gmastrapas
commited on
Commit
·
56fe6da
1
Parent(s):
949a53f
feat: initial commit
Browse files- README.md +14 -0
- configuration_clip.py +300 -0
- eva_model.py +763 -0
- hf_model.py +425 -0
- modeling_clip.py +317 -0
- processing_clip.py +66 -0
- rope_embeddings.py +165 -0
- transform.py +458 -0
README.md
ADDED
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Jina CLIP
|
2 |
+
|
3 |
+
The Jina CLIP implementation is hosted in this repository. The model uses:
|
4 |
+
* the EVA 02 architecture for the vision tower
|
5 |
+
* the Jina BERT with Flash Attention model as a text tower
|
6 |
+
|
7 |
+
To use the Jina CLIP model, the following packages are required:
|
8 |
+
* `torch`
|
9 |
+
* `timm`
|
10 |
+
* `transformers`
|
11 |
+
* `einops`
|
12 |
+
* `xformers` to use x-attention
|
13 |
+
* `flash-attn` to use flash attention
|
14 |
+
* `apex` to use fused layer normalization
|
configuration_clip.py
ADDED
@@ -0,0 +1,300 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
#
|
3 |
+
# Code mainly copied from:
|
4 |
+
# https://github.com/huggingface/transformers/blob/main/src/transformers/models/clip/configuration_clip.py
|
5 |
+
# and adjusted for Jina CLIP
|
6 |
+
|
7 |
+
import os
|
8 |
+
from copy import deepcopy
|
9 |
+
from typing import Any, Dict, Optional, Union
|
10 |
+
|
11 |
+
from transformers import PretrainedConfig, logging
|
12 |
+
|
13 |
+
logger = logging.get_logger(__name__)
|
14 |
+
|
15 |
+
|
16 |
+
""" Jina CLIP model configuration """
|
17 |
+
|
18 |
+
|
19 |
+
class JinaCLIPTextConfig(PretrainedConfig):
|
20 |
+
model_type = 'jina_clip_text'
|
21 |
+
|
22 |
+
def __init__(
|
23 |
+
self,
|
24 |
+
embed_dim: int = 768,
|
25 |
+
hf_model_name_or_path: str = 'jinaai/jina-bert-v2-base-en-flash',
|
26 |
+
hf_model_config_kwargs: Optional[Dict[str, Any]] = None,
|
27 |
+
pooler_type: Optional[str] = None,
|
28 |
+
proj_type: Optional[str] = None,
|
29 |
+
proj_bias: bool = False,
|
30 |
+
**kwargs,
|
31 |
+
):
|
32 |
+
super().__init__(**kwargs)
|
33 |
+
|
34 |
+
self.embed_dim = embed_dim
|
35 |
+
self.hf_model_name_or_path = hf_model_name_or_path
|
36 |
+
self.hf_model_config_kwargs = hf_model_config_kwargs or {}
|
37 |
+
self.pooler_type = pooler_type
|
38 |
+
self.proj_type = proj_type
|
39 |
+
self.proj_bias = proj_bias
|
40 |
+
|
41 |
+
@classmethod
|
42 |
+
def from_pretrained(
|
43 |
+
cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs
|
44 |
+
) -> 'PretrainedConfig':
|
45 |
+
cls._set_token_in_kwargs(kwargs)
|
46 |
+
|
47 |
+
configdict, kwargs = cls.get_config_dict(
|
48 |
+
pretrained_model_name_or_path, **kwargs
|
49 |
+
)
|
50 |
+
|
51 |
+
# get the text config dict if we are loading from JinaCLIPConfig
|
52 |
+
if configdict.get('model_type') == 'jina_clip':
|
53 |
+
configdict = configdict['text_config']
|
54 |
+
|
55 |
+
if (
|
56 |
+
'model_type' in configdict
|
57 |
+
and hasattr(cls, 'model_type')
|
58 |
+
and configdict['model_type'] != cls.model_type
|
59 |
+
):
|
60 |
+
logger.warning(
|
61 |
+
f'You are using a model of type {configdict["model_type"]} to '
|
62 |
+
f'instantiate a model of type {cls.model_type}. This is not supported '
|
63 |
+
'for all configurations of models and can yield errors.'
|
64 |
+
)
|
65 |
+
|
66 |
+
return cls.from_dict(configdict, **kwargs)
|
67 |
+
|
68 |
+
|
69 |
+
class JinaCLIPVisionConfig(PretrainedConfig):
|
70 |
+
model_type = 'jina_clip_vision'
|
71 |
+
|
72 |
+
def __init__(
|
73 |
+
self,
|
74 |
+
embed_dim: int = 768,
|
75 |
+
width: int = 768,
|
76 |
+
image_size: int = 224,
|
77 |
+
patch_size: int = 16,
|
78 |
+
layers: int = 12,
|
79 |
+
head_width: int = 64,
|
80 |
+
mlp_ratio: float = 4.0,
|
81 |
+
ls_init_value: Optional[float] = None,
|
82 |
+
patch_dropout: float = 0.0,
|
83 |
+
qkv_bias: bool = True,
|
84 |
+
fused_layer_norm: bool = False,
|
85 |
+
x_attention: bool = False,
|
86 |
+
post_norm: bool = False,
|
87 |
+
rope_embeddings: bool = False,
|
88 |
+
pt_hw_seq_len: int = 16,
|
89 |
+
intp_freq: bool = False,
|
90 |
+
naive_swiglu: bool = False,
|
91 |
+
subln: bool = False,
|
92 |
+
drop_path_rate: float = 0.0,
|
93 |
+
proj_type: Optional[str] = None,
|
94 |
+
**kwargs,
|
95 |
+
):
|
96 |
+
super().__init__(**kwargs)
|
97 |
+
|
98 |
+
self.layers = layers
|
99 |
+
self.embed_dim = embed_dim
|
100 |
+
self.width = width
|
101 |
+
self.head_width = head_width
|
102 |
+
self.mlp_ratio = mlp_ratio
|
103 |
+
self.image_size = image_size
|
104 |
+
self.patch_size = patch_size
|
105 |
+
self.ls_init_value = ls_init_value
|
106 |
+
self.patch_dropout = patch_dropout
|
107 |
+
self.qkv_bias = qkv_bias
|
108 |
+
self.fused_layer_norm = fused_layer_norm
|
109 |
+
self.x_attention = x_attention
|
110 |
+
self.post_norm = post_norm
|
111 |
+
self.rope_embeddings = rope_embeddings
|
112 |
+
self.pt_hw_seq_len = pt_hw_seq_len
|
113 |
+
self.intp_freq = intp_freq
|
114 |
+
self.naive_swiglu = naive_swiglu
|
115 |
+
self.subln = subln
|
116 |
+
self.drop_path_rate = drop_path_rate
|
117 |
+
self.proj_type = proj_type
|
118 |
+
|
119 |
+
@classmethod
|
120 |
+
def from_pretrained(
|
121 |
+
cls, pretrained_model_name_or_path: Union[str, os.PathLike], **kwargs
|
122 |
+
) -> 'PretrainedConfig':
|
123 |
+
cls._set_token_in_kwargs(kwargs)
|
124 |
+
|
125 |
+
configdict, kwargs = cls.get_config_dict(
|
126 |
+
pretrained_model_name_or_path, **kwargs
|
127 |
+
)
|
128 |
+
|
129 |
+
# get the vision config dict if we are loading from JinaCLIPConfig
|
130 |
+
if configdict.get('model_type') == 'jina_clip':
|
131 |
+
configdict = configdict['vision_config']
|
132 |
+
|
133 |
+
if (
|
134 |
+
'model_type' in configdict
|
135 |
+
and hasattr(cls, 'model_type')
|
136 |
+
and configdict['model_type'] != cls.model_type
|
137 |
+
):
|
138 |
+
logger.warning(
|
139 |
+
f'You are using a model of type {configdict["model_type"]} to '
|
140 |
+
f'instantiate a model of type {cls.model_type}. This is not supported '
|
141 |
+
'for all configurations of models and can yield errors.'
