File size: 14,660 Bytes
8241db4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 |
# coding=utf-8
# Copyright 2024 Microsoft Research and The HuggingFace Inc. team. All rights reserved.
#
# 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.
"""KOSMOS-2.5.5 model configuration"""
import os
from typing import Union
from transformers.configuration_utils import PretrainedConfig
from transformers.utils import logging
logger = logging.get_logger(__name__)
class Kosmos2_5TextConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a [`Kosmos2_5TextModel`]. It is used to instantiate a
KOSMOS-2.5 text decoder according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the text decoder of the KOSMOS-2.5
[microsoft/KOSMOS-2.5](https://huggingface.co/microsoft/KOSMOS-2.5) architecture.
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 108481):
Vocabulary size of the Kosmos2_5 model. Defines the number of different tokens that can be represented by the
`inputs_ids` passed when calling [`Kosmos2_5Model`].
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).
embed_dim (`int`, *optional*, defaults to 2048):
Dimensionality of the layers and the pooler layer.
layers (`int`, *optional*, defaults to 24):
Number of hidden layers in the Transformer encoder.
ffn_dim (`int`, *optional*, defaults to 8192):
Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder.
attention_heads (`int`, *optional*, defaults to 32):
Number of attention heads for each attention layer in the Transformer encoder.
activation_function (`str` or `function`, *optional*, defaults to `"gelu"`):
The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
`"relu"`, `"silu"` and `"gelu_new"` are supported.
dropout (`float`, *optional*, defaults to 0.1):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
attention_dropout (`float`, *optional*, defaults to 0.1):
The dropout ratio for the attention probabilities.
activation_dropout (`float`, *optional*, defaults to 0.0):
The dropout ratio for activations inside the fully connected layer.
layerdrop (`float`, *optional*, defaults to 0.0):
The LayerDrop probability for the decoder. See the [LayerDrop paper](see https://arxiv.org/abs/1909.11556)
for more details.
layer_norm_eps (`float`, *optional*, defaults to 1e-5):
The epsilon used by the layer normalization layers.
init_std (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
scale_embedding (`bool`, *optional*, defaults to `True`):
Scale embeddings by diving by sqrt(embed_dim).
use_cache (`bool`, *optional*, defaults to `True`):
Whether or not the model should return the last key/values attentions (not used by all models).
```"""
model_type = "kosmos_2_5_text_model"
keys_to_ignore_at_inference = ["past_key_values"]
attribute_map = {
"num_attention_heads": "attention_heads",
"hidden_size": "embed_dim",
"num_hidden_layers": "layers",
}
def __init__(
self,
vocab_size=108481,
max_position_embeddings=4096,
embed_dim=1536,
layers=24,
ffn_dim=6144,
attention_heads=16,
activation_function="gelu",
dropout=0.1,
attention_dropout=0,
activation_dropout=0.0,
layerdrop=0.0,
layer_norm_eps=1e-5,
init_std=0.02,
scale_embedding=True,
use_cache=True,
pad_token_id=1,
bos_token_id=0,
eos_token_id=2,
**kwargs,
):
super().__init__(
pad_token_id=pad_token_id,
bos_token_id=bos_token_id,
eos_token_id=eos_token_id,
**kwargs,
)
self.vocab_size = vocab_size
self.max_position_embeddings = max_position_embeddings
self.embed_dim = embed_dim
self.layers = layers
self.ffn_dim = ffn_dim
self.attention_heads = attention_heads
self.activation_function = activation_function
self.dropout = dropout
self.attention_dropout = attention_dropout
self.activation_dropout = activation_dropout
self.layerdrop = layerdrop
self.layer_norm_eps = layer_norm_eps
self.init_std = init_std
self.scale_embedding = scale_embedding
self.use_cache = use_cache
@classmethod
def from_pretrained(cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs) -> "PretrainedConfig":
cls._set_token_in_kwargs(kwargs)
config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)
# get the text config dict if we are loading from Kosmos2_5Config
if config_dict.get("model_type") == "kosmos-2.5":
config_dict = config_dict["text_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)
class Kosmos2_5VisionConfig(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a [`Kosmos2_5VisionModel`]. It is used to
instantiate a Kosmos2_5 vision model according to the specified arguments, defining the model architecture.
Instantiating a configuration defaults will yield a similar configuration to that of the kosmos-2.5
[microsoft/kosmos-2.5](https://huggingface.co/microsoft/kosmos-2.5) architecture.
Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
documentation from [`PretrainedConfig`] for more information.
