Upload folder using huggingface_hub

added_tokens.json

@@ -0,0 +1,11 @@
+{
+  "</box>": 92552,
+  "</img>": 92545,
+  "</quad>": 92548,
+  "</ref>": 92550,
+  "<IMG_CONTEXT>": 92546,
+  "<box>": 92551,
+  "<img>": 92544,
+  "<quad>": 92547,
+  "<ref>": 92549
+}

configuration_intern_vit.py

@@ -0,0 +1,119 @@
+# --------------------------------------------------------
+# InternVL
+# Copyright (c) 2024 OpenGVLab
+# Licensed under The MIT License [see LICENSE for details]
+# --------------------------------------------------------
+import os
+from typing import Union
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+logger = logging.get_logger(__name__)
+
+
+class InternVisionConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`InternVisionModel`]. It is used to
+    instantiate a vision encoder according to the specified arguments, defining the model architecture.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+    Args:
+        num_channels (`int`, *optional*, defaults to 3):
+            Number of color channels in the input images (e.g., 3 for RGB).
+        patch_size (`int`, *optional*, defaults to 14):
+            The size (resolution) of each patch.
+        image_size (`int`, *optional*, defaults to 224):
+            The size (resolution) of each image.
+        qkv_bias (`bool`, *optional*, defaults to `False`):
+            Whether to add a bias to the queries and values in the self-attention layers.
+        hidden_size (`int`, *optional*, defaults to 3200):
+            Dimensionality of the encoder layers and the pooler layer.
+        num_attention_heads (`int`, *optional*, defaults to 25):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        intermediate_size (`int`, *optional*, defaults to 12800):
+            Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
+        qk_normalization (`bool`, *optional*, defaults to `True`):
+            Whether to normalize the queries and keys in the self-attention layers.
+        num_hidden_layers (`int`, *optional*, defaults to 48):
+            Number of hidden layers in the Transformer encoder.
+        use_flash_attn (`bool`, *optional*, defaults to `True`):
+            Whether to use flash attention mechanism.
+        hidden_act (`str` or `function`, *optional*, defaults to `"gelu"`):
+            The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
+            `"relu"`, `"selu"` and `"gelu_new"` ``"gelu"` are supported.
+        layer_norm_eps (`float`, *optional*, defaults to 1e-6):
+            The epsilon used by the layer normalization layers.
+        dropout (`float`, *optional*, defaults to 0.0):
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
+        drop_path_rate (`float`, *optional*, defaults to 0.0):
+            Dropout rate for stochastic depth.
+        attention_dropout (`float`, *optional*, defaults to 0.0):
+            The dropout ratio for the attention probabilities.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        initializer_factor (`float`, *optional*, defaults to 0.1):
+            A factor for layer scale.
+    """
+
+    model_type = 'intern_vit_6b'
+
+    def __init__(
+            self,
+            num_channels=3,
+            patch_size=14,
+            image_size=224,
+            qkv_bias=False,
+            hidden_size=3200,
+            num_attention_heads=25,
+            intermediate_size=12800,
+            qk_normalization=True,
+            num_hidden_layers=48,
+            use_flash_attn=True,
+            hidden_act='gelu',
+            norm_type='rms_norm',
+            layer_norm_eps=1e-6,
+            dropout=0.0,
+            drop_path_rate=0.0,
+            attention_dropout=0.0,
+            initializer_range=0.02,
+            initializer_factor=0.1,
+            **kwargs,
+    ):
+        super().__init__(**kwargs)
+
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.dropout = dropout
+        self.drop_path_rate = drop_path_rate
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.num_channels = num_channels
+        self.patch_size = patch_size
+        self.image_size = image_size
+        self.initializer_range = initializer_range
+        self.initializer_factor = initializer_factor
+        self.attention_dropout = attention_dropout
+        self.layer_norm_eps = layer_norm_eps
+        self.hidden_act = hidden_act
+        self.norm_type = norm_type
+        self.qkv_bias = qkv_bias
+        self.qk_normalization = qk_normalization
+        self.use_flash_attn = use_flash_attn
+
+    @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)
+
+        if 'vision_config' in config_dict:
+            config_dict = config_dict['vision_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)