|
142 |
+
)
|
143 |
+
|
144 |
+
return cls.from_dict(configdict, **kwargs)
|
145 |
+
|
146 |
+
|
147 |
+
class JinaCLIPConfig(PretrainedConfig):
|
148 |
+
model_type = 'jina_clip'
|
149 |
+
is_composition = True
|
150 |
+
|
151 |
+
def __init__(
|
152 |
+
self,
|
153 |
+
text_config: Optional[Dict] = None,
|
154 |
+
vision_config: Optional[Dict] = None,
|
155 |
+
add_projections: bool = False,
|
156 |
+
projection_dim: int = 768,
|
157 |
+
logit_scale_init_value: float = 2.6592,
|
158 |
+
**kwargs,
|
159 |
+
):
|
160 |
+
# If `_config_dict` exist, we use them for the backward compatibility.
|
161 |
+
# We pop out these 2 attributes before calling `super().__init__` to avoid
|
162 |
+
# them being saved (which causes a lot of confusion!).
|
163 |
+
|
164 |
+
text_config_dict: Optional[Dict] = kwargs.pop('text_config_dict', None)
|
165 |
+
vision_config_dict: Optional[Dict] = kwargs.pop('vision_config_dict', None)
|
166 |
+
|
167 |
+
super().__init__(**kwargs)
|
168 |
+
|
169 |
+
if text_config_dict is not None:
|
170 |
+
if text_config is None:
|
171 |
+
text_config = {}
|
172 |
+
|
173 |
+
# This is the complete result when using `text_config_dict`.
|
174 |
+
_text_config_dict = JinaCLIPTextConfig(**text_config_dict).to_dict()
|
175 |
+
|
176 |
+
# Give a warning if the values exist in both `_text_config_dict` and
|
177 |
+
# `text_config` but being different.
|
178 |
+
for key, value in _text_config_dict.items():
|
179 |
+
if (
|
180 |
+
key in text_config
|
181 |
+
and value != text_config[key]
|
182 |
+
and key not in ['transformers_version']
|
183 |
+
):
|
184 |
+
# If specified in `text_config_dict`
|
185 |
+
if key in text_config_dict:
|
186 |
+
message = (
|
187 |
+
f'`{key}` is found in both `text_config_dict` and '
|
188 |
+
f'`text_config` but with different values. '
|
189 |
+
f'The value `text_config_dict["{key}"]` will be used '
|
190 |
+
f'instead.'
|
191 |
+
)
|
192 |
+
# If inferred from default argument values (
|
193 |
+
# just to be super careful)
|
194 |
+
else:
|
195 |
+
message = (
|
196 |
+
f'`text_config_dict` is provided which will be used to '
|
197 |
+
f'initialize `JinaCLIPTextConfig`. The '
|
198 |
+
f'value `text_config["{key}"]` will be overriden.'
|
199 |
+
)
|
200 |
+
logger.info(message)
|
201 |
+
|
202 |
+
# Update all values in `text_config` with the ones in `_text_config_dict`.
|
203 |
+
text_config.update(_text_config_dict)
|
204 |
+
|
205 |
+
if vision_config_dict is not None:
|
206 |
+
if vision_config is None:
|
207 |
+
vision_config = {}
|
208 |
+
|
209 |
+
# This is the complete result when using `vision_config_dict`.
|
210 |
+
_vision_config_dict = JinaCLIPVisionConfig(**vision_config_dict).to_dict()
|
211 |
+
# convert keys to string instead of integer
|
212 |
+
if 'id2label' in _vision_config_dict:
|
213 |
+
_vision_config_dict['id2label'] = {
|
214 |
+
str(key): value
|
215 |
+
for key, value in _vision_config_dict['id2label'].items()
|
216 |
+
}
|
217 |
+
|
218 |
+
# Give a warning if the values exist in both `_vision_config_dict`
|
219 |
+
# and `vision_config` but being different.
|
220 |
+
for key, value in _vision_config_dict.items():
|
221 |
+
if (
|
222 |
+
key in vision_config
|
223 |
+
and value != vision_config[key]
|
224 |
+
and key not in ['transformers_version']
|
225 |
+
):
|
226 |
+
# If specified in `vision_config_dict`
|
227 |
+
if key in vision_config_dict:
|
228 |
+
message = (
|
229 |
+
f'`{key}` is found in both `vision_config_dict` and '
|
230 |
+
f'`vision_config` but with different '
|
231 |
+
f'values. The value `vision_config_dict["{key}"]` will '
|
232 |
+
f'be used instead.'
|
233 |
+
)
|
234 |
+
# If inferred from default argument values
|
235 |
+
# (just to be super careful)
|
236 |
+
else:
|
237 |
+
message = (
|
238 |
+
f'`vision_config_dict` is provided which will be used to '
|
239 |
+
f'initialize `JinaCLIPVisionConfig`. '
|
240 |
+
f'The value `vision_config["{key}"]` will be overriden.'
|
241 |
+
)
|
242 |
+
logger.info(message)
|
243 |
+
|
244 |
+
# Update all values in `vision_config` with the ones in
|
245 |
+
# `_vision_config_dict`.
|
246 |
+
vision_config.update(_vision_config_dict)
|
247 |
+
|
248 |
+
if text_config is None:
|
249 |
+
text_config = {}
|
250 |
+
logger.info(
|
251 |
+
'`text_config` is `None`. Initializing the `JinaCLIPTextConfig` with '
|
252 |
+
'default values.'
|
253 |
+
)
|
254 |
+
|
255 |
+
if vision_config is None:
|
256 |
+
vision_config = {}
|
257 |
+
logger.info(
|
258 |
+
'`vision_config` is `None`. initializing the `JinaCLIPVisionConfig` '
|
259 |
+
'with default values.'
|
260 |
+
)
|
261 |
+
|
262 |
+
self.text_config = JinaCLIPTextConfig(**text_config)
|
263 |
+
self.vision_config = JinaCLIPVisionConfig(**vision_config)
|
264 |
+
|
265 |
+
self.add_projections = add_projections
|
266 |
+
self.projection_dim = projection_dim
|
267 |
+
self.logit_scale_init_value = logit_scale_init_value
|
268 |
+
self.initializer_factor = 1.0
|
269 |
+
|
270 |
+
if not self.add_projections:
|
271 |
+
if self.text_config.embed_dim != self.vision_config.embed_dim:
|
272 |
+
raise ValueError(
|
273 |
+
'When projections are disabled (`add_projections=False`), text '
|
274 |
+
'and vision towers need to have the same embedding dimensionality. '
|
275 |
+
f'Currently text embedding dim is {self.text_config.embed_dim} != '
|
276 |
+
f'{self.vision_config.embed_dim} of the vision tower. '
|
277 |
+
'Either set the same output dim for both towers, or enable '
|
278 |
+
'projections with `add_projections=True`.'
