Upload folder using huggingface_hub

chat_template.jinja

@@ -0,0 +1,2 @@
+<|begin_of_text|>{% for message in messages %}{{message['role'] | capitalize}}{% if message['content'][0]['type'] == 'image' %}{{':'}}{% else %}{{': '}}{% endif %}{% for line in message['content'] %}{% if line['type'] == 'text' %}{{line['text']}}{% elif line['type'] == 'image' %}{{ '<image>' }}{% endif %}{% endfor %}<end_of_utterance>
+{% endfor %}{% if add_generation_prompt %}{{ 'Assistant:' }}{% endif %}

chat_template.json

@@ -0,0 +1,3 @@
+{
+    "chat_template": "<|begin_of_text|>{% for message in messages %}{{message['role'] | capitalize}}{% if message['content'][0]['type'] == 'image' %}{{':'}}{% else %}{{': '}}{% endif %}{% for line in message['content'] %}{% if line['type'] == 'text' %}{{line['text']}}{% elif line['type'] == 'image' %}{{ '<image>' }}{% endif %}{% endfor %}<end_of_utterance>\n{% endfor %}{% if add_generation_prompt %}{{ 'Assistant:' }}{% endif %}"
+}

config.json

@@ -0,0 +1,50 @@
+{
+  "additional_vocab_size": 40,
+  "architectures": [
+    "VBertForMaskedLM"
+  ],
+  "auto_map": {
+    "AutoConfig": "configuration_vbert.VBertConfig",
+    "AutoModel": "modeling_vbert.VBertModel",
+    "AutoModelForMaskedLM": "modeling_vbert.VBertForMaskedLM"
+  },
+  "freeze_config": {
+    "freeze_lm_head": true,
+    "freeze_text_layers": true,
+    "freeze_vision_layers": true
+  },
+  "hidden_size": 768,
+  "image_token_id": 50407,
+  "initializer_range": 0.02,
+  "max_position_embeddings": 8192,
+  "model_type": "vbert",
+  "neftune_noise_alpha": 0.0,
+  "output_attentions": false,
+  "pixel_shuffle_factor": 4,
+  "qk_layer_norms": false,
+  "scale_factor": 4,
+  "text_config": {
+    "hidden_size": 768,
+    "intermediate_size": 1152,
+    "mlp_bias": false,
+    "model_type": "vbert",
+    "num_hidden_layers": 22,
+    "text_model_name": "jhu-clsp/ettin-encoder-150m",
+    "vocab_size": 50368
+  },
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": null,
+  "use_cache": true,
+  "use_resampler": false,
+  "vision_config": {
+    "embed_dim": 768,
+    "image_size": 512,
+    "intermediate_size": 3072,
+    "model_type": "vbert",
+    "num_hidden_layers": 12,
+    "patch_size": 16,
+    "vision_model_name": "google/siglip2-base-patch16-512"
+  },
+  "vocab_size": 50368
+}