configuration_internlm2.py

@@ -0,0 +1,150 @@
+# Copyright (c) The InternLM team and The HuggingFace Inc. team. All rights reserved.
+#
+# This code is based on transformers/src/transformers/models/llama/configuration_llama.py
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+""" InternLM2 model configuration"""
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+logger = logging.get_logger(__name__)
+
+INTERNLM2_PRETRAINED_CONFIG_ARCHIVE_MAP = {}
+
+
+# Modified from transformers.model.llama.configuration_llama.LlamaConfig
+class InternLM2Config(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`InternLM2Model`]. It is used to instantiate
+    an InternLM2 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 InternLM2-7B.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 32000):
+            Vocabulary size of the InternLM2 model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`InternLM2Model`]
+        hidden_size (`int`, *optional*, defaults to 4096):
+            Dimension of the hidden representations.
+        intermediate_size (`int`, *optional*, defaults to 11008):
+            Dimension of the MLP representations.
+        num_hidden_layers (`int`, *optional*, defaults to 32):
+            Number of hidden layers in the Transformer encoder.
+        num_attention_heads (`int`, *optional*, defaults to 32):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        num_key_value_heads (`int`, *optional*):
+            This is the number of key_value heads that should be used to implement Grouped Query Attention. If
+            `num_key_value_heads=num_attention_heads`, the model will use Multi Head Attention (MHA), if
+            `num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. When
+            converting a multi-head checkpoint to a GQA checkpoint, each group key and value head should be constructed
+            by meanpooling all the original heads within that group. For more details checkout [this
+            paper](https://arxiv.org/pdf/2305.13245.pdf). If it is not specified, will default to
+            `num_attention_heads`.
+        hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
+            The non-linear activation function (function or string) in the decoder.
+        max_position_embeddings (`int`, *optional*, defaults to 2048):
+            The maximum sequence length that this model might ever be used with. Typically set this to something large
+            just in case (e.g., 512 or 1024 or 2048).
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        rms_norm_eps (`float`, *optional*, defaults to 1e-12):
+            The epsilon used by the rms normalization layers.
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models). Only
+            relevant if `config.is_decoder=True`.
+        tie_word_embeddings(`bool`, *optional*, defaults to `False`):
+            Whether to tie weight embeddings
+        Example:
+
+    """
+    model_type = 'internlm2'
+    _auto_class = 'AutoConfig'
+
+    def __init__(  # pylint: disable=W0102
+        self,
+        vocab_size=103168,
+        hidden_size=4096,
+        intermediate_size=11008,
+        num_hidden_layers=32,
+        num_attention_heads=32,
+        num_key_value_heads=None,
+        hidden_act='silu',
+        max_position_embeddings=2048,
+        initializer_range=0.02,
+        rms_norm_eps=1e-6,
+        use_cache=True,
+        pad_token_id=0,
+        bos_token_id=1,
+        eos_token_id=2,
+        tie_word_embeddings=False,
+        bias=True,
+        rope_theta=10000,
+        rope_scaling=None,
+        attn_implementation='eager',
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.bias = bias
+
+        if num_key_value_heads is None:
+            num_key_value_heads = num_attention_heads
+        self.num_key_value_heads = num_key_value_heads
+
+        self.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.use_cache = use_cache
+        self.rope_theta = rope_theta
+        self.rope_scaling = rope_scaling
+        self._rope_scaling_validation()
+
+        self.attn_implementation = attn_implementation
+        if self.attn_implementation is None:
+            self.attn_implementation = 'eager'
+        super().__init__(
+            pad_token_id=pad_token_id,
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )
+
+    def _rope_scaling_validation(self):
+        """
+        Validate the `rope_scaling` configuration.
+        """
+        if self.rope_scaling is None:
+            return
+
+        if not isinstance(self.rope_scaling, dict) or len(self.rope_scaling) != 2:
+            raise ValueError(
+                '`rope_scaling` must be a dictionary with with two fields, `type` and `factor`, '
+                f'got {self.rope_scaling}'
+            )
+        rope_scaling_type = self.rope_scaling.get('type', None)
+        rope_scaling_factor = self.rope_scaling.get('factor', None)
+        if rope_scaling_type is None or rope_scaling_type not in ['linear', 'dynamic']:
+            raise ValueError(
+                f"`rope_scaling`'s type field must be one of ['linear', 'dynamic'], got {rope_scaling_type}"
+            )
+        if rope_scaling_factor is None or not isinstance(rope_scaling_factor, float) or rope_scaling_factor < 1.0:
+            raise ValueError(f"`rope_scaling`'s factor field must be a float >= 1, got {rope_scaling_factor}")

configuration_internvl_chat.py

@@ -0,0 +1,96 @@
+# --------------------------------------------------------
+# InternVL
+# Copyright (c) 2024 OpenGVLab
+# Licensed under The MIT License [see LICENSE for details]
+# --------------------------------------------------------
+
+import copy
+
+from transformers import AutoConfig, LlamaConfig
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+from .configuration_intern_vit import InternVisionConfig
+from .configuration_internlm2 import InternLM2Config
+
+logger = logging.get_logger(__name__)
+
+
+class InternVLChatConfig(PretrainedConfig):
+    model_type = 'internvl_chat'
+    is_composition = True
+
+    def __init__(
+            self,
+            vision_config=None,
+            llm_config=None,
+            use_backbone_lora=0,
+            use_llm_lora=0,
+            select_layer=-1,
+            force_image_size=None,
+            downsample_ratio=0.5,
+            template=None,
+            dynamic_image_size=False,
+            use_thumbnail=False,
+            ps_version='v1',
+            min_dynamic_patch=1,
+            max_dynamic_patch=6,
+            **kwargs):
+        super().__init__(**kwargs)
+
+        if vision_config is None:
+            vision_config = {}
+            logger.info('vision_config is None. Initializing the InternVisionConfig with default values.')
+
+        if llm_config is None:
+            llm_config = {}
+            logger.info('llm_config is None. Initializing the LlamaConfig config with default values (`LlamaConfig`).')
+
+        self.vision_config = InternVisionConfig(**vision_config)
+        if llm_config['architectures'][0] == 'LlamaForCausalLM':
+            self.llm_config = LlamaConfig(**llm_config)
+        elif llm_config['architectures'][0] == 'InternLM2ForCausalLM':
+            self.llm_config = InternLM2Config(**llm_config)
+        else:
+            raise ValueError('Unsupported architecture: {}'.format(llm_config['architectures'][0]))
+        self.use_backbone_lora = use_backbone_lora
+        self.use_llm_lora = use_llm_lora
+        self.select_layer = select_layer
+        self.force_image_size = force_image_size
+        self.downsample_ratio = downsample_ratio
+        self.template = template
+        self.dynamic_image_size = dynamic_image_size
+        self.use_thumbnail = use_thumbnail
+        self.ps_version = ps_version  # pixel shuffle version
+        self.min_dynamic_patch = min_dynamic_patch
+        self.max_dynamic_patch = max_dynamic_patch
+
+        logger.info(f'vision_select_layer: {self.select_layer}')
+        logger.info(f'ps_version: {self.ps_version}')
+        logger.info(f'min_dynamic_patch: {self.min_dynamic_patch}')
+        logger.info(f'max_dynamic_patch: {self.max_dynamic_patch}')
+
+    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['vision_config'] = self.vision_config.to_dict()
+        output['llm_config'] = self.llm_config.to_dict()
+        output['model_type'] = self.__class__.model_type
+        output['use_backbone_lora'] = self.use_backbone_lora
+        output['use_llm_lora'] = self.use_llm_lora
+        output['select_layer'] = self.select_layer
+        output['force_image_size'] = self.force_image_size
+        output['downsample_ratio'] = self.downsample_ratio
+        output['template'] = self.template
+        output['dynamic_image_size'] = self.dynamic_image_size
+        output['use_thumbnail'] = self.use_thumbnail
+        output['ps_version'] = self.ps_version
+        output['min_dynamic_patch'] = self.min_dynamic_patch
+        output['max_dynamic_patch'] = self.max_dynamic_patch
+
+        return output