|
279 |
+
)
|
280 |
+
|
281 |
+
@classmethod
|
282 |
+
def from_text_vision_configs(
|
283 |
+
cls,
|
284 |
+
text_config: JinaCLIPTextConfig,
|
285 |
+
vision_config: JinaCLIPVisionConfig,
|
286 |
+
**kwargs,
|
287 |
+
):
|
288 |
+
return cls(
|
289 |
+
text_config=text_config.to_dict(),
|
290 |
+
vision_config=vision_config.to_dict(),
|
291 |
+
projection_dim=text_config.projection_dim,
|
292 |
+
**kwargs,
|
293 |
+
)
|
294 |
+
|
295 |
+
def to_dict(self):
|
296 |
+
output = deepcopy(self.__dict__)
|
297 |
+
output['text_config'] = self.text_config.to_dict()
|
298 |
+
output['vision_config'] = self.vision_config.to_dict()
|
299 |
+
output['model_type'] = self.__class__.model_type
|
300 |
+
return output
|
eva_model.py
ADDED
@@ -0,0 +1,763 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# --------------------------------------------------------
|
2 |
+
# Adapted from EVA CLIP
|
3 |
+
# https://github.com/baaivision/EVA/tree/master/EVA-CLIP/rei/eva_clip
|
4 |
+
# --------------------------------------------------------
|
5 |
+
|
6 |
+
import math
|
7 |
+
import os
|
8 |
+
from functools import partial
|
9 |
+
|
10 |
+
import torch
|
11 |
+
import torch.nn as nn
|
12 |
+
import torch.nn.functional as F
|
13 |
+
|
14 |
+
try:
|
15 |
+
from timm.models.layers import drop_path, to_2tuple, trunc_normal_
|
16 |
+
except ImportError or ModuleNotFoundError:
|
17 |
+
from timm.layers import drop_path, to_2tuple, trunc_normal_
|
18 |
+
|
19 |
+
from .rope_embeddings import VisionRotaryEmbeddingFast
|
20 |
+
|
21 |
+
if os.getenv('ENV_TYPE') == 'deepspeed':
|
22 |
+
try:
|
23 |
+
from deepspeed.runtime.activation_checkpointing.checkpointing import checkpoint
|
24 |
+
except ImportError or ModuleNotFoundError:
|
25 |
+
from torch.utils.checkpoint import checkpoint
|
26 |
+
else:
|
27 |
+
from torch.utils.checkpoint import checkpoint
|
28 |
+
|
29 |
+
try:
|
30 |
+
import xformers.ops as xops
|
31 |
+
except ImportError:
|
32 |
+
xops = None
|
33 |
+
|
34 |
+
|
35 |
+
class PatchDropout(nn.Module):
|
36 |
+
"""
|
37 |
+
https://arxiv.org/abs/2212.00794
|
38 |
+
"""
|
39 |
+
|
40 |
+
def __init__(self, prob, exclude_first_token=True):
|
41 |
+
super().__init__()
|
42 |
+
assert 0 <= prob < 1.0
|
43 |
+
self.prob = prob
|
44 |
+
self.exclude_first_token = exclude_first_token # exclude CLS token
|
45 |
+
|
46 |
+
def forward(self, x):
|
47 |
+
if not self.training or self.prob == 0.0:
|
48 |
+
return x
|
49 |
+
|
50 |
+
if self.exclude_first_token:
|
51 |
+
cls_tokens, x = x[:, :1], x[:, 1:]
|
52 |
+
else:
|
53 |
+
cls_tokens = torch.jit.annotate(torch.Tensor, x[:, :1])
|
54 |
+
|
55 |
+
batch = x.size()[0]
|
56 |
+
num_tokens = x.size()[1]
|
57 |
+
|
58 |
+
batch_indices = torch.arange(batch)
|
59 |
+
batch_indices = batch_indices[..., None]
|
60 |
+
|
61 |
+
keep_prob = 1 - self.prob
|
62 |
+
num_patches_keep = max(1, int(num_tokens * keep_prob))
|
63 |
+
|
64 |
+
rand = torch.randn(batch, num_tokens)
|
65 |
+
patch_indices_keep = rand.topk(num_patches_keep, dim=-1).indices
|
66 |
+
|
67 |
+
x = x[batch_indices, patch_indices_keep]
|
68 |
+
|
69 |
+
if self.exclude_first_token:
|
70 |
+
x = torch.cat((cls_tokens, x), dim=1)
|
71 |
+
|
72 |
+
return x, patch_indices_keep
|
73 |
+
|
74 |
+
|
75 |
+
class DropPath(nn.Module):
|
76 |
+
"""Drop paths (Stochastic Depth) per sample (when applied in main path of
|
77 |
+
residual blocks)."""
|
78 |
+
|
79 |
+
def __init__(self, drop_prob=None):
|
80 |
+
super(DropPath, self).__init__()
|
81 |
+
self.drop_prob = drop_prob
|
82 |
+
|
83 |
+
def forward(self, x):
|
84 |
+
return drop_path(x, self.drop_prob, self.training)
|
85 |
+
|
86 |
+
def extra_repr(self) -> str:
|
87 |
+
return 'p={}'.format(self.drop_prob)
|
88 |
+
|
89 |
+
|
90 |
+
class Mlp(nn.Module):
|
91 |
+
def __init__(
|
92 |
+
self,
|
93 |
+
in_features,
|
94 |
+
hidden_features=None,
|
95 |
+
out_features=None,
|
96 |
+
act_layer=nn.GELU,
|
97 |
+
norm_layer=nn.LayerNorm,
|
98 |
+
drop=0.0,
|
99 |
+
subln=False,
|
100 |
+
):
|
101 |
+
super().__init__()
|
102 |
+
out_features = out_features or in_features
|
103 |
+
hidden_features = hidden_features or in_features
|
104 |
+
self.fc1 = nn.Linear(in_features, hidden_features)
|
105 |
+
self.act = act_layer()
|
106 |
+
|
107 |
+
self.ffn_ln = norm_layer(hidden_features) if subln else nn.Identity()
|
108 |
+
|
109 |
+
self.fc2 = nn.Linear(hidden_features, out_features)
|
110 |
+
self.drop = nn.Dropout(drop)
|
111 |
+
|
112 |
+
def forward(self, x):
|
113 |
+
x = self.fc1(x)
|
114 |
+
x = self.act(x)
|
115 |
+
# x = self.drop(x)
|
116 |
+
# commit this for the orignal BERT implement
|
117 |
+
x = self.ffn_ln(x)
|
118 |
+
|
119 |
+
x = self.fc2(x)
|
120 |
+
x = self.drop(x)
|
121 |
+
return x
|
122 |
+
|
123 |
+
|
124 |
+
class SwiGLU(nn.Module):
|
125 |
+
def __init__(
|
126 |
+
self,
|
127 |
+
in_features,
|
128 |
+
hidden_features=None,
|
129 |
+
out_features=None,
|
130 |
+
act_layer=nn.SiLU,
|
131 |
+
drop=0.0,
|
132 |
+
norm_layer=nn.LayerNorm,
|
133 |
+
subln=False,
|
134 |
+
):
|
135 |
+
super().__init__()
|
136 |
+
out_features = out_features or in_features
|
137 |
+
hidden_features = hidden_features or in_features
|
138 |
+
|
139 |
+
self.w1 = nn.Linear(in_features, hidden_features)
|
140 |
+
self.w2 = nn.Linear(in_features, hidden_features)
|
141 |
+
|
142 |
+
self.act = act_layer()
|
143 |
+
self.ffn_ln = norm_layer(hidden_features) if subln else nn.Identity()
|
144 |
+
self.w3 = nn.Linear(hidden_features, out_features)
|
145 |
+
|
146 |
+
self.drop = nn.Dropout(drop)
|
147 |
+
|
148 |
+
def forward(self, x):
|
149 |
+
x1 = self.w1(x)
|
150 |
+
x2 = self.w2(x)
|
151 |