configuration_vbert.py

@@ -0,0 +1,233 @@
+import copy
+import os
+
+from typing import Union, Any, Dict
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+from transformers import CONFIG_MAPPING, AutoConfig
+
+logger = logging.get_logger(__name__)
+
+def collect_arg_in_candidates(config, candidates, default = None) -> Any:
+    """ Gets the argument in a config given a list of candidates """
+    for c in candidates:
+        if hasattr(config, c):
+            return getattr(config, c)
+        elif c in config:
+            return config[c]
+    if default is not None:
+        return default
+    raise ValueError("No matching arguments found in candidates. Candidates: {}, Config: {}".format(candidates, config))
+
+class VBertTextConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`LlamaModel`]. It is used to instantiate an LLaMA
+    model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
+    defaults will yield a similar configuration to that of the LLaMA-7B.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+    Args:
+        embed_dim (`int`, *optional*, defaults to 1152):
+            Dimensionality of the encoder layers and the pooler layer. (elsewhere referred to as `embed_dim`)
+        image_size (`int`, *optional*, defaults to 384):
+            The size (resolution) of each image.
+    """
+    model_type = "vbert"
+
+    def __init__(
+        self,
+        # Case for when vllama3 is from the hub with no vision_model_name
+        text_model_name="EuroBERT/EuroBERT-210m",
+        **kwargs,
+    ):
+        self.text_model_name = text_model_name
+        text_config = AutoConfig.from_pretrained(text_model_name, trust_remote_code=True)
+        if hasattr(text_config, "text_config"):
+            text_config = text_config.text_config
+        
+        self.hidden_size = collect_arg_in_candidates(text_config, ["hidden_size", "embed_dim"])
+        self.num_hidden_layers = collect_arg_in_candidates(text_config, ["num_hidden_layers", "num_hidden_blocks"])
+        self.intermediate_size = collect_arg_in_candidates(text_config, ["intermediate_size", "mlp_dim"])
+        self.mlp_bias = collect_arg_in_candidates(text_config, ["mlp_bias", "mlp_hidden_bias"], default = False)
+        self.vocab_size = collect_arg_in_candidates(text_config, ["vocab_size"])
+    
+        super().__init__(text_model_name=text_model_name, **kwargs)
+
+class VBertVisionConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`LlamaModel`]. It is used to instantiate an LLaMA
+    model according to the specified arguments, defining the model architecture. Instantiating a configuration with the
+    defaults will yield a similar configuration to that of the LLaMA-7B.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+    Args:
+        embed_dim (`int`, *optional*, defaults to 1152):
+            Dimensionality of the encoder layers and the pooler layer. (elsewhere referred to as `embed_dim`)
+        image_size (`int`, *optional*, defaults to 384):
+            The size (resolution) of each image.
+    """
+    model_type = "vbert"
+    attribute_map = {
+        "hidden_size": "embed_dim",
+    }
+
+    def __init__(
+        self,
+        # Case for when vllama3 is from the hub with no vision_model_name
+        vision_model_name="google/siglip2-base-patch16-512",
+        **kwargs,
+    ):
+        self.vision_model_name = vision_model_name
+        vision_config = AutoConfig.from_pretrained(vision_model_name, trust_remote_code=True)
+        if hasattr(vision_config, "vision_config"):
+            vision_config = vision_config.vision_config
+
+        self.embed_dim = collect_arg_in_candidates(vision_config, ["embed_dim", "hidden_size"])
+        self.image_size = collect_arg_in_candidates(vision_config, ["image_size", "img_size"])
+        self.patch_size = collect_arg_in_candidates(vision_config, ["patch_size"])
+        self.num_hidden_layers = collect_arg_in_candidates(vision_config, ["num_hidden_layers", "num_hidden_blocks"])
+        self.intermediate_size = collect_arg_in_candidates(vision_config, ["intermediate_size", "mlp_dim"])
+
+        super().__init__(vision_model_name=vision_model_name, **kwargs)
+
+class VBertConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`SmolVLMModel`]. It is used to instantiate a
+    SmolVLM model according to the specified arguments, defining the model architecture. Instantiating a
+    configuration with the defaults will yield a similar configuration to that of the model of the SmolVLM
+    [HuggingFaceTB/SmolVLM2-2.2B-Instruct](https://huggingface.co/HuggingFaceTB/SmolVLM2-2.2B-Instruct) architecture.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+    Args:
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should cache the key/value pairs of the attention mechanism. Only
+            relevant if `config.is_decoder=True`.
+        image_token_id (`int`, *optional*, defaults to 128257):
+            The id of the "image" token.
+        tie_word_embeddings (`bool`, *optional*, defaults to `False`):
+            Whether or not to tie the word embeddings with the token embeddings.
+        vision_config (`IdeficsVisionConfig` or `dict`, *optional*, defaults to `IdeficsVisionConfig`):
+            Custom vision config or dict for the vision tower
+        text_config (`PretrainedConfig` or `dict`, *optional*, defaults to `LlamaConfig`):
+            Custom text config or dict for the text model
+        scale_factor (`int`, *optional*, defaults to 2):
+            The scale factor for the image encoder.
+        pad_token_id (`int`, *optional*, defaults to 128002):
+            The id of the padding token.
+
+    Example:
+    ```python
+    >>> from transformers import SmolVLMModel, SmolVLMConfig
+    >>> # Initializing configuration
+    >>> configuration = SmolVLMConfig()
+    >>> # Initializing a model from the configuration
+    >>> model = SmolVLMModel(configuration)
+    >>> # Accessing the model configuration
+    >>> configuration = model.config
+    ```"""
+
+    model_type = "vbert"
+    is_composition = True
+    # sub_configs = {"text_config": VBertTextConfig, "vision_config": VBertVisionConfig}
+
+    DEFAULT_TEXT_MODEL_NAME = "EuroBERT/EuroBERT-210m"
+    DEFAULT_VISION_MODEL_NAME = "google/siglip2-base-patch16-512"
+
+    def __init__(
+        self,
+        text_config: Union[PretrainedConfig, Dict[str, Any]] = None,
+        vision_config: Union[PretrainedConfig, Dict[str, Any]] = None,
+        image_token_id: int = 128_257,
+        vocab_size=128_256,
+        use_cache = True,
+        tie_word_embeddings = False,
+        freeze_config = None,
+        pad_token_id = None,
+        initializer_range = 0.02,
+        pixel_shuffle_factor = 4,
+        use_resampler = False,
+        additional_vocab_size = 0,
+        neftune_noise_alpha = 0.0,
+        **kwargs,
+    ):        
+        self.image_token_id = image_token_id
+        self.use_cache = use_cache
+        self.tie_word_embeddings = tie_word_embeddings
+        self.scale_factor = pixel_shuffle_factor
+        self.additional_vocab_size = additional_vocab_size
+
+        if text_config is None:
+            text_config = AutoConfig.from_pretrained(self.DEFAULT_TEXT_MODEL_NAME, trust_remote_code=True)
+        elif isinstance(text_config, dict):
+            text_config = VBertTextConfig(text_config["text_model_name"])
+        self.text_config = text_config
+        
+        if vision_config is None:
+            vision_config = AutoConfig.from_pretrained(self.DEFAULT_VISION_MODEL_NAME, trust_remote_code=True)
+        elif isinstance(vision_config, dict):
+            vision_config = VBertVisionConfig(vision_config["vision_model_name"])
+        self.vision_config = vision_config
+
+        self.freeze_config = freeze_config
+
+        # Pixel shuffle factor
+        self.pixel_shuffle_factor = pixel_shuffle_factor
+        self.use_resampler = use_resampler
+
+        self.neftune_noise_alpha = neftune_noise_alpha
+
+        self.initializer_range = initializer_range
+
+        hidden_size = kwargs.pop("hidden_size", self.text_config.hidden_size)
+
+        super().__init__(
+            **kwargs, 
+            pad_token_id=pad_token_id, 
+            tie_word_embeddings=tie_word_embeddings,
+            vocab_size=vocab_size,
+            hidden_size=hidden_size,
+        )
+
+    def to_dict(self):
+        """
+        Serializes this instance to a Python dictionary. Override the default [`~PretrainedConfig.to_dict`].
+        Returns:
+            `Dict[str, any]`: Dictionary of all the attributes that make up this configuration instance,
+        """
+        output = copy.deepcopy(self.__dict__)
+
+        output["model_type"] = self.__class__.model_type
+        output["vision_config"] = self.vision_config.to_dict()
+        output["text_config"] = self.text_config.to_dict()
+        # output["freeze_config"] = self.freeze_config.to_dict()
+
+        return output
+    
+    # @classmethod
+    # def from_pretrained(cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs) -> "PretrainedConfig":
+    #     outputs = super(VBertConfig, cls).from_pretrained(pretrained_model_name_or_path, **kwargs)
+    #     return outputs
+
+    @classmethod
+    def from_pretrained_models(
+        cls, 
+        text_model_name: Union[str, os.PathLike],
+        vision_model_name: Union[str, os.PathLike],
+        **kwargs
+    ) -> "PretrainedConfig":
+        # text_model_config = AutoConfig.from_pretrained(text_model_name, trust_remote_code=True)
+        # vision_model_config = AutoConfig.from_pretrained(vision_model_name, trust_remote_code=True)
+        text_model_config = VBertTextConfig(text_model_name)
+        vision_model_config = VBertVisionConfig(vision_model_name)
+        return cls(
+            text_config=text_model_config,
+            vision_config=vision_model_config,
+            **kwargs
+        )