generation_config.json

@@ -0,0 +1,4 @@
+{
+  "_from_model_config": true,
+  "transformers_version": "4.39.3"
+}

model-00001-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ffa49c03299ed3c3fb1aa86e04f26b511f38047356473ad72d5d42ec2c3503f
+size 4939944336

model-00002-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5088b9dba4d87bd022e35ab00cdc40d2f15e4ad9c61e44375f523f2f331ce883
+size 4915914584

model-00003-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d761b20908afb4fdbdbe9c0f74ba2f533eb85755308e0cfc478156e3ba27a12d
+size 4915914592

model-00004-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4404e9def49d54a79a21cf21daa87245d4b4ae6612e78a5c770efc541f36eb0e
+size 1379026920

model.safetensors.index.json

@@ -0,0 +1,580 @@
+{
+  "metadata": {
+    "total_size": 16150730752
+  },
+  "weight_map": {
+    "language_model.model.layers.0.attention.wo.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.0.attention.wqkv.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.0.attention_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.0.feed_forward.w1.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.0.feed_forward.w2.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.0.feed_forward.w3.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.0.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.1.attention.wo.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.1.attention.wqkv.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.1.attention_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.1.feed_forward.w1.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.1.feed_forward.w2.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.1.feed_forward.w3.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.1.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.10.attention.wo.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.10.attention.wqkv.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.10.attention_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.10.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.10.feed_forward.w2.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.10.feed_forward.w3.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.10.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.11.attention.wo.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.11.attention.wqkv.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.11.attention_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.11.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.11.feed_forward.w2.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.11.feed_forward.w3.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.11.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.12.attention.wo.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.12.attention.wqkv.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.12.attention_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.12.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.12.feed_forward.w2.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.12.feed_forward.w3.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.12.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.13.attention.wo.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.13.attention.wqkv.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.13.attention_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.13.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.13.feed_forward.w2.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.13.feed_forward.w3.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.13.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.14.attention.wo.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.14.attention.wqkv.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.14.attention_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.14.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.14.feed_forward.w2.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.14.feed_forward.w3.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.14.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.15.attention.wo.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.15.attention.wqkv.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.15.attention_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.15.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.15.feed_forward.w2.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.15.feed_forward.w3.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.15.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.16.attention.wo.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.16.attention.wqkv.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.16.attention_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.16.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.16.feed_forward.w2.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.16.feed_forward.w3.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.16.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.17.attention.wo.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.17.attention.wqkv.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.17.attention_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.17.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.17.feed_forward.w2.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.17.feed_forward.w3.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.17.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.18.attention.wo.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.18.attention.wqkv.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.18.attention_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.18.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.18.feed_forward.w2.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.18.feed_forward.w3.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.18.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.19.attention.wo.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.19.attention.wqkv.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.19.attention_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.19.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.19.feed_forward.w2.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.19.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.19.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.2.attention.wo.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.2.attention.wqkv.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.2.attention_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.2.feed_forward.w1.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.2.feed_forward.w2.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.2.feed_forward.w3.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.2.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.20.attention.wo.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.20.attention.wqkv.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.20.attention_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.20.feed_forward.w1.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.20.feed_forward.w2.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.20.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.20.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.21.attention.wo.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.21.attention.wqkv.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.21.attention_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.21.feed_forward.w1.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.21.feed_forward.w2.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.21.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.21.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.22.attention.wo.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.22.attention.wqkv.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.22.attention_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.22.feed_forward.w1.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.22.feed_forward.w2.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.22.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.22.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.23.attention.wo.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.23.attention.wqkv.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.23.attention_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.23.feed_forward.w1.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.23.feed_forward.w2.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.23.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.23.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.24.attention.wo.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.24.attention.wqkv.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.24.attention_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.24.feed_forward.w1.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.24.feed_forward.w2.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.24.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.24.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.25.attention.wo.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.25.attention.wqkv.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.25.attention_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.25.feed_forward.w1.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.25.feed_forward.w2.