+
hidden = self.act(x1) * x2
|
152 |
+
x = self.ffn_ln(hidden)
|
153 |
+
x = self.w3(x)
|
154 |
+
x = self.drop(x)
|
155 |
+
return x
|
156 |
+
|
157 |
+
|
158 |
+
class Attention(nn.Module):
|
159 |
+
def __init__(
|
160 |
+
self,
|
161 |
+
dim,
|
162 |
+
num_heads=8,
|
163 |
+
qkv_bias=False,
|
164 |
+
qk_scale=None,
|
165 |
+
attn_drop=0.0,
|
166 |
+
proj_drop=0.0,
|
167 |
+
window_size=None,
|
168 |
+
attn_head_dim=None,
|
169 |
+
xattn=False,
|
170 |
+
rope=None,
|
171 |
+
subln=False,
|
172 |
+
norm_layer=nn.LayerNorm,
|
173 |
+
):
|
174 |
+
super().__init__()
|
175 |
+
self.num_heads = num_heads
|
176 |
+
head_dim = dim // num_heads
|
177 |
+
if attn_head_dim is not None:
|
178 |
+
head_dim = attn_head_dim
|
179 |
+
all_head_dim = head_dim * self.num_heads
|
180 |
+
self.scale = qk_scale or head_dim**-0.5
|
181 |
+
|
182 |
+
self.subln = subln
|
183 |
+
if self.subln:
|
184 |
+
self.q_proj = nn.Linear(dim, all_head_dim, bias=False)
|
185 |
+
self.k_proj = nn.Linear(dim, all_head_dim, bias=False)
|
186 |
+
self.v_proj = nn.Linear(dim, all_head_dim, bias=False)
|
187 |
+
else:
|
188 |
+
self.qkv = nn.Linear(dim, all_head_dim * 3, bias=False)
|
189 |
+
|
190 |
+
if qkv_bias:
|
191 |
+
self.q_bias = nn.Parameter(torch.zeros(all_head_dim))
|
192 |
+
self.v_bias = nn.Parameter(torch.zeros(all_head_dim))
|
193 |
+
else:
|
194 |
+
self.q_bias = None
|
195 |
+
self.v_bias = None
|
196 |
+
|
197 |
+
if window_size:
|
198 |
+
self.window_size = window_size
|
199 |
+
self.num_relative_distance = (2 * window_size[0] - 1) * (
|
200 |
+
2 * window_size[1] - 1
|
201 |
+
) + 3
|
202 |
+
self.relative_position_bias_table = nn.Parameter(
|
203 |
+
torch.zeros(self.num_relative_distance, num_heads)
|
204 |
+
) # 2*Wh-1 * 2*Ww-1, nH
|
205 |
+
# cls to token & token 2 cls & cls to cls
|
206 |
+
|
207 |
+
# get pair-wise relative position index for each token inside the window
|
208 |
+
coords_h = torch.arange(window_size[0])
|
209 |
+
coords_w = torch.arange(window_size[1])
|
210 |
+
coords = torch.stack(torch.meshgrid([coords_h, coords_w])) # 2, Wh, Ww
|
211 |
+
coords_flatten = torch.flatten(coords, 1) # 2, Wh*Ww
|
212 |
+
relative_coords = (
|
213 |
+
coords_flatten[:, :, None] - coords_flatten[:, None, :]
|
214 |
+
) # 2, Wh*Ww, Wh*Ww
|
215 |
+
relative_coords = relative_coords.permute(
|
216 |
+
1, 2, 0
|
217 |
+
).contiguous() # Wh*Ww, Wh*Ww, 2
|
218 |
+
relative_coords[:, :, 0] += window_size[0] - 1 # shift to start from 0
|
219 |
+
relative_coords[:, :, 1] += window_size[1] - 1
|
220 |
+
relative_coords[:, :, 0] *= 2 * window_size[1] - 1
|
221 |
+
relative_position_index = torch.zeros(
|
222 |
+
size=(window_size[0] * window_size[1] + 1,) * 2,
|
223 |
+
dtype=relative_coords.dtype,
|
224 |
+
)
|
225 |
+
relative_position_index[1:, 1:] = relative_coords.sum(-1) # Wh*Ww, Wh*Ww
|
226 |
+
relative_position_index[0, 0:] = self.num_relative_distance - 3
|
227 |
+
relative_position_index[0:, 0] = self.num_relative_distance - 2
|
228 |
+
relative_position_index[0, 0] = self.num_relative_distance - 1
|
229 |
+
|
230 |
+
self.register_buffer('relative_position_index', relative_position_index)
|
231 |
+
else:
|
232 |
+
self.window_size = None
|
233 |
+
self.relative_position_bias_table = None
|
234 |
+
self.relative_position_index = None
|
235 |
+
|
236 |
+
self.attn_drop = nn.Dropout(attn_drop)
|
237 |
+
self.inner_attn_ln = norm_layer(all_head_dim) if subln else nn.Identity()
|
238 |
+
# self.proj = nn.Linear(all_head_dim, all_head_dim)
|
239 |
+
self.proj = nn.Linear(all_head_dim, dim)
|
240 |
+
self.proj_drop = nn.Dropout(proj_drop)
|
241 |
+
self.xattn = xattn
|
242 |
+
self.xattn_drop = attn_drop
|
243 |
+
|
244 |
+
self.rope = rope
|
245 |
+
|
246 |
+
def forward(self, x, rel_pos_bias=None, attn_mask=None):
|
247 |
+
B, N, C = x.shape
|
248 |
+
if self.subln:
|
249 |
+
q = F.linear(input=x, weight=self.q_proj.weight, bias=self.q_bias)
|
250 |
+
k = F.linear(input=x, weight=self.k_proj.weight, bias=None)
|
251 |
+
v = F.linear(input=x, weight=self.v_proj.weight, bias=self.v_bias)
|
252 |
+
|
253 |
+
q = q.reshape(B, N, self.num_heads, -1).permute(
|
254 |
+
0, 2, 1, 3
|
255 |
+
) # B, num_heads, N, C
|
256 |
+
k = k.reshape(B, N, self.num_heads, -1).permute(0, 2, 1, 3)
|
257 |
+
v = v.reshape(B, N, self.num_heads, -1).permute(0, 2, 1, 3)
|
258 |
+
else:
|
259 |
+
qkv_bias = None
|
260 |
+
if self.q_bias is not None:
|
261 |
+
qkv_bias = torch.cat(
|
262 |
+
(
|
263 |
+
self.q_bias,
|
264 |
+
torch.zeros_like(self.v_bias, requires_grad=False),
|
265 |
+
self.v_bias,
|
266 |
+
)
|
267 |
+
)
|
268 |
+
|
269 |
+
qkv = F.linear(input=x, weight=self.qkv.weight, bias=qkv_bias)
|
270 |
+
qkv = qkv.reshape(B, N, 3, self.num_heads, -1).permute(
|
271 |
+
2, 0, 3, 1, 4
|
272 |
+
) # 3, B, num_heads, N, C
|
273 |
+
q, k, v = qkv[0], qkv[1], qkv[2]
|
274 |
+
|
275 |
+
if self.rope:
|
276 |
+
# slightly fast impl
|
277 |
+
q_t = q[:, :, 1:, :]
|
278 |
+
ro_q_t = self.rope(q_t)
|
279 |
+
q = torch.cat((q[:, :, :1, :], ro_q_t), -2).type_as(v)
|
280 |
+
|
281 |
+
k_t = k[:, :, 1:, :]
|
282 |
+
ro_k_t = self.rope(k_t)
|
283 |
+
k = torch.cat((k[:, :, :1, :], ro_k_t), -2).type_as(v)
|
284 |
+
|
285 |
+
if self.xattn:
|
286 |
+
if xops is None:
|
287 |
+
raise ValueError(
|
288 |
+
"Can't use xattn without xformers. Please 'pip install xformers'"
|
289 |
+
)
|
290 |
+
q = q.permute(0, 2, 1, 3) # B, num_heads, N, C -> B, N, num_heads, C
|
291 |
+
k = k.permute(0, 2, 1, 3)
|
292 |
+
v = v.permute(0, 2, 1, 3)
|
293 |
+
|
294 |
+
x = xops.memory_efficient_attention(
|
295 |
+
q,
|
296 |
+
k,
|
297 |
+
v,
|
298 |
+
p=self.xattn_drop,
|
299 |
+
scale=self.scale,