Args:
hidden_size (`int`, *optional*, defaults to 768):
Dimensionality of the encoder layers and the pooler layer.
patch_embed_hidden_size (`int`, *optional*, defaults to 768):
Dimensionality of the input patch_embedding layer in the Transformer encoder.
d_ff (`int`, *optional*, defaults to 2048):
Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
d_kv (`int`, *optional*, defaults to 64):
Dimensionality of the key, query, value projections per attention head.
num_hidden_layers (`int`, *optional*, defaults to 12):
Number of hidden layers in the Transformer encoder.
num_attention_heads (`int`, *optional*, defaults to 12):
Number of attention heads for each attention layer in the Transformer encoder.
dense_act_fn (`str` or `function`, *optional*, defaults to `"gelu_new"`):
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-06):
The epsilon used by the layer normalization layers.
dropout_rate (`float`, *optional*, defaults to 0.0):
The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
attention_dropout (`float`, *optional*, defaults to 0.0):
The dropout ratio for the attention probabilities.
initializer_range (`float`, *optional*, defaults to 1e-10):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
initializer_factor (`float`, *optional*, defaults to 1.0):
A factor for initializing all weight matrices (should be kept to 1, used internally for initialization
testing).
seq_len (`int`, *optional*, defaults to 4096):
Maximum sequence length (here number of patches) supported by the model.
Example:
```python
>>> from transformers import Kosmos2_5VisionConfig, Kosmos2_5VisionModel
>>> # Initializing a Kosmos2_5VisionConfig with microsoft/kosmos-2.5 style configuration
>>> configuration = Kosmos2_5VisionConfig()
>>> # Initializing a Kosmos2_5VisionModel (with random weights) from the microsoft/kosmos-2.5 style configuration
>>> model = Kosmos2_5VisionModel(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
```"""
model_type = "kosmos_2_5_vision_model"
def __init__(
self,
hidden_size=1536,
patch_embed_hidden_size=768,
d_ff=3968,
d_kv=64,
num_hidden_layers=18,
num_attention_heads=24,
dense_act_fn="gelu_new",
layer_norm_eps=1e-6,
dropout_rate=0.0,
attention_dropout=0.0,
initializer_range=1e-10,
initializer_factor=1.0,
seq_len=4096,
**kwargs,
):
super().__init__(**kwargs)
self.hidden_size = hidden_size
self.patch_embed_hidden_size = patch_embed_hidden_size
self.d_ff = d_ff
self.dropout_rate = dropout_rate
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.initializer_range = initializer_range
self.initializer_factor = initializer_factor
self.attention_dropout = attention_dropout
self.layer_norm_eps = layer_norm_eps
self.dense_act_fn = dense_act_fn
self.seq_len = seq_len
self.d_kv = d_kv
@classmethod
def from_pretrained(
cls, pretrainehidden_size_name_or_path: Union[str, os.PathLike], **kwargs
) -> "PretrainedConfig":
cls._set_token_in_kwargs(kwargs)
config_dict, kwargs = cls.get_config_dict(pretrainehidden_size_name_or_path, **kwargs)
# get the vision config dict if we are loading from Kosmos2_5Config
if config_dict.get("model_type") == "Kosmos2_5":
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)
class Kosmos2_5Config(PretrainedConfig):
r"""
This is the configuration class to store the configuration of a [`Kosmos2_5Model`]. It is used to instantiate a
KOSMOS-2.5 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 KOSMOS-2.5
[microsoft/KOSMOS-2.5-patch14-224](https://huggingface.co/microsoft/KOSMOS-2.5-patch14-224) architecture.
Args:
text_config (`dict`, *optional*):
Dictionary of configuration options used to initialize [`Kosmos2_5TextConfig`].
vision_config (`dict`, *optional*):
Dictionary of configuration options used to initialize [`Kosmos2_5VisionConfig`].
latent_query_num (`int`, *optional*, defaults to 2048):
The number of latent query tokens that represent the image features used in the text decoder component.
kwargs (*optional*):
Dictionary of keyword arguments.
Example:
```python
>>> from .. import Kosmos2_5Config, Kosmos2_5Model
>>> # Initializing a KOSMOS-2.5 KOSMOS-2.5-patch14-224 style configuration
>>> configuration = Kosmos2_5Config()
>>> # Initializing a model (with random weights) from the KOSMOS-2.5-patch14-224 style configuration
>>> model = Kosmos2_5Model(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
```"""
model_type = "kosmos-2.5"
is_composition = True
def __init__(
self,
text_config=None,
vision_config=None,
latent_query_num=2048,
**kwargs,
):
super().__init__(**kwargs)
if text_config is None:
text_config = {}
logger.info("text_config is None. Initializing the Kosmos2_5TextConfig with default values.")
if vision_config is None:
vision_config = {}
logger.info("vision_config is None. Initializing the Kosmos2_5VisionConfig with default values.")
self.text_config = Kosmos2_5TextConfig(**text_config)
self.vision_config = Kosmos2_5VisionConfig(**vision_config)
self.latent_query_num = latent_query_num
@classmethod
def from_text_vision_configs(
cls,
text_config: Kosmos2_5TextConfig,
vision_config: Kosmos2_5VisionConfig,
**kwargs,
):
r"""
Instantiate a [`Pix2StructConfig`] (or a derived class) from pix2struct text model configuration and pix2struct
vision model configuration.
Returns:
[`Pix2StructConfig`]: An instance of a configuration object
"""
return cls(
text_config=text_config.to_dict(),
vision_config=vision_config.to_dict(),
**kwargs,
) |