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.25.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.25.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.26.attention.wo.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.26.attention.wqkv.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.26.attention_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.26.feed_forward.w1.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.26.feed_forward.w2.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.26.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.26.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.27.attention.wo.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.27.attention.wqkv.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.27.attention_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.27.feed_forward.w1.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.27.feed_forward.w2.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.27.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.27.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.28.attention.wo.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.28.attention.wqkv.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.28.attention_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.28.feed_forward.w1.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.28.feed_forward.w2.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.28.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.28.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.29.attention.wo.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.29.attention.wqkv.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.29.attention_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.29.feed_forward.w1.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.29.feed_forward.w2.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.29.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.29.ffn_norm.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.3.attention.wo.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.3.attention.wqkv.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.3.attention_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.3.feed_forward.w1.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.3.feed_forward.w2.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.3.feed_forward.w3.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.3.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.30.attention.wo.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.30.attention.wqkv.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.30.attention_norm.weight": "model-00004-of-00004.safetensors",
+    "language_model.model.layers.30.feed_forward.w1.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.30.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
+    "language_model.model.layers.30.feed_forward.w3.weight": "model-00003-of-00004.safetensors",
+    "language_model.model.layers.30.ffn_norm.weight": "model-00004-of-00004.safetensors",
+    "language_model.model.layers.31.attention.wo.weight": "model-00004-of-00004.safetensors",
+    "language_model.model.layers.31.attention.wqkv.weight": "model-00004-of-00004.safetensors",
+    "language_model.model.layers.31.attention_norm.weight": "model-00004-of-00004.safetensors",
+    "language_model.model.layers.31.feed_forward.w1.weight": "model-00004-of-00004.safetensors",
+    "language_model.model.layers.31.feed_forward.w2.weight": "model-00004-of-00004.safetensors",
+    "language_model.model.layers.31.feed_forward.w3.weight": "model-00004-of-00004.safetensors",
+    "language_model.model.layers.31.ffn_norm.weight": "model-00004-of-00004.safetensors",
+    "language_model.model.layers.4.attention.wo.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.4.attention.wqkv.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.4.attention_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.4.feed_forward.w1.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.4.feed_forward.w2.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.4.feed_forward.w3.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.4.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.5.attention.wo.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.5.attention.wqkv.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.5.attention_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.5.feed_forward.w1.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.5.feed_forward.w2.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.5.feed_forward.w3.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.5.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.6.attention.wo.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.6.attention.wqkv.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.6.attention_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.6.feed_forward.w1.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.6.feed_forward.w2.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.6.feed_forward.w3.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.6.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.7.attention.wo.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.7.attention.wqkv.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.7.attention_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.7.feed_forward.w1.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.7.feed_forward.w2.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.7.feed_forward.w3.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.7.ffn_norm.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.8.attention.wo.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.8.attention.wqkv.weight": "model-00001-of-00004.safetensors",
+    "language_model.model.layers.8.attention_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.8.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.8.feed_forward.w2.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.8.feed_forward.w3.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.8.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.9.attention.wo.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.9.attention.wqkv.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.9.attention_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.9.feed_forward.w1.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.9.feed_forward.w2.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.9.feed_forward.w3.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.layers.9.ffn_norm.weight": "model-00002-of-00004.safetensors",
+    "language_model.model.norm.weight": "model-00004-of-00004.safetensors",
+    "language_model.model.tok_embeddings.weight": "model-00001-of-00004.safetensors",
+    "language_model.output.weight": "model-00004-of-00004.safetensors",
+    "mlp1.0.bias": "model-00004-of-00004.safetensors",
+    "mlp1.0.weight": "model-00004-of-00004.safetensors",
+    "mlp1.1.bias": "model-00004-of-00004.safetensors",
+    "mlp1.1.weight": "model-00004-of-00004.safetensors",
+    "mlp1.3.bias": "model-00004-of-00004.safetensors",
+    "mlp1.3.weight": "model-00004-of-00004.safetensors",
+    "vision_model.embeddings.class_embedding": "model-00001-of-00004.safetensors",
+    "vision_model.embeddings.patch_embedding.bias": "model-00001-of-00004.safetensors",
+    "vision_model.embeddings.patch_embedding.weight": "model-00001-of-00004.safetensors",
+    "vision_model.embeddings.position_embedding": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.0.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.1.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.10.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.11.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.12.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.13.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.14.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.15.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.16.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.17.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.18.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.19.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.2.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.20.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.21.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.22.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.23.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.3.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.4.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.5.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.6.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.7.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.8.norm2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.attn.proj.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.attn.proj.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.attn.qkv.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.attn.qkv.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.ls1": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.ls2": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.norm1.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.norm1.weight": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.norm2.bias": "model-00001-of-00004.safetensors",
+    "vision_model.encoder.layers.9.norm2.weight": "model-00001-of-00004.safetensors"
+  }
+}