|
300 |
+
)
|
301 |
+
x = x.reshape(B, N, -1)
|
302 |
+
x = self.inner_attn_ln(x)
|
303 |
+
x = self.proj(x)
|
304 |
+
x = self.proj_drop(x)
|
305 |
+
else:
|
306 |
+
q = q * self.scale
|
307 |
+
attn = q @ k.transpose(-2, -1)
|
308 |
+
|
309 |
+
if self.relative_position_bias_table is not None:
|
310 |
+
relative_position_bias = self.relative_position_bias_table[
|
311 |
+
self.relative_position_index.view(-1)
|
312 |
+
].view(
|
313 |
+
self.window_size[0] * self.window_size[1] + 1,
|
314 |
+
self.window_size[0] * self.window_size[1] + 1,
|
315 |
+
-1,
|
316 |
+
) # Wh*Ww,Wh*Ww,nH
|
317 |
+
relative_position_bias = relative_position_bias.permute(
|
318 |
+
2, 0, 1
|
319 |
+
).contiguous() # nH, Wh*Ww, Wh*Ww
|
320 |
+
attn = attn + relative_position_bias.unsqueeze(0).type_as(attn)
|
321 |
+
|
322 |
+
if rel_pos_bias is not None:
|
323 |
+
attn = attn + rel_pos_bias.type_as(attn)
|
324 |
+
|
325 |
+
if attn_mask is not None:
|
326 |
+
attn_mask = attn_mask.bool()
|
327 |
+
attn = attn.masked_fill(~attn_mask[:, None, None, :], float('-inf'))
|
328 |
+
|
329 |
+
attn = attn.softmax(dim=-1)
|
330 |
+
attn = self.attn_drop(attn)
|
331 |
+
|
332 |
+
x = (attn @ v).transpose(1, 2).reshape(B, N, -1)
|
333 |
+
x = self.inner_attn_ln(x)
|
334 |
+
x = self.proj(x)
|
335 |
+
x = self.proj_drop(x)
|
336 |
+
return x
|
337 |
+
|
338 |
+
|
339 |
+
class Block(nn.Module):
|
340 |
+
def __init__(
|
341 |
+
self,
|
342 |
+
dim,
|
343 |
+
num_heads,
|
344 |
+
mlp_ratio=4.0,
|
345 |
+
qkv_bias=False,
|
346 |
+
qk_scale=None,
|
347 |
+
drop=0.0,
|
348 |
+
attn_drop=0.0,
|
349 |
+
drop_path=0.0,
|
350 |
+
init_values=None,
|
351 |
+
act_layer=nn.GELU,
|
352 |
+
norm_layer=nn.LayerNorm,
|
353 |
+
window_size=None,
|
354 |
+
attn_head_dim=None,
|
355 |
+
xattn=False,
|
356 |
+
rope=None,
|
357 |
+
postnorm=False,
|
358 |
+
subln=False,
|
359 |
+
naiveswiglu=False,
|
360 |
+
):
|
361 |
+
super().__init__()
|
362 |
+
self.norm1 = norm_layer(dim)
|
363 |
+
self.attn = Attention(
|
364 |
+
dim,
|
365 |
+
num_heads=num_heads,
|
366 |
+
qkv_bias=qkv_bias,
|
367 |
+
qk_scale=qk_scale,
|
368 |
+
attn_drop=attn_drop,
|
369 |
+
proj_drop=drop,
|
370 |
+
window_size=window_size,
|
371 |
+
attn_head_dim=attn_head_dim,
|
372 |
+
xattn=xattn,
|
373 |
+
rope=rope,
|
374 |
+
subln=subln,
|
375 |
+
norm_layer=norm_layer,
|
376 |
+
)
|
377 |
+
# NOTE: drop path for stochastic depth, we shall see if this is better
|
378 |
+
# than dropout here
|
379 |
+
self.drop_path = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()
|
380 |
+
self.norm2 = norm_layer(dim)
|
381 |
+
mlp_hidden_dim = int(dim * mlp_ratio)
|
382 |
+
|
383 |
+
if naiveswiglu:
|
384 |
+
self.mlp = SwiGLU(
|
385 |
+
in_features=dim,
|
386 |
+
hidden_features=mlp_hidden_dim,
|
387 |
+
subln=subln,
|
388 |
+
norm_layer=norm_layer,
|
389 |
+
)
|
390 |
+
else:
|
391 |
+
self.mlp = Mlp(
|
392 |
+
in_features=dim,
|
393 |
+
hidden_features=mlp_hidden_dim,
|
394 |
+
act_layer=act_layer,
|
395 |
+
subln=subln,
|
396 |
+
drop=drop,
|
397 |
+
)
|
398 |
+
|
399 |
+
if init_values is not None and init_values > 0:
|
400 |
+
self.gamma_1 = nn.Parameter(
|
401 |
+
init_values * torch.ones((dim,)), requires_grad=True
|
402 |
+
)
|
403 |
+
self.gamma_2 = nn.Parameter(
|
404 |
+
init_values * torch.ones((dim,)), requires_grad=True
|
405 |
+
)
|
406 |
+
else:
|
407 |
+
self.gamma_1, self.gamma_2 = None, None
|
408 |
+
|
409 |
+
self.postnorm = postnorm
|
410 |
+
|
411 |
+
def forward(self, x, rel_pos_bias=None, attn_mask=None):
|
412 |
+
if self.gamma_1 is None:
|
413 |
+
if self.postnorm:
|
414 |
+
x = x + self.drop_path(
|
415 |
+
self.norm1(
|
416 |
+
self.attn(x, rel_pos_bias=rel_pos_bias, attn_mask=attn_mask)
|
417 |
+
)
|
418 |
+
)
|
419 |
+
x = x + self.drop_path(self.norm2(self.mlp(x)))
|
420 |
+
else:
|
421 |
+
x = x + self.drop_path(
|
422 |
+
self.attn(
|
423 |
+
self.norm1(x), rel_pos_bias=rel_pos_bias, attn_mask=attn_mask
|
424 |
+
)
|
425 |
+
)
|
426 |
+
x = x + self.drop_path(self.mlp(self.norm2(x)))
|
427 |
+
else:
|
428 |
+
if self.postnorm:
|
429 |
+
x = x + self.drop_path(
|
430 |
+
self.gamma_1
|
431 |
+
* self.norm1(
|
432 |
+
self.attn(x, rel_pos_bias=rel_pos_bias, attn_mask=attn_mask)
|
433 |
+
)
|
434 |
+
)
|
435 |
+
x = x + self.drop_path(self.gamma_2 * self.norm2(self.mlp(x)))
|
436 |
+
else:
|
437 |
+
x = x + self.drop_path(
|
438 |
+
self.gamma_1
|
439 |
+
* self.attn(
|
440 |
+
self.norm1(x), rel_pos_bias=rel_pos_bias, attn_mask=attn_mask
|
441 |
+
)
|
442 |
+
)
|
443 |
+
x = x + self.drop_path(self.gamma_2 * self.mlp(self.norm2(x)))
|
444 |
+
return x
|
445 |
+
|
446 |
+
|
447 |
+
class PatchEmbed(nn.Module):
|
448 |
+
"""Image to Patch Embedding"""
|
449 |
+
|
450 |
+
def __init__(self, img_size=224, patch_size=16, in_chans=3, embed_dim=768):
|
451 |
+
super().__init__()
|
452 |
+
img_size = to_2tuple(img_size)
|
453 |
+
patch_size = to_2tuple(patch_size)
|
454 |
+
num_patches = (img_size[1] // patch_size[1]) * (img_size[0] // patch_size[0])
|
455 |
+
self.patch_shape = (img_size[0] // patch_size[0], img_size[1] // patch_size[1])
|
456 |
+
self.img_size = img_size
|
457 |
+
self.patch_size = patch_size
|
458 |
+
self.num_patches = num_patches
|
459 |
+
|
460 |
+
self.proj = nn.Conv2d(
|
461 |
+
in_chans, embed_dim, kernel_size=patch_size, stride=patch_size
|
462 |
+
)
|
463 |
+
|
464 |
+
def forward(self, x, **kwargs):
|
465 |
+
B, C, H, W = x.shape
|
466 |
+
# FIXME look at relaxing size constraints
|
467 |
+
assert H == self.img_size[0] and W == self.img_size[1], (
|
468 |
+
f"Input image size ({H}*{W}) doesn't match model "
|
469 |
+
f'({self.img_size[0]}*{self.img_size[1]}).'