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7258bef432f17b1010de6f93d9651d8298da83cea3430dac26b6caf43864162c
+size 1165468824

modeling_vbert.py

@@ -0,0 +1,633 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torch.nn import CrossEntropyLoss
+from typing import Optional, Tuple, Union, List
+
+from transformers.cache_utils import DynamicCache
+
+from .configuration_vbert import VBertConfig
+
+from transformers import AutoModel, AutoConfig, AutoModelForMaskedLM, PreTrainedModel
+from transformers.modeling_outputs import BaseModelOutput
+from transformers.models.bert.modeling_bert import BaseModelOutputWithPoolingAndCrossAttentions, MaskedLMOutput
+
+from typing import List, Optional, Tuple, Union
+
+import torch
+import torch.utils.checkpoint
+
+from dataclasses import dataclass
+
+from transformers import logging 
+
+logger = logging.get_logger(__name__)
+
+
+class DecoupledEmbedding(nn.Embedding):
+    # Derived from https://pytorch.org/docs/stable/_modules/torch/nn/modules/sparse.html#Embedding
+    """
+    Implements a decoupling of parameters to allow freezing (or not) a subset of the embeddings.
+    In practise, the regular `weight` can be trained or frozen (i.e. `partially_freeze=True`), and if `num_additional_embeddings` > 0, then it will create `num_additional_embeddings` additional parameters that are always trained.
+    If `num_additional_embeddings=0`, then the module defaults back to the regular behavior of `nn.Embedding`.
+    """
+
+    def __init__(
+        self,
+        num_embeddings,
+        num_additional_embeddings,
+        embedding_dim,
+        partially_freeze=False,
+        device=None,
+        dtype=None,
+        padding_idx=None,
+        **kwargs,
+    ) -> None:
+        """
+        num_additional_embeddings: int. Number of additional embeddings. Only useful when you `partially_freeze=True`.
+        partially_freeze: bool. If True, the regular `weight` will be frozen. `additional_weight` is never frozen.
+
+        Note: there are a lot of other parameters to initialize a standard `nn.Embedding` such as `padding_idx`, `max_norm` or `norm_type`. We are not supporting these.
+        """
+        if padding_idx is not None and padding_idx > num_embeddings:
+            raise ValueError(f"padding_idx must be within num_embeddings. Got {padding_idx} and {num_embeddings}")
+        super().__init__(
+            num_embeddings=num_embeddings,
+            embedding_dim=embedding_dim,
+            device=device,
+            dtype=dtype,
+            padding_idx=padding_idx,
+            **kwargs,
+        )
+        self.num_embeddings = num_embeddings
+        self.padding_idx = padding_idx
+        self.num_additional_embeddings = num_additional_embeddings
+        self.partially_freeze = partially_freeze
+
+        if partially_freeze:
+            self.weight.requires_grad_(False)
+
+        if self.num_additional_embeddings > 0:
+            self.additional_embedding = nn.Embedding(
+                num_embeddings=self.num_additional_embeddings,
+                embedding_dim=embedding_dim,
+                device=device,
+                dtype=dtype,
+            )
+
+    def forward(self, input_ids):
+        """
+        we have 2 embeddings, with different indices - one pretrained self.weight and another
+        self.additional_embedding.weight that is being trained.
+
+        in order to make a lookup of the input ids, we:
+        1. find out the indices of the entries belonging to the 2nd embedding
+        2. extract those values while subtracting the size of the first embedding (num_embeddings),
+           since the 2nd embedding starts from 0 and not num_embeddings
+        3. perform the 2nd embedding lookup
+        4. now we handle the 1st embedding, we overwrite indices belonging to the 2nd embedding with a padding index
+        5. perform the 1st embedding lookup
+        6. now we overwrite the values in the 1st embedding lookup with the values of the 2nd embedding lookup
+
+        note: for the 1st embedding lookup we could have looked up only the low indices and not do
+        the padding, but then we have to create a new tensor and populate it with 2 tensors that are
+        spread out across various indices - i.e. not a simple concat - I haven't benchmarked the
+        complex case if it's any faster, given that seqlens are usually relatively short it's
+        probably not faster or if faster not by much - but might be a good idea to measure.
+
+        """
+        if self.num_additional_embeddings == 0:
+            return self.additional_embedding(input_ids)
+
+        # Clone so that we don't modify the original input_ids later on
+        input_ids = input_ids.clone()
+        additional_vocab_indices = torch.where(input_ids >= self.num_embeddings)
+        input_ids_additional_vocab = input_ids[additional_vocab_indices]
+        additional_embeddings = self.additional_embedding(input_ids_additional_vocab - self.num_embeddings)
+
+        # for successful lookup replace input_ids with 0, the results of these will be discarded anyway
+        input_ids[additional_vocab_indices] = 0
+        full_vector = F.embedding(input_ids, self.weight)
+
+        # overwrite the records with high indices
+        full_vector[additional_vocab_indices] = additional_embeddings
+
+        return full_vector
+
+    def extra_repr(self) -> str:
+        return "num_embeddings={}, num_additional_embeddings={}, embedding_dim={}, partially_freeze={}".format(
+            self.num_embeddings,
+            self.num_additional_embeddings,
+            self.embedding_dim,
+            self.partially_freeze,
+        )
+    
+@dataclass
+class VBertBaseModelOutput(BaseModelOutput):
+    """
+    Base class for SmolVLM model's outputs that may also contain a past key/values (to speed up sequential decoding).
+    Args:
+        last_hidden_state (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`):
+            Sequence of hidden-states at the output of the last layer of the model.
+            If `past_key_values` is used only the last hidden-state of the sequences of shape `(batch_size, 1,
+            hidden_size)` is output.
+        past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`):
+            Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of shape
+            `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and optionally if
+            `config.is_encoder_decoder=True` 2 additional tensors of shape `(batch_size, num_heads,
+            encoder_sequence_length, embed_size_per_head)`.
+            Contains pre-computed hidden-states (key and values in the self-attention blocks and optionally if
+            `config.is_encoder_decoder=True` in the cross-attention blocks) that can be used (see `past_key_values`
+            input) to speed up sequential decoding.