modeling_internvl_chat.py

@@ -0,0 +1,346 @@
+# --------------------------------------------------------
+# InternVL
+# Copyright (c) 2024 OpenGVLab
+# Licensed under The MIT License [see LICENSE for details]
+# --------------------------------------------------------
+import warnings
+from typing import Any, List, Optional, Tuple, Union
+
+import torch.utils.checkpoint
+import transformers
+from torch import nn
+from torch.nn import CrossEntropyLoss
+from transformers import (AutoModel, GenerationConfig, LlamaForCausalLM,
+                          LlamaTokenizer)
+from transformers.modeling_outputs import CausalLMOutputWithPast
+from transformers.modeling_utils import PreTrainedModel
+from transformers.utils import ModelOutput, logging
+
+from .configuration_internvl_chat import InternVLChatConfig
+from .conversation import get_conv_template
+from .modeling_intern_vit import InternVisionModel
+from .modeling_internlm2 import InternLM2ForCausalLM
+
+logger = logging.get_logger(__name__)
+
+
+def version_cmp(v1, v2, op='eq'):
+    import operator
+
+    from packaging import version
+    op_func = getattr(operator, op)
+    return op_func(version.parse(v1), version.parse(v2))
+
+
+class InternVLChatModel(PreTrainedModel):
+    config_class = InternVLChatConfig
+    main_input_name = 'pixel_values'
+    _supports_flash_attn_2 = True
+    _no_split_modules = ['InternVisionModel', 'LlamaDecoderLayer', 'InternLM2DecoderLayer']
+
+    def __init__(self, config: InternVLChatConfig, vision_model=None, language_model=None):
+        super().__init__(config)
+
+        assert version_cmp(transformers.__version__, '4.36.2', 'ge')
+        image_size = config.force_image_size or config.vision_config.image_size
+        patch_size = config.vision_config.patch_size
+        self.patch_size = patch_size
+        self.select_layer = config.select_layer
+        self.template = config.template
+        self.num_image_token = int((image_size // patch_size) ** 2 * (config.downsample_ratio ** 2))
+        self.downsample_ratio = config.downsample_ratio
+        self.ps_version = config.ps_version
+
+        logger.info(f'num_image_token: {self.num_image_token}')
+        logger.info(f'ps_version: {self.ps_version}')
+        if vision_model is not None:
+            self.vision_model = vision_model
+        else:
+            self.vision_model = InternVisionModel(config.vision_config)
+        if language_model is not None:
+            self.language_model = language_model
+        else:
+            if config.llm_config.architectures[0] == 'LlamaForCausalLM':
+                self.language_model = LlamaForCausalLM(config.llm_config)
+            elif config.llm_config.architectures[0] == 'InternLM2ForCausalLM':
+                self.language_model = InternLM2ForCausalLM(config.llm_config)
+            else:
+                raise NotImplementedError(f'{config.llm_config.architectures[0]} is not implemented.')
+
+        vit_hidden_size = config.vision_config.hidden_size
+        llm_hidden_size = config.llm_config.hidden_size
+
+        self.mlp1 = nn.Sequential(
+            nn.LayerNorm(vit_hidden_size * int(1 / self.downsample_ratio) ** 2),
+            nn.Linear(vit_hidden_size * int(1 / self.downsample_ratio) ** 2, llm_hidden_size),
+            nn.GELU(),
+            nn.Linear(llm_hidden_size, llm_hidden_size)
+        )
+
+        self.img_context_token_id = None
+        self.conv_template = get_conv_template(self.template)
+        self.system_message = self.conv_template.system_message
+
+    def forward(
+            self,
+            pixel_values: torch.FloatTensor,
+            input_ids: torch.LongTensor = None,
+            attention_mask: Optional[torch.Tensor] = None,
+            position_ids: Optional[torch.LongTensor] = None,
+            image_flags: Optional[torch.LongTensor] = None,
+            past_key_values: Optional[List[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, CausalLMOutputWithPast]:
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        image_flags = image_flags.squeeze(-1)
+        input_embeds = self.language_model.get_input_embeddings()(input_ids)
+
+        vit_embeds = self.extract_feature(pixel_values)
+        vit_embeds = vit_embeds[image_flags == 1]
+        vit_batch_size = pixel_values.shape[0]
+
+        B, N, C = input_embeds.shape
+        input_embeds = input_embeds.reshape(B * N, C)
+
+        if torch.distributed.get_rank() == 0:
+            print(f'dynamic ViT batch size: {vit_batch_size}, images per sample: {vit_batch_size / B}, dynamic token length: {N}')
+
+        input_ids = input_ids.reshape(B * N)
+        selected = (input_ids == self.img_context_token_id)
+        try:
+            input_embeds[selected] = input_embeds[selected] * 0.0 + vit_embeds.reshape(-1, C)
+        except Exception as e:
+            vit_embeds = vit_embeds.reshape(-1, C)
+            print(f'warning: {e}, input_embeds[selected].shape={input_embeds[selected].shape}, '
+                  f'vit_embeds.shape={vit_embeds.shape}')
+            n_token = selected.sum()
+            input_embeds[selected] = input_embeds[selected] * 0.0 + vit_embeds[:n_token]
+
+        input_embeds = input_embeds.reshape(B, N, C)
+
+        outputs = self.language_model(
+            inputs_embeds=input_embeds,
+            attention_mask=attention_mask,
+            position_ids=position_ids,
+            past_key_values=past_key_values,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        logits = outputs.logits
+
+        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.language_model.config.vocab_size)
+            shift_labels = shift_labels.view(-1)
+            # Enable model 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 pixel_shuffle(self, x, scale_factor=0.5):
+        n, w, h, c = x.size()
+        # N, W, H, C --> N, W, H * scale, C // scale
+        x = x.view(n, w, int(h * scale_factor), int(c / scale_factor))
+        # N, W, H * scale, C // scale --> N, H * scale, W, C // scale
+        x = x.permute(0, 2, 1, 3).contiguous()
+        # N, H * scale, W, C // scale --> N, H * scale, W * scale, C // (scale ** 2)
+        x = x.view(n, int(h * scale_factor), int(w * scale_factor),
+                   int(c / (scale_factor * scale_factor)))
+        if self.ps_version == 'v1':
+            warnings.warn("In ps_version 'v1', the height and width have not been swapped back, "
+                          'which results in a transposed image.')
+        else:
+            x = x.permute(0, 2, 1, 3).contiguous()
+        return x
+
+    def extract_feature(self, pixel_values):
+        if self.select_layer == -1:
+            vit_embeds = self.vision_model(
+                pixel_values=pixel_values,
+                output_hidden_states=False,
+                return_dict=True).last_hidden_state
+        else:
+            vit_embeds = self.vision_model(
+                pixel_values=pixel_values,
+                output_hidden_states=True,
+                return_dict=True).hidden_states[self.select_layer]
+        vit_embeds = vit_embeds[:, 1:, :]
+
+        h = w = int(vit_embeds.shape[1] ** 0.5)