|
470 |
+
)
|
471 |
+
x = self.proj(x).flatten(2).transpose(1, 2)
|
472 |
+
return x
|
473 |
+
|
474 |
+
|
475 |
+
class RelativePositionBias(nn.Module):
|
476 |
+
def __init__(self, window_size, num_heads):
|
477 |
+
super().__init__()
|
478 |
+
self.window_size = window_size
|
479 |
+
self.num_relative_distance = (2 * window_size[0] - 1) * (
|
480 |
+
2 * window_size[1] - 1
|
481 |
+
) + 3
|
482 |
+
self.relative_position_bias_table = nn.Parameter(
|
483 |
+
torch.zeros(self.num_relative_distance, num_heads)
|
484 |
+
) # 2*Wh-1 * 2*Ww-1, nH
|
485 |
+
# cls to token & token 2 cls & cls to cls
|
486 |
+
|
487 |
+
# get pair-wise relative position index for each token inside the window
|
488 |
+
coords_h = torch.arange(window_size[0])
|
489 |
+
coords_w = torch.arange(window_size[1])
|
490 |
+
coords = torch.stack(torch.meshgrid([coords_h, coords_w])) # 2, Wh, Ww
|
491 |
+
coords_flatten = torch.flatten(coords, 1) # 2, Wh*Ww
|
492 |
+
relative_coords = (
|
493 |
+
coords_flatten[:, :, None] - coords_flatten[:, None, :]
|
494 |
+
) # 2, Wh*Ww, Wh*Ww
|
495 |
+
relative_coords = relative_coords.permute(
|
496 |
+
1, 2, 0
|
497 |
+
).contiguous() # Wh*Ww, Wh*Ww, 2
|
498 |
+
relative_coords[:, :, 0] += window_size[0] - 1 # shift to start from 0
|
499 |
+
relative_coords[:, :, 1] += window_size[1] - 1
|
500 |
+
relative_coords[:, :, 0] *= 2 * window_size[1] - 1
|
501 |
+
relative_position_index = torch.zeros(
|
502 |
+
size=(window_size[0] * window_size[1] + 1,) * 2, dtype=relative_coords.dtype
|
503 |
+
)
|
504 |
+
relative_position_index[1:, 1:] = relative_coords.sum(-1) # Wh*Ww, Wh*Ww
|
505 |
+
relative_position_index[0, 0:] = self.num_relative_distance - 3
|
506 |
+
relative_position_index[0:, 0] = self.num_relative_distance - 2
|
507 |
+
relative_position_index[0, 0] = self.num_relative_distance - 1
|
508 |
+
|
509 |
+
self.register_buffer('relative_position_index', relative_position_index)
|
510 |
+
|
511 |
+
def forward(self):
|
512 |
+
relative_position_bias = self.relative_position_bias_table[
|
513 |
+
self.relative_position_index.view(-1)
|
514 |
+
].view(
|
515 |
+
self.window_size[0] * self.window_size[1] + 1,
|
516 |
+
self.window_size[0] * self.window_size[1] + 1,
|
517 |
+
-1,
|
518 |
+
) # Wh*Ww,Wh*Ww,nH
|
519 |
+
return relative_position_bias.permute(2, 0, 1).contiguous() # nH, Wh*Ww, Wh*Ww
|
520 |
+
|
521 |
+
|
522 |
+
class EVAVisionTransformer(nn.Module):
|
523 |
+
"""Vision Transformer with support for patch or hybrid CNN input stage"""
|
524 |
+
|
525 |
+
def __init__(
|
526 |
+
self,
|
527 |
+
img_size=224,
|
528 |
+
patch_size=16,
|
529 |
+
in_chans=3,
|
530 |
+
num_classes=0,
|
531 |
+
embed_dim=768,
|
532 |
+
depth=12,
|
533 |
+
num_heads=12,
|
534 |
+
mlp_ratio=4.0,
|
535 |
+
qkv_bias=False,
|
536 |
+
qk_scale=None,
|
537 |
+
drop_rate=0.0,
|
538 |
+
attn_drop_rate=0.0,
|
539 |
+
drop_path_rate=0.0,
|
540 |
+
norm_layer=nn.LayerNorm,
|
541 |
+
init_values=None,
|
542 |
+
patch_dropout=0.0,
|
543 |
+
use_abs_pos_emb=True,
|
544 |
+
use_rel_pos_bias=False,
|
545 |
+
use_shared_rel_pos_bias=False,
|
546 |
+
rope=False,
|
547 |
+
use_mean_pooling=True,
|
548 |
+
init_scale=0.001,
|
549 |
+
grad_checkpointing=False,
|
550 |
+
xattn=False,
|
551 |
+
postnorm=False,
|
552 |
+
pt_hw_seq_len=16,
|
553 |
+
intp_freq=False,
|
554 |
+
naiveswiglu=False,
|
555 |
+
subln=False,
|
556 |
+
proj_type=None,
|
557 |
+
):
|
558 |
+
super().__init__()
|
559 |
+
self.image_size = img_size
|
560 |
+
self.num_classes = num_classes
|
561 |
+
self.num_features = (
|
562 |
+
self.embed_dim
|
563 |
+
) = embed_dim # num_features for consistency with other models
|
564 |
+
|
565 |
+
self.patch_embed = PatchEmbed(
|
566 |
+
img_size=img_size,
|
567 |
+
patch_size=patch_size,
|
568 |
+
in_chans=in_chans,
|
569 |
+
embed_dim=embed_dim,
|
570 |
+
)
|
571 |
+
num_patches = self.patch_embed.num_patches
|
572 |
+
|
573 |
+
self.cls_token = nn.Parameter(torch.zeros(1, 1, embed_dim))
|
574 |
+
# self.mask_token = nn.Parameter(torch.zeros(1, 1, embed_dim))
|
575 |
+
if use_abs_pos_emb:
|
576 |
+
self.pos_embed = nn.Parameter(torch.zeros(1, num_patches + 1, embed_dim))
|
577 |
+
else:
|
578 |
+
self.pos_embed = None
|
579 |
+
self.pos_drop = nn.Dropout(p=drop_rate)
|
580 |
+
|
581 |
+
if use_shared_rel_pos_bias:
|
582 |
+
self.rel_pos_bias = RelativePositionBias(
|
583 |
+
window_size=self.patch_embed.patch_shape, num_heads=num_heads
|
584 |
+
)
|
585 |
+
else:
|
586 |
+
self.rel_pos_bias = None
|
587 |
+
|
588 |
+
if rope:
|
589 |
+
half_head_dim = embed_dim // num_heads // 2
|
590 |
+
hw_seq_len = img_size // patch_size
|
591 |
+
self.rope = VisionRotaryEmbeddingFast(
|
592 |
+
dim=half_head_dim,
|
593 |
+
pt_seq_len=pt_hw_seq_len,
|
594 |
+
ft_seq_len=hw_seq_len if intp_freq else None,