+        hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
+            Tuple of `torch.FloatTensor` (one for the output of the embeddings, if the model has an embedding layer, +
+            one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`.
+            Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.
+        attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
+            Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
+            sequence_length)`.
+            Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
+            heads.
+        image_hidden_states (`tuple(torch.FloatTensor)`, *optional*):
+            Tuple of `torch.FloatTensor` (one for the output of the image embeddings, `(batch_size, num_images,
+            sequence_length, hidden_size)`.
+            image_hidden_states of the model produced by the vision encoder
+    """
+
+    last_hidden_state: torch.FloatTensor = None
+    hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+    attentions: Optional[Tuple[torch.FloatTensor]] = None
+    image_hidden_states: Optional[Tuple[torch.FloatTensor]] = None
+
+@dataclass
+class VBertMaskedLMOutput(MaskedLMOutput):
+    """
+    Base class for SmolVLM model's outputs that may also contain a past key/values (to speed up sequential decoding).
+    Args:
+        loss (`torch.FloatTensor`, *optional*, returned when `labels` is provided):
+            Masked language modeling (MLM) loss.
+        logits (`torch.FloatTensor`): 
+            Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
+        hidden_states (`tuple(torch.FloatTensor)`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`):
+            Tuple of `torch.FloatTensor` (one for the output of the embeddings, if the model has an embedding layer, +
+            one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`.
+            Hidden-states of the model at the output of each layer plus the optional initial embedding outputs.
+        attentions (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`):
+            Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length,
+            sequence_length)`.
+            Attentions weights after the attention softmax, used to compute the weighted average in the self-attention
+            heads.
+        image_hidden_states (`tuple(torch.FloatTensor)`, *optional*):
+            Tuple of `torch.FloatTensor` (one for the output of the image embeddings, `(batch_size, num_images,
+            sequence_length, hidden_size)`.
+            image_hidden_states of the model produced by the vision encoder
+    """
+    loss: Optional[torch.FloatTensor] = None
+    logits: torch.FloatTensor = None
+    hidden_states: Optional[Tuple[torch.FloatTensor, ...]] = None
+    attentions: Optional[Tuple[torch.FloatTensor, ...]] = None
+    image_hidden_states: Optional[torch.FloatTensor] = None
+
+class VBertSimpleMLP(nn.Module):
+    def __init__(self, input_size, output_size):
+        super().__init__()
+        self.proj = nn.Linear(input_size, output_size, bias=False)
+
+    def forward(self, x):
+        return self.proj(x)
+
+class VBertConnector(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.scale_factor = config.pixel_shuffle_factor
+        self.modality_projection = VBertSimpleMLP(
+            input_size=config.vision_config.hidden_size * (config.scale_factor**2),
+            output_size=config.text_config.hidden_size
+        )
+
+    def pixel_shuffle(self, x, scale_factor):
+        bsz, seq, embed_dim = x.size()
+        height = width = int(seq**0.5)
+        x = x.view(bsz, height, width, embed_dim)
+        x = x.view(bsz, height, int(width / scale_factor), embed_dim * scale_factor)
+        x = x.permute(0, 2, 1, 3)
+        x = x.reshape(bsz, int(width / scale_factor), int(height / scale_factor), embed_dim * (scale_factor**2))
+        x = x.permute(0, 2, 1, 3)
+        x = x.reshape(bsz, int(seq / (scale_factor**2)), embed_dim * (scale_factor**2))
+        return x
+
+    def forward(self, image_hidden_states):
+        image_hidden_states = self.pixel_shuffle(image_hidden_states, self.scale_factor)
+        image_hidden_states = self.modality_projection(image_hidden_states)
+        return image_hidden_states
+
+class VBertPreTrainedModel(PreTrainedModel):
+    config_class = VBertConfig
+    base_model_prefix = "model"
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["VBertDecoderLayer"]
+    _skip_keys_device_placement = "past_key_values"
+    _supports_flash_attn_2 = True
+    _supports_sdpa = True
+    _supports_cache_class = True
+
+    def _init_weights(self, module):
+        """Initialize the weights."""
+
+        std = (
+            self.config.initializer_range
+            if hasattr(self.config, "initializer_range")
+            else self.config.text_config.initializer_range
+        )
+
+        if hasattr(module, "class_embedding"):
+            module.class_embedding.data.normal_(mean=0.0, std=std)
+
+        if isinstance(module, (nn.Linear, nn.Conv2d)):
+            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_()
+
+class VBertModel(VBertPreTrainedModel):
+    """
+    A subclass of Idefics3Model. We do *not* remove or block the call to inputs_merger
+    in forward. Instead, we override inputs_merger here with custom logic.
+    """
+
+    def __init__(self, config: VBertConfig, **kwargs):
+        super().__init__(config)
+
+        self.vision_model = VBertModel.init_vision_model(config, **kwargs)
+        self.connector = VBertConnector(config)
+        self.text_model = VBertModel.init_language_model(config, **kwargs)
+
+        self.image_seq_len = int(
+            ((config.vision_config.image_size // config.vision_config.patch_size) ** 2) / (config.scale_factor**2)
+        )
+        self.image_token_id = self.config.image_token_id
+
+        self.post_init()
+
+    @staticmethod
+    def init_vision_model(config: VBertConfig, **kwargs):
+        vision_model_config = AutoConfig.from_pretrained(
+            config.vision_config.vision_model_name,
+            trust_remote_code=True,
+            **kwargs,
+        )
+
+        vision_model = AutoModel.from_config(vision_model_config, trust_remote_code=True, **kwargs)
+
+        if hasattr(vision_model, "vision_model"):
+            # If the model has a vision_model attribute, it means it's a wrapper around another model
+            vision_model = vision_model.vision_model
+
+        return vision_model
+
+    @staticmethod
+    def init_language_model(config: VBertConfig, **kwargs):
+        text_model_config = AutoConfig.from_pretrained(
+            config.text_config.text_model_name,
+            trust_remote_code=True,
+            **kwargs,
+        )
+
+        text_model = AutoModel.from_config(text_model_config, trust_remote_code=True, **kwargs)