+        vit_embeds = vit_embeds.reshape(vit_embeds.shape[0], h, w, -1)
+        vit_embeds = self.pixel_shuffle(vit_embeds, scale_factor=self.downsample_ratio)
+        vit_embeds = vit_embeds.reshape(vit_embeds.shape[0], -1, vit_embeds.shape[-1])
+        vit_embeds = self.mlp1(vit_embeds)
+        return vit_embeds
+
+    def batch_chat(self, tokenizer, pixel_values, questions, generation_config, num_patches_list=None,
+                   history=None, return_history=False, IMG_START_TOKEN='<img>', IMG_END_TOKEN='</img>',
+                   IMG_CONTEXT_TOKEN='<IMG_CONTEXT>', verbose=False, image_counts=None):
+        if history is not None or return_history:
+            print('Now multi-turn chat is not supported in batch_chat.')
+            raise NotImplementedError
+
+        if image_counts is not None:
+            num_patches_list = image_counts
+            print('Warning: `image_counts` is deprecated. Please use `num_patches_list` instead.')
+
+        img_context_token_id = tokenizer.convert_tokens_to_ids(IMG_CONTEXT_TOKEN)
+        self.img_context_token_id = img_context_token_id
+
+        if verbose and pixel_values is not None:
+            image_bs = pixel_values.shape[0]
+            print(f'dynamic ViT batch size: {image_bs}')
+
+        queries = []
+        for idx, num_patches in enumerate(num_patches_list):
+            question = questions[idx]
+            if pixel_values is not None and '<image>' not in question:
+                question = '<image>\n' + question
+            template = get_conv_template(self.template)
+            template.system_message = self.system_message
+            template.append_message(template.roles[0], question)
+            template.append_message(template.roles[1], None)
+            query = template.get_prompt()
+
+            image_tokens = IMG_START_TOKEN + IMG_CONTEXT_TOKEN * self.num_image_token * num_patches + IMG_END_TOKEN
+            query = query.replace('<image>', image_tokens, 1)
+            queries.append(query)
+
+        tokenizer.padding_side = 'left'
+        model_inputs = tokenizer(queries, return_tensors='pt', padding=True)
+        input_ids = model_inputs['input_ids'].cuda()
+        attention_mask = model_inputs['attention_mask'].cuda()
+        eos_token_id = tokenizer.convert_tokens_to_ids(template.sep)
+        generation_config['eos_token_id'] = eos_token_id
+        generation_output = self.generate(
+            pixel_values=pixel_values,
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            **generation_config
+        )
+        responses = tokenizer.batch_decode(generation_output, skip_special_tokens=True)
+        responses = [response.split(template.sep)[0].strip() for response in responses]
+        return responses
+
+    def chat(self, tokenizer, pixel_values, question, generation_config, history=None, return_history=False,
+             num_patches_list=None, IMG_START_TOKEN='<img>', IMG_END_TOKEN='</img>', IMG_CONTEXT_TOKEN='<IMG_CONTEXT>',
+             verbose=False):
+
+        if history is None and pixel_values is not None and '<image>' not in question:
+            question = '<image>\n' + question
+
+        if num_patches_list is None:
+            num_patches_list = [pixel_values.shape[0]] if pixel_values is not None else []
+        assert pixel_values is None or len(pixel_values) == sum(num_patches_list)
+
+        img_context_token_id = tokenizer.convert_tokens_to_ids(IMG_CONTEXT_TOKEN)
+        self.img_context_token_id = img_context_token_id
+
+        template = get_conv_template(self.template)
+        template.system_message = self.system_message
+        eos_token_id = tokenizer.convert_tokens_to_ids(template.sep)
+
+        history = [] if history is None else history
+        for (old_question, old_answer) in history:
+            template.append_message(template.roles[0], old_question)
+            template.append_message(template.roles[1], old_answer)
+        template.append_message(template.roles[0], question)
+        template.append_message(template.roles[1], None)
+        query = template.get_prompt()
+
+        if verbose and pixel_values is not None:
+            image_bs = pixel_values.shape[0]
+            print(f'dynamic ViT batch size: {image_bs}')
+
+        for num_patches in num_patches_list:
+            image_tokens = IMG_START_TOKEN + IMG_CONTEXT_TOKEN * self.num_image_token * num_patches + IMG_END_TOKEN
+            query = query.replace('<image>', image_tokens, 1)
+
+        model_inputs = tokenizer(query, return_tensors='pt')
+        input_ids = model_inputs['input_ids'].cuda()
+        attention_mask = model_inputs['attention_mask'].cuda()
+        generation_config['eos_token_id'] = eos_token_id
+        generation_output = self.generate(
+            pixel_values=pixel_values,
+            input_ids=input_ids,
+            attention_mask=attention_mask,
+            **generation_config
+        )
+        response = tokenizer.batch_decode(generation_output, skip_special_tokens=True)[0]
+        response = response.split(template.sep)[0].strip()
+        history.append((question, response))
+        if return_history:
+            return response, history
+        else:
+            query_to_print = query.replace(IMG_CONTEXT_TOKEN, '')
+            query_to_print = query_to_print.replace(f'{IMG_START_TOKEN}{IMG_END_TOKEN}', '<image>')
+            if verbose:
+                print(query_to_print, response)
+            return response
+
+    @torch.no_grad()
+    def generate(
+            self,
+            pixel_values: Optional[torch.FloatTensor] = None,
+            input_ids: Optional[torch.FloatTensor] = None,
+            attention_mask: Optional[torch.LongTensor] = None,
+            visual_features: Optional[torch.FloatTensor] = None,
+            generation_config: Optional[GenerationConfig] = None,
+            output_hidden_states: Optional[bool] = None,
+            return_dict: Optional[bool] = None,
+            **generate_kwargs,
+    ) -> torch.LongTensor:
+
+        assert self.img_context_token_id is not None
+        if pixel_values is not None:
+            if visual_features is not None:
+                vit_embeds = visual_features
+            else:
+                vit_embeds = self.extract_feature(pixel_values)
+            input_embeds = self.language_model.get_input_embeddings()(input_ids)
+            B, N, C = input_embeds.shape
+            input_embeds = input_embeds.reshape(B * N, C)
+
+            input_ids = input_ids.reshape(B * N)
+            selected = (input_ids == self.img_context_token_id)
+            assert selected.sum() != 0
+            input_embeds[selected] = vit_embeds.reshape(-1, C).to(input_embeds.device)
+
+            input_embeds = input_embeds.reshape(B, N, C)
+        else:
+            input_embeds = self.language_model.get_input_embeddings()(input_ids)
+
+        outputs = self.language_model.generate(
+            inputs_embeds=input_embeds,
+            attention_mask=attention_mask,
+            generation_config=generation_config,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+            use_cache=True,
+            **generate_kwargs,
+        )
+
+        return outputs