|
595 |
+
patch_dropout=patch_dropout,
|
596 |
+
)
|
597 |
+
else:
|
598 |
+
self.rope = None
|
599 |
+
|
600 |
+
self.naiveswiglu = naiveswiglu
|
601 |
+
|
602 |
+
dpr = [
|
603 |
+
x.item() for x in torch.linspace(0, drop_path_rate, depth)
|
604 |
+
] # stochastic depth decay rule
|
605 |
+
self.use_rel_pos_bias = use_rel_pos_bias
|
606 |
+
self.blocks = nn.ModuleList(
|
607 |
+
[
|
608 |
+
Block(
|
609 |
+
dim=embed_dim,
|
610 |
+
num_heads=num_heads,
|
611 |
+
mlp_ratio=mlp_ratio,
|
612 |
+
qkv_bias=qkv_bias,
|
613 |
+
qk_scale=qk_scale,
|
614 |
+
drop=drop_rate,
|
615 |
+
attn_drop=attn_drop_rate,
|
616 |
+
drop_path=dpr[i],
|
617 |
+
norm_layer=norm_layer,
|
618 |
+
init_values=init_values,
|
619 |
+
window_size=self.patch_embed.patch_shape
|
620 |
+
if use_rel_pos_bias
|
621 |
+
else None,
|
622 |
+
xattn=xattn,
|
623 |
+
rope=self.rope,
|
624 |
+
postnorm=postnorm,
|
625 |
+
subln=subln,
|
626 |
+
naiveswiglu=naiveswiglu,
|
627 |
+
)
|
628 |
+
for i in range(depth)
|
629 |
+
]
|
630 |
+
)
|
631 |
+
self.norm = nn.Identity() if use_mean_pooling else norm_layer(embed_dim)
|
632 |
+
self.fc_norm = norm_layer(embed_dim) if use_mean_pooling else None
|
633 |
+
if (num_classes == embed_dim) and (proj_type is None):
|
634 |
+
self.head = nn.Identity()
|
635 |
+
elif proj_type == 'linear':
|
636 |
+
self.head = nn.Linear(embed_dim, num_classes, bias=qkv_bias)
|
637 |
+
elif proj_type == 'mlp':
|
638 |
+
hidden_size = (embed_dim + num_classes) // 2
|
639 |
+
self.proj = nn.Sequential(
|
640 |
+
nn.Linear(embed_dim, hidden_size, bias=qkv_bias),
|
641 |
+
nn.GELU(),
|
642 |
+
nn.Linear(hidden_size, num_classes, bias=qkv_bias),
|
643 |
+
)
|
644 |
+
|
645 |
+
if self.pos_embed is not None:
|
646 |
+
trunc_normal_(self.pos_embed, std=0.02)
|
647 |
+
|
648 |
+
trunc_normal_(self.cls_token, std=0.02)
|
649 |
+
|
650 |
+
self.apply(self._init_weights)
|
651 |
+
self.fix_init_weight()
|
652 |
+
|
653 |
+
if isinstance(self.head, nn.Linear):
|
654 |
+
trunc_normal_(self.head.weight, std=0.02)
|
655 |
+
self.head.weight.data.mul_(init_scale)
|
656 |
+
if qkv_bias:
|
657 |
+
self.head.bias.data.mul_(init_scale)
|
658 |
+
|
659 |
+
# setting a patch_dropout of 0. would mean it is disabled and this function
|
660 |
+
# would be the identity fn
|
661 |
+
self.patch_dropout = (
|
662 |
+
PatchDropout(patch_dropout) if patch_dropout > 0.0 else nn.Identity()
|
663 |
+
)
|
664 |
+
|
665 |
+
self.grad_checkpointing = grad_checkpointing
|
666 |
+
|
667 |
+
def fix_init_weight(self):
|
668 |
+
def rescale(param, layer_id):
|
669 |
+
param.div_(math.sqrt(2.0 * layer_id))
|
670 |
+
|
671 |
+
for layer_id, layer in enumerate(self.blocks):
|
672 |
+
rescale(layer.attn.proj.weight.data, layer_id + 1)
|
673 |
+
if self.naiveswiglu:
|
674 |
+
rescale(layer.mlp.w3.weight.data, layer_id + 1)
|
675 |
+
else:
|
676 |
+
rescale(layer.mlp.fc2.weight.data, layer_id + 1)
|
677 |
+
|
678 |
+
def get_cast_dtype(self) -> torch.dtype:
|
679 |
+
return self.blocks[0].mlp.fc2.weight.dtype
|
680 |
+
|
681 |
+
def _init_weights(self, m):
|
682 |
+
if isinstance(m, nn.Linear):
|
683 |
+
trunc_normal_(m.weight, std=0.02)
|
684 |
+
if m.bias is not None:
|
685 |
+
nn.init.constant_(m.bias, 0)
|
686 |
+
elif isinstance(m, nn.LayerNorm):
|
687 |
+
nn.init.constant_(m.bias, 0)
|
688 |
+
nn.init.constant_(m.weight, 1.0)
|
689 |
+
|
690 |
+
def get_num_layers(self):
|
691 |
+
return len(self.blocks)
|
692 |
+
|
693 |
+
def lock(self, unlocked_groups=0, freeze_bn_stats=False):
|
694 |
+
assert (
|
695 |
+
unlocked_groups == 0
|
696 |
+
), 'partial locking not currently supported for this model'
|
697 |
+
for param in self.parameters():
|
698 |
+
param.requires_grad = False
|
699 |
+
|
700 |
+
@torch.jit.ignore
|
701 |
+
def set_grad_checkpointing(self, enable=True):
|
702 |
+
self.grad_checkpointing = enable
|
703 |
+
|
704 |
+
@torch.jit.ignore
|
705 |
+
def no_weight_decay(self):
|
706 |
+
return {'pos_embed', 'cls_token'}
|
707 |
+
|
708 |
+
def get_classifier(self):
|
709 |
+
return self.head
|
710 |
+
|
711 |
+
def reset_classifier(self, num_classes, global_pool=''):
|
712 |
+
self.num_classes = num_classes
|
713 |
+
self.head = (
|
714 |
+
nn.Linear(self.embed_dim, num_classes) if num_classes > 0 else nn.Identity()
|
715 |
+
)
|
716 |
+
|
717 |
+
def forward_features(self, x, return_all_features=False):
|
718 |
+
x = self.patch_embed(x)
|
719 |
+
batch_size, seq_len, _ = x.size()
|
720 |
+
|
721 |
+
cls_tokens = self.cls_token.expand(
|
722 |
+
batch_size, -1, -1
|
723 |
+
) # stole cls_tokens impl from Phil Wang, thanks
|
724 |
+
x = torch.cat((cls_tokens, x), dim=1)
|
725 |
+
if self.pos_embed is not None:
|
726 |
+
x = x + self.pos_embed
|
727 |
+
x = self.pos_drop(x)
|
728 |
+
|
729 |
+