+        # extractor = regex_lookup(language_model_name, language_model_name2model)
+
+        embed_layer = DecoupledEmbedding(
+            num_embeddings=text_model_config.vocab_size,
+            num_additional_embeddings=config.additional_vocab_size,
+            embedding_dim=config.hidden_size,
+            partially_freeze=config.freeze_config["freeze_text_layers"],
+            padding_idx=config.pad_token_id,
+        )
+
+        text_model.set_input_embeddings(embed_layer)
+
+        return text_model
+
+    def enable_input_require_grads(self):
+        """
+        Enables the gradients for the input embeddings.
+
+        This is useful for lora when using gradient checkpointing.
+        c.f. https://github.com/huggingface/peft/issues/1402#issuecomment-1913675032
+
+        Override to set output.requires_grad = True for both the decoder's and vision model's embeddings.
+        """
+
+        def get_lowest_module(module):
+            if len(list(module.children())) == 0:
+                # If the module has no children, it is a leaf module (e.g., Linear, Conv2d, etc.)
+                return module
+            else:
+                # Recursively call the function on each child module
+                return get_lowest_module(list(module.children())[0])
+
+        def make_inputs_require_grads(module, input, output):
+            output.requires_grad_(True)
+
+        self._text_require_grads_hook = self.get_input_embeddings().register_forward_hook(make_inputs_require_grads)
+        self._vision_require_grads_hook = get_lowest_module(self.vision_model).register_forward_hook(
+            make_inputs_require_grads
+        )
+
+    def disable_input_require_grads(self):
+        self._text_require_grads_hook.remove()
+        self._vision_require_grads_hook.remove()
+
+    def get_input_embeddings(self):
+        return self.text_model.get_input_embeddings()
+
+    def set_input_embeddings(self, value):
+        self.text_model.set_input_embeddings(value)
+
+    def inputs_merger(
+        self, input_ids: torch.LongTensor, inputs_embeds: torch.Tensor, image_hidden_states: torch.Tensor
+    ):
+        """
+        This method aims at merging the token embeddings with the image hidden states into one single sequence of vectors that are fed to the transformer LM.
+        The merging happens as follows:
+        - The text token sequence is: `tok_1 tok_2 tok_3 <fake_token_around_image> <image> <image> ... <image> <fake_token_around_image> tok_4`.
+        - We get the image hidden states for the image through the vision encoder and that hidden state, after a pixel shuffle operation, is then projected into the text embedding space.
+        We thus have a sequence of image hidden states of size (1, image_seq_len, hidden_dim), where 1 is for batch_size of 1 image and hidden_dim is the hidden_dim of the LM transformer.
+        - The merging happens so that we obtain the following sequence: `vector_tok_1 vector_tok_2 vector_tok_3 vector_fake_tok_around_image {sequence of image_seq_len image hidden states} vector_fake_toke_around_image vector_tok_4`. That sequence is fed to the LM.
+        - To fit the format of that sequence, `input_ids`, `input_embeds`, `attention_mask` are all 3 adapted to insert the image hidden states.
+        """
+        _, patch_size, _ = image_hidden_states.shape
+
+        image_mask = input_ids == self.image_token_id
+        num_image_tokens = image_mask.sum(dim=1)
+        if not torch.all(num_image_tokens % patch_size == 0):
+            raise ValueError("At least one sample has <image> tokens not divisible by patch_size.")
+
+        blocks_per_sample = num_image_tokens // patch_size
+
+        offsets = torch.nn.functional.pad(blocks_per_sample.cumsum(dim=0), (1, 0), value=0)
+        block_offset = offsets[:-1]
+        row_cum = image_mask.cumsum(dim=-1)
+        chunk_idx = (row_cum - 1) // patch_size
+        local_idx = (row_cum - 1) % patch_size
+        block_idx = block_offset.unsqueeze(1) + chunk_idx
+
+        image_embeds = torch.zeros_like(inputs_embeds)
+        image_embeds[image_mask] = image_hidden_states[block_idx[image_mask], local_idx[image_mask], :]
+
+        merged_embeds = torch.where(image_mask.unsqueeze(-1), image_embeds, inputs_embeds)
+        return merged_embeds
+
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        pixel_values: Optional[torch.FloatTensor] = None,
+        pixel_attention_mask: Optional[torch.BoolTensor] = None,
+        image_hidden_states: 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,
+        cache_position: Optional[torch.LongTensor] = None,
+    ) -> Union[Tuple, BaseModelOutputWithPoolingAndCrossAttentions]:    
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        use_cache = use_cache if use_cache is not None else self.config.use_cache
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        if self.training and self.text_model.gradient_checkpointing and use_cache:
+            logger.warning_once(
+                "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
+            )
+            use_cache = False
+
+        # retrieve input_ids and inputs_embeds
+        if 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 input_ids or inputs_embeds")
+
+        past_seen_tokens = 0
+        if use_cache:
+            if past_key_values is None:
+                past_key_values = DynamicCache()
+            past_seen_tokens = past_key_values.get_seq_length()
+
+        if inputs_embeds is not None and input_ids is None and past_seen_tokens == 0:
+            raise ValueError("When first calling the model, if input_embeds are passed, input_ids should not be None.")
+
+        if inputs_embeds is None:
+            inputs_embeds = self.text_model.get_input_embeddings()(input_ids).to(input_ids.device)
+
+        # START VISUAL INPUTS INTEGRATION
+        if pixel_values is not None and image_hidden_states is not None:
+            raise ValueError("You cannot specify both pixel_values and image_hidden_states at the same time")
+        elif pixel_values is not None:
+            batch_size, num_images, num_channels, height, width = pixel_values.shape
+            pixel_values = pixel_values
+            pixel_values = pixel_values.view(batch_size * num_images, *pixel_values.shape[2:])
+
+            # Remove padding images - padding images are full 0.
+            nb_values_per_image = pixel_values.shape[1:].numel()
+            real_images_inds = (pixel_values == 0.0).sum(dim=(-1, -2, -3)) != nb_values_per_image
+
+            if not any(real_images_inds):
+                # no images, leave one empty image.
+                real_images_inds[0] = True
+
+            pixel_values = pixel_values[real_images_inds].contiguous()
+
+            # Handle the vision attention mask
+            if pixel_attention_mask is None:
+                pixel_attention_mask = torch.ones(
+                    size=[pixel_values.shape[i] for i in (0, 2, 3)],
+                    dtype=torch.bool,
+                    device=pixel_values.device,
+                )
+            else:
+                # Remove padding images from the mask
+                pixel_attention_mask = pixel_attention_mask.view(
+                    batch_size * num_images, *pixel_attention_mask.shape[2:]
+                )
+                pixel_attention_mask = pixel_attention_mask[real_images_inds].contiguous()
+
+            # patch_size = self.config.vision_config.patch_size
+            # patches_subgrid = pixel_attention_mask.unfold(dimension=1, size=patch_size, step=patch_size)
+            # patches_subgrid = patches_subgrid.unfold(dimension=2, size=patch_size, step=patch_size)
+            # patch_attention_mask = (patches_subgrid.sum(dim=(-1, -2)) > 0).bool()
+
+            # Get sequence from the vision encoder
+            image_hidden_states = self.vision_model(
+                pixel_values=pixel_values,
+                # patch_attention_mask=patch_attention_mask,
+            ).last_hidden_state
+
+            # Modality projection & resampling
+            image_hidden_states = self.connector(image_hidden_states)
+
+        elif image_hidden_states is not None:
+            image_hidden_states = image_hidden_states.to(dtype=self.dtype, device=input_ids.device)
+
+        if inputs_embeds is not None and image_hidden_states is not None:
+            # When we embed, we don't want to replace the potential image_token_id that we generated by images
+            # that simply don't exist
+            inputs_embeds = self.inputs_merger(
+                input_ids=input_ids,
+                inputs_embeds=inputs_embeds,
+                image_hidden_states=image_hidden_states,
+            )
+
+        outputs = self.text_model(
+            inputs_embeds=inputs_embeds,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+            # past_key_values=past_key_values,
+            # use_cache=use_cache,
+            # cache_position=cache_position,
+        )
+
+        if not return_dict:
+            return tuple(v for v in [*outputs, image_hidden_states] if v is not None)
+
+        return VBertBaseModelOutput(
+            last_hidden_state=outputs.last_hidden_state,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+            image_hidden_states=image_hidden_states,
+        )
+
+class VBertLMHead(nn.Module):
+    def __init__(self, config, **kwargs):
+        super().__init__()
+        pretrained_config = AutoConfig.from_pretrained(
+            config.text_config.text_model_name,
+            trust_remote_code=True,
+            **kwargs,
+        ) 
+        pretrained_model = AutoModelForMaskedLM.from_config(pretrained_config, trust_remote_code=True, **kwargs)
+
+        self.head = pretrained_model.head
+        self.decoder = pretrained_model.decoder
+
+    def forward(self, hidden_states):
+        hidden_states = self.head(hidden_states)
+        hidden_states = self.decoder(hidden_states)
+        return hidden_states
+
+class VBertForMaskedLM(VBertPreTrainedModel):
+    # _tied_weights_keys = ["predictions.decoder.bias", "cls.predictions.decoder.weight"]
+
+    def __init__(self, config, **kwargs):
+        super().__init__(config)
+
+        self.image_token_id = config.image_token_id
+        self.in_features = config.hidden_size
+        self.out_additional_features = config.additional_vocab_size
+        self.vocab_size = config.vocab_size
+
+        if config.is_decoder:
+            logger.warning(
+                "If you want to use `BertForMaskedLM` make sure `config.is_decoder=False` for "
+                "bi-directional self-attention."
+            )
+
+        self.model = VBertModel(config, **kwargs)
+        self.lm_head = VBertLMHead(config, **kwargs)
+        if self.out_additional_features > 0:
+            self.additional_fc = nn.Linear(
+                in_features=self.in_features,
+                out_features=self.out_additional_features,
+                bias=False,
+            )
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def forward(
+            self,
+            input_ids: torch.LongTensor = None,
+            attention_mask: Optional[torch.Tensor] = None,
+            position_ids: Optional[torch.LongTensor] = None,
+            past_key_values: Optional[List[torch.FloatTensor]] = None,
+            inputs_embeds: Optional[torch.FloatTensor] = None,
+            pixel_values: Optional[torch.FloatTensor] = None,
+            pixel_attention_mask: Optional[torch.BoolTensor] = None,
+            image_hidden_states: 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,
+            return_dict: Optional[bool] = None,
+    ) -> Union[Tuple, VBertMaskedLMOutput]:
+        r"""
+        Args:
+            labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+                Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+                config.vocab_size]` or `model.image_token_id` (where `model` is your instance of `Idefics3ForConditionalGeneration`).
+                Tokens with indices set to `model.image_token_id` are ignored (masked), the loss is only
+                computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+        ```"""
+        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
+
+
+        # Pass the inputs to VBertModel
+        outputs = self.model(
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            inputs_embeds=inputs_embeds,
+            pixel_values=pixel_values,
+            pixel_attention_mask=pixel_attention_mask,
+            image_hidden_states=image_hidden_states,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        # Pass the outputs to the MLM head
+        hidden_states = outputs[0]
+
+        logits = self.lm_head(hidden_states)
+        if self.out_additional_features > 0:
+            proj_states = self.lm_head.head(hidden_states)
+            additional_features = self.additional_fc(proj_states)
+            logits = torch.cat((logits, additional_features), -1)
+        logits = logits.float()
+
+        masked_lm_loss = None
+        if labels is not None:
+            # print the ratio of not ignored tokens
+            loss_fct = CrossEntropyLoss()
+            masked_lm_loss = loss_fct(logits.view(-1, self.vocab_size + self.out_additional_features), labels.view(-1))
+
+        if not return_dict:
+            output = (logits,) + outputs[2:]
+            return ((masked_lm_loss,) + output) if masked_lm_loss is not None else output
+
+        return VBertMaskedLMOutput(
+            loss=masked_lm_loss,
+            logits=logits,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+            image_hidden_states=outputs.image_hidden_states,
+        )

