
                result.paste(pil_img, (0, (width - height) // 2))
                return result
            else:
                result = Image.new(pil_img.mode, (height, height), background_color)
                result.paste(pil_img, ((height - width) // 2, 0))
                return result

        vit_is = vit_is # vit_input_size, for simplicity

        num_slid_window = 5

        image = expand2square(image, tuple(int(x*255) for x in image_processor.image_mean))
        image = get_proper_imgsize(image, vit_is)
        image_x2 = image_processor.preprocess(image, return_tensors='pt', do_resize=False, do_center_crop=False)['pixel_values'][0]
        image_x1 = image_processor.preprocess(image, return_tensors='pt')['pixel_values'][0]
        image = [image_x2, image_x1]
        image = image_slid_window(image, num_slid_window)

        return image