special_tokens_map.json

@@ -0,0 +1,47 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|action_start|>",
+    "<|action_end|>",
+    "<|interpreter|>",
+    "<|plugin|>",
+    "<img>",
+    "</img>",
+    "<IMG_CONTEXT>",
+    "<quad>",
+    "</quad>",
+    "<ref>",
+    "</ref>",
+    "<box>",
+    "</box>"
+  ],
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenization_internlm2.py

@@ -0,0 +1,235 @@
+# Copyright (c) The InternLM team and The HuggingFace Inc. team. All rights reserved.
+#
+# This code is based on transformers/src/transformers/models/llama/tokenization_llama.py
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Tokenization classes for InternLM."""
+import os
+from shutil import copyfile
+from typing import Any, Dict, List, Optional, Tuple
+
+import sentencepiece as spm
+from transformers.tokenization_utils import PreTrainedTokenizer
+from transformers.utils import logging
+
+logger = logging.get_logger(__name__)
+
+VOCAB_FILES_NAMES = {'vocab_file': './tokenizer.model'}
+
+PRETRAINED_VOCAB_FILES_MAP = {}
+
+
+# Modified from transformers.model.llama.tokenization_llama.LlamaTokenizer
+class InternLM2Tokenizer(PreTrainedTokenizer):
+    """
+    Construct a InternLM2 tokenizer. Based on byte-level Byte-Pair-Encoding.
+
+    Args:
+        vocab_file (`str`):
+            Path to the vocabulary file.
+    """
+
+    vocab_files_names = VOCAB_FILES_NAMES
+    pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
+    model_input_names = ['input_ids', 'attention_mask']
+    _auto_class = 'AutoTokenizer'
+
+    def __init__(
+        self,
+        vocab_file,
+        unk_token='<unk>',
+        bos_token='<s>',
+        eos_token='</s>',
+        pad_token='</s>',
+        sp_model_kwargs: Optional[Dict[str, Any]] = None,
+        add_bos_token=True,
+        add_eos_token=False,
+        decode_with_prefix_space=False,
+        clean_up_tokenization_spaces=False,
+        **kwargs,
+    ):
+        self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs
+        self.vocab_file = vocab_file
+        self.add_bos_token = add_bos_token
+        self.add_eos_token = add_eos_token
+        self.decode_with_prefix_space = decode_with_prefix_space
+        self.sp_model = spm.SentencePieceProcessor(**self.sp_model_kwargs)
+        self.sp_model.Load(vocab_file)
+        self._no_prefix_space_tokens = None
+        super().__init__(
+            bos_token=bos_token,
+            eos_token=eos_token,
+            unk_token=unk_token,
+            pad_token=pad_token,
+            clean_up_tokenization_spaces=clean_up_tokenization_spaces,
+            **kwargs,
+        )
+
+    @property
+    def no_prefix_space_tokens(self):
+        if self._no_prefix_space_tokens is None:
+            vocab = self.convert_ids_to_tokens(list(range(self.vocab_size)))
+            self._no_prefix_space_tokens = {i for i, tok in enumerate(vocab) if not tok.startswith('▁')}
+        return self._no_prefix_space_tokens
+
+    @property
+    def vocab_size(self):
+        """Returns vocab size"""
+        return self.sp_model.get_piece_size()
+
+    @property
+    def bos_token_id(self) -> Optional[int]:
+        return self.sp_model.bos_id()
+
+    @property
+    def eos_token_id(self) -> Optional[int]:
+        return self.sp_model.eos_id()
+
+    def get_vocab(self):
+        """Returns vocab as a dict"""
+        vocab = {self.convert_ids_to_tokens(i): i for i in range(self.vocab_size)}
+        vocab.update(self.added_tokens_encoder)
+        return vocab
+
+    def _tokenize(self, text):
+        """Returns a tokenized string."""
+        return self.sp_model.encode(text, out_type=str)
+
+    def _convert_token_to_id(self, token):
+        """Converts a token (str) in an id using the vocab."""
+        return self.sp_model.piece_to_id(token)
+
+    def _convert_id_to_token(self, index):
+        """Converts an index (integer) in a token (str) using the vocab."""
+        token = self.sp_model.IdToPiece(index)
+        return token
+
+    def _maybe_add_prefix_space(self, tokens, decoded):
+        if tokens and tokens[0] not in self.no_prefix_space_tokens:
+            return ' ' + decoded
+        else:
+            return decoded
+
+    def convert_tokens_to_string(self, tokens):
+        """Converts a sequence of tokens (string) in a single string."""
+        current_sub_tokens = []
+        out_string = ''
+        prev_is_special = False
+        for token in tokens:
+            # make sure that special tokens are not decoded using sentencepiece model
+            if token in self.all_special_tokens:
+                if not prev_is_special:
+                    out_string += ' '
+                out_string += self.sp_model.decode(current_sub_tokens) + token
+                prev_is_special = True
+                current_sub_tokens = []
+            else:
+                current_sub_tokens.append(token)
+                prev_is_special = False
+        out_string += self.sp_model.decode(current_sub_tokens)
+        out_string = self.clean_up_tokenization(out_string)
+        out_string = self._maybe_add_prefix_space(tokens=tokens, decoded=out_string)
+        return out_string[1:]
+
+    def save_vocabulary(self, save_directory, filename_prefix: Optional[str] = None) -> Tuple[str]:
+        """
+        Save the vocabulary and special tokens file to a directory.
+
+        Args:
+            save_directory (`str`):
+                The directory in which to save the vocabulary.
+
+        Returns:
+            `Tuple(str)`: Paths to the files saved.
+        """
+        if not os.path.isdir(save_directory):
+            logger.error(f'Vocabulary path ({save_directory}) should be a directory')
+            return
+        out_vocab_file = os.path.join(
+            save_directory, (filename_prefix + '-' if filename_prefix else '') + VOCAB_FILES_NAMES['vocab_file']
+        )
+
+        if os.path.abspath(self.vocab_file) != os.path.abspath(out_vocab_file) and os.path.isfile(self.vocab_file):
+            copyfile(self.vocab_file, out_vocab_file)
+        elif not os.path.isfile(self.vocab_file):
+            with open(out_vocab_file, 'wb') as fi:
+                content_spiece_model = self.sp_model.serialized_model_proto()
+                fi.write(content_spiece_model)
+
+        return (out_vocab_file,)
+
+    def build_inputs_with_special_tokens(self, token_ids_0, token_ids_1=None):
+        if self.add_bos_token:
+            bos_token_ids = [self.bos_token_id]
+        else:
+            bos_token_ids = []
+
+        output = bos_token_ids + token_ids_0
+
+        if token_ids_1 is not None:
+            output = output + token_ids_1
+
+        if self.add_eos_token:
+            output = output + [self.eos_token_id]
+
+        return output
+
+    def get_special_tokens_mask(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None, already_has_special_tokens: bool = False
+    ) -> List[int]:
+        """
+        Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding
+        special tokens using the tokenizer `prepare_for_model` method.
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+            already_has_special_tokens (`bool`, *optional*, defaults to `False`):
+                Whether or not the token list is already formatted with special tokens for the model.
+
+        Returns:
+            `List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
+        """
+        if already_has_special_tokens:
+            return super().get_special_tokens_mask(
+                token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=True
+            )
+
+        if token_ids_1 is None:
+            return [1] + ([0] * len(token_ids_0)) + [1]
+        return [1] + ([0] * len(token_ids_0)) + [1, 1] + ([0] * len(token_ids_1)) + [1]
+
+    def create_token_type_ids_from_sequences(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
+    ) -> List[int]:
+        """
+        Create a mask from the two sequences passed to be used in a sequence-pair classification task. T5 does not make
+        use of token type ids, therefore a list of zeros is returned.
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+
+        Returns:
+            `List[int]`: List of zeros.
+        """
+        eos = [self.eos_token_id]
+
+        if token_ids_1 is None:
+            return len(token_ids_0 + eos) * [0]
+        return len(token_ids_0 + eos + token_ids_1 + eos) * [0]

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f868398fc4e05ee1e8aeba95ddf18ddcc45b8bce55d5093bead5bbf80429b48b
+size 1477754

tokenizer_config.json

@@ -0,0 +1,179 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92538": {
+      "content": "<|plugin|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92539": {
+      "content": "<|interpreter|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92540": {
+      "content": "<|action_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92541": {
+      "content": "<|action_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92542": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92543": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92544": {
+      "content": "<img>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92545": {
+      "content": "</img>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92546": {
+      "content": "<IMG_CONTEXT>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92547": {
+      "content": "<quad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92548": {
+      "content": "</quad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92549": {
+      "content": "<ref>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92550": {
+      "content": "</ref>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92551": {
+      "content": "<box>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "92552": {
+      "content": "</box>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|action_start|>",
+    "<|action_end|>",
+    "<|interpreter|>",
+    "<|plugin|>",
+    "<img>",
+    "</img>",
+    "<IMG_CONTEXT>",
+    "<quad>",
+    "</quad>",
+    "<ref>",
+    "</ref>",
+    "<box>",
+    "</box>"
+  ],
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_internlm2.InternLM2Tokenizer",
+      null
+    ]
+  },
+  "bos_token": "<s>",
+  "chat_template": "{{ bos_token }}{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "</s>",
+  "model_max_length": 2048,
+  "pad_token": "</s>",
+  "tokenizer_class": "InternLM2Tokenizer",
+  "unk_token": "<unk>"
+}