preprocessor_config.json

@@ -0,0 +1,28 @@
+{
+    "do_convert_rgb": true,
+    "do_image_splitting": true,
+    "do_normalize": true,
+    "do_pad": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "image_processor_type": "Idefics3ImageProcessor",
+    "image_std": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "max_image_size": {
+        "longest_edge": 512
+    },
+    "processor_class": "Idefics3Processor",
+    "resample": 1,
+    "rescale_factor": 0.00392156862745098,
+    "size": {
+        "longest_edge": 2048
+    }
+}

processor_config.json

@@ -0,0 +1,4 @@
+{
+    "image_seq_len": 64,
+    "processor_class": "Idefics3Processor"
+}

special_tokens_map.json

@@ -0,0 +1,79 @@
+{
+  "additional_special_tokens": [
+    "<global-img>",
+    "<row_1_col_1>",
+    "<row_1_col_2>",
+    "<row_1_col_3>",
+    "<row_1_col_4>",
+    "<row_1_col_5>",
+    "<row_1_col_6>",
+    "<row_2_col_1>",
+    "<row_2_col_2>",
+    "<row_2_col_3>",
+    "<row_2_col_4>",
+    "<row_2_col_5>",
+    "<row_2_col_6>",
+    "<row_3_col_1>",
+    "<row_3_col_2>",
+    "<row_3_col_3>",
+    "<row_3_col_4>",
+    "<row_3_col_5>",
+    "<row_3_col_6>",
+    "<row_4_col_1>",
+    "<row_4_col_2>",
+    "<row_4_col_3>",
+    "<row_4_col_4>",
+    "<row_4_col_5>",
+    "<row_4_col_6>",
+    "<row_5_col_1>",
+    "<row_5_col_2>",
+    "<row_5_col_3>",
+    "<row_5_col_4>",
+    "<row_5_col_5>",
+    "<row_5_col_6>",
+    "<row_6_col_1>",
+    "<row_6_col_2>",
+    "<row_6_col_3>",
+    "<row_6_col_4>",
+    "<row_6_col_5>",
+    "<row_6_col_6>",
+    "<end_of_utterance>",
+    "<fake_token_around_image>",
+    "<image>"
+  ],
+  "cls_token": {
+    "content": "[CLS]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "mask_token": {
+    "content": "[MASK]",
+    "lstrip": true,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "[PAD]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "sep_token": {
+    "content": "[SEP]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "[UNK]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer_config.json

@@ -0,0 +1,1310 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "|||IP_ADDRESS|||",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "1": {
+      "content": "<|padding|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50254": {
+      "content": "                        ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50255": {
+      "content": "                       ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50256": {
+      "content": "                      ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50257": {
+      "content": "                     ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50258": {
+      "content": "                    ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50259": {
+      "content": "                   ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50260": {
+      "content": "                  ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50261": {
+      "content": "                 ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50262": {
+      "content": "                ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50263": {
+      "content": "               ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50264": {
+      "content": "              ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50265": {
+      "content": "             ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50266": {
+      "content": "            ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50267": {
+      "content": "           ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50268": {
+      "content": "          ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50269": {
+      "content": "         ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50270": {
+      "content": "        ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50271": {
+      "content": "       ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50272": {
+      "content": "      ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50273": {
+      "content": "     ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50274": {
+      "content": "    ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50275": {
+      "content": "   ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50276": {
+      "content": "  ",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50277": {
+      "content": "|||EMAIL_ADDRESS|||",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50278": {
+      "content": "|||PHONE_NUMBER|||",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50279": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50280": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50281": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50282": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50283": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50284": {
+      "content": "[MASK]",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50285": {
+      "content": "[unused0]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50286": {
+      "content": "[unused1]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50287": {
+      "content": "[unused2]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50288": {
+      "content": "[unused3]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50289": {
+      "content": "[unused4]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50290": {
+      "content": "[unused5]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50291": {
+      "content": "[unused6]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50292": {
+      "content": "[unused7]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50293": {
+      "content": "[unused8]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50294": {
+      "content": "[unused9]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50295": {
+      "content": "[unused10]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50296": {
+      "content": "[unused11]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50297": {
+      "content": "[unused12]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50298": {
+      "content": "[unused13]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50299": {
+      "content": "[unused14]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50300": {
+      "content": "[unused15]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50301": {
+      "content": "[unused16]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50302": {
+      "content": "[unused17]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50303": {
+      "content": "[unused18]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50304": {
+      "content": "[unused19]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50305": {
+      "content": "[unused20]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50306": {
+      "content": "[unused21]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50307": {
+      "content": "[unused22]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50308": {
+      "content": "[unused23]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50309": {
+      "content": "[unused24]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50310": {
+      "content": "[unused25]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50311": {
+      "content": "[unused26]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50312": {
+      "content": "[unused27]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50313": {
+      "content": "[unused28]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50314": {
+      "content": "[unused29]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50315": {
+      "content": "[unused30]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50316": {
+      "content": "[unused31]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50317": {
+      "content": "[unused32]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50318": {
+      "content": "[unused33]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50319": {
+      "content": "[unused34]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50320": {
+      "content": "[unused35]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50321": {
+      "content": "[unused36]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50322": {
+      "content": "[unused37]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50323": {
+      "content": "[unused38]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50324": {
+      "content": "[unused39]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50325": {
+      "content": "[unused40]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50326": {
+      "content": "[unused41]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50327": {
+      "content": "[unused42]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50328": {
+      "content": "[unused43]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50329": {
+      "content": "[unused44]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50330": {
+      "content": "[unused45]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50331": {
+      "content": "[unused46]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50332": {
+      "content": "[unused47]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50333": {
+      "content": "[unused48]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50334": {
+      "content": "[unused49]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50335": {
+      "content": "[unused50]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50336": {
+      "content": "[unused51]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50337": {
+      "content": "[unused52]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50338": {
+      "content": "[unused53]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50339": {
+      "content": "[unused54]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50340": {
+      "content": "[unused55]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50341": {
+      "content": "[unused56]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50342": {
+      "content": "[unused57]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50343": {
+      "content": "[unused58]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50344": {
+      "content": "[unused59]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50345": {
+      "content": "[unused60]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50346": {
+      "content": "[unused61]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50347": {
+      "content": "[unused62]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50348": {
+      "content": "[unused63]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50349": {
+      "content": "[unused64]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50350": {
+      "content": "[unused65]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50351": {
+      "content": "[unused66]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50352": {
+      "content": "[unused67]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50353": {
+      "content": "[unused68]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50354": {
+      "content": "[unused69]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50355": {
+      "content": "[unused70]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50356": {
+      "content": "[unused71]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50357": {
+      "content": "[unused72]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50358": {
+      "content": "[unused73]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50359": {
+      "content": "[unused74]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50360": {
+      "content": "[unused75]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50361": {
+      "content": "[unused76]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50362": {
+      "content": "[unused77]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50363": {
+      "content": "[unused78]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50364": {
+      "content": "[unused79]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50365": {
+      "content": "[unused80]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50366": {
+      "content": "[unused81]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50367": {
+      "content": "[unused82]",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50368": {
+      "content": "<global-img>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50369": {
+      "content": "<row_1_col_1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50370": {
+      "content": "<row_1_col_2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50371": {
+      "content": "<row_1_col_3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50372": {
+      "content": "<row_1_col_4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50373": {
+      "content": "<row_1_col_5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50374": {
+      "content": "<row_1_col_6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50375": {
+      "content": "<row_2_col_1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50376": {
+      "content": "<row_2_col_2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50377": {
+      "content": "<row_2_col_3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50378": {
+      "content": "<row_2_col_4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50379": {
+      "content": "<row_2_col_5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50380": {
+      "content": "<row_2_col_6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50381": {
+      "content": "<row_3_col_1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50382": {
+      "content": "<row_3_col_2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50383": {
+      "content": "<row_3_col_3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50384": {
+      "content": "<row_3_col_4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50385": {
+      "content": "<row_3_col_5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50386": {
+      "content": "<row_3_col_6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50387": {
+      "content": "<row_4_col_1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50388": {
+      "content": "<row_4_col_2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50389": {
+      "content": "<row_4_col_3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50390": {
+      "content": "<row_4_col_4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50391": {
+      "content": "<row_4_col_5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50392": {
+      "content": "<row_4_col_6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50393": {
+      "content": "<row_5_col_1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50394": {
+      "content": "<row_5_col_2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50395": {
+      "content": "<row_5_col_3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50396": {
+      "content": "<row_5_col_4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50397": {
+      "content": "<row_5_col_5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50398": {
+      "content": "<row_5_col_6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50399": {
+      "content": "<row_6_col_1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50400": {
+      "content": "<row_6_col_2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50401": {
+      "content": "<row_6_col_3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50402": {
+      "content": "<row_6_col_4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50403": {
+      "content": "<row_6_col_5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50404": {
+      "content": "<row_6_col_6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50405": {
+      "content": "<end_of_utterance>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50406": {
+      "content": "<fake_token_around_image>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "50407": {
+      "content": "<image>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<global-img>",
+    "<row_1_col_1>",
+    "<row_1_col_2>",
+    "<row_1_col_3>",
+    "<row_1_col_4>",
+    "<row_1_col_5>",
+    "<row_1_col_6>",
+    "<row_2_col_1>",
+    "<row_2_col_2>",
+    "<row_2_col_3>",
+    "<row_2_col_4>",
+    "<row_2_col_5>",
+    "<row_2_col_6>",
+    "<row_3_col_1>",
+    "<row_3_col_2>",
+    "<row_3_col_3>",
+    "<row_3_col_4>",
+    "<row_3_col_5>",
+    "<row_3_col_6>",
+    "<row_4_col_1>",
+    "<row_4_col_2>",
+    "<row_4_col_3>",
+    "<row_4_col_4>",
+    "<row_4_col_5>",
+    "<row_4_col_6>",
+    "<row_5_col_1>",
+    "<row_5_col_2>",
+    "<row_5_col_3>",
+    "<row_5_col_4>",
+    "<row_5_col_5>",
+    "<row_5_col_6>",
+    "<row_6_col_1>",
+    "<row_6_col_2>",
+    "<row_6_col_3>",
+    "<row_6_col_4>",
+    "<row_6_col_5>",
+    "<row_6_col_6>",
+    "<end_of_utterance>",
+    "<fake_token_around_image>",
+    "<image>"
+  ],
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "extra_special_tokens": {},
+  "legacy": false,
+  "mask_token": "[MASK]",
+  "model_input_names": [
+    "input_ids",
+    "attention_mask",
+    "pixel_values",
+    "pixel_attention_mask"
+  ],
+  "model_max_length": 8192,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "tokenizer_class": "PreTrainedTokenizerFast",
+  "unk_token": "[UNK]"
+}