# a patch_dropout of 0. would mean it is disabled and this function would do
|
730 |
+
# nothing but return what was passed in
|
731 |
+
if self.rope is not None:
|
732 |
+
if self.training and not isinstance(self.patch_dropout, nn.Identity):
|
733 |
+
x, patch_indices_keep = self.patch_dropout(x)
|
734 |
+
self.rope.forward = partial(
|
735 |
+
self.rope.forward, patch_indices_keep=patch_indices_keep
|
736 |
+
)
|
737 |
+
else:
|
738 |
+
self.rope.forward = partial(self.rope.forward, patch_indices_keep=None)
|
739 |
+
x = self.patch_dropout(x)
|
740 |
+
else:
|
741 |
+
x = self.patch_dropout(x)
|
742 |
+
|
743 |
+
rel_pos_bias = self.rel_pos_bias() if self.rel_pos_bias is not None else None
|
744 |
+
for blk in self.blocks:
|
745 |
+
if self.grad_checkpointing:
|
746 |
+
x = checkpoint(blk, x, (rel_pos_bias,))
|
747 |
+
else:
|
748 |
+
x = blk(x, rel_pos_bias=rel_pos_bias)
|
749 |
+
|
750 |
+
if not return_all_features:
|
751 |
+
x = self.norm(x)
|
752 |
+
if self.fc_norm is not None:
|
753 |
+
return self.fc_norm(x.mean(1))
|
754 |
+
else:
|
755 |
+
return x[:, 0]
|
756 |
+
return x
|
757 |
+
|
758 |
+
def forward(self, x, return_all_features=False):
|
759 |
+
if return_all_features:
|
760 |
+
return self.forward_features(x, return_all_features)
|
761 |
+
x = self.forward_features(x)
|
762 |
+
x = self.head(x)
|
763 |
+
return x
|
hf_model.py
ADDED
@@ -0,0 +1,425 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import re
|
2 |
+
from typing import Dict, Optional, Tuple
|
3 |
+
|
4 |
+
import torch
|
5 |
+
import torch.nn as nn
|
6 |
+
from transformers import AutoConfig, AutoModel, PretrainedConfig
|
7 |
+
from transformers.modeling_outputs import (
|
8 |
+
BaseModelOutput,
|
9 |
+
BaseModelOutputWithPooling,
|
10 |
+
BaseModelOutputWithPoolingAndCrossAttentions,
|
11 |
+
)
|
12 |
+
|
13 |
+
"""
|
14 |
+
HF architecture mapping
|
15 |
+
"""
|
16 |
+
|
17 |
+
_HF_ARCH_DICT = {
|
18 |
+
# https://huggingface.co/docs/transformers/model_doc/roberta#roberta
|
19 |
+
'roberta': {
|
20 |
+
'config_names': {
|
21 |
+
'context_length': 'max_position_embeddings',
|
22 |
+
'vocab_size': 'vocab_size',
|
23 |
+
'width': 'hidden_size',
|
24 |
+
'heads': 'num_attention_heads',
|
25 |
+
'layers': 'num_hidden_layers',
|
26 |
+
'layer_attr': 'layer',
|
27 |
+
'token_embeddings_attr': 'embeddings',
|
28 |
+
},
|
29 |
+
'pooler': 'mean_pooler',
|
30 |
+
},
|
31 |
+
# https://huggingface.co/docs/transformers/model_doc/xlm-roberta#transformers.XLMRobertaConfig
|
32 |
+
'xlm-roberta': {
|
33 |
+
'config_names': {
|
34 |
+
'context_length': 'max_position_embeddings',
|
35 |
+
'vocab_size': 'vocab_size',
|
36 |
+
'width': 'hidden_size',
|
37 |
+
'heads': 'num_attention_heads',
|
38 |
+
'layers': 'num_hidden_layers',
|
39 |
+
'layer_attr': 'layer',
|
40 |
+
'token_embeddings_attr': 'embeddings',
|
41 |
+
},
|
42 |
+
'pooler': 'mean_pooler',
|
43 |
+
},
|
44 |
+
# https://huggingface.co/docs/transformers/model_doc/mt5#mt5
|
45 |
+
'mt5': {
|
46 |
+
'config_names': {
|
47 |
+
# unlimited seqlen
|
48 |
+
# https://github.com/google-research/text-to-text-transfer-transformer/issues/273
|
49 |
+
# https://github.com/huggingface/transformers/blob/v4.24.0/src/transformers/models/t5/modeling_t5.py#L374
|
50 |
+
'context_length': '',
|
51 |
+
'vocab_size': 'vocab_size',
|
52 |
+
'width': 'd_model',
|
53 |
+
'heads': 'num_heads',
|
54 |
+
'layers': 'num_layers',
|
55 |
+
'layer_attr': 'block',
|
56 |
+
'token_embeddings_attr': 'embed_tokens',
|
57 |
+
},
|
58 |
+
'pooler': 'mean_pooler',
|
59 |
+
},
|
60 |
+
# https://huggingface.co/docs/transformers/model_doc/bert
|
61 |
+
'bert': {
|
62 |
+
'config_names': {
|
63 |
+
'context_length': 'max_position_embeddings',
|
64 |
+
'vocab_size': 'vocab_size',
|
65 |
+
'width': 'hidden_size',
|
66 |
+
'heads': 'num_attention_heads',
|
67 |
+
'layers': 'num_hidden_layers',
|
68 |
+
},
|
69 |
+
'pooler': 'cls_pooler',
|
70 |
+
},
|
71 |
+
# https://huggingface.co/docs/transformers/model_doc/m2m_100
|
72 |
+
'm2m_100': {
|
73 |
+
'config_names': {
|
74 |
+
'context_length': 'max_position_embeddings',
|
75 |
+
'vocab_size': 'vocab_size',
|
76 |
+
'width': 'd_model',
|
77 |
+
'heads': 'encoder_attention_heads',
|
78 |
+
'layers': 'encoder_layers',
|
79 |
+
},
|
80 |
+
'pooler': 'cls_pooler',
|
81 |
+
},
|
82 |
+
}
|
83 |
+
|
84 |
+
|
85 |
+
"""
|
86 |
+
Pooling functions
|
87 |
+
"""
|
88 |
+
|
89 |
+
_POOLERS = {}
|
90 |
+
|
91 |
+
|
92 |
+
def _camel2snake(s):
|
93 |
+
return re.sub(r'(?<!^)(?=[A-Z])', '_', s).lower()
|
94 |
+
|
95 |
+
|
96 |
+
def register_pooler(cls):
|
97 |
+
"""Decorator registering pooler class"""
|
98 |
+
_POOLERS[_camel2snake(cls.__name__)] = cls
|
99 |
+
return cls
|
100 |
+
|
101 |
+
|
102 |
+
@register_pooler
|
103 |
+
class MeanPooler(nn.Module):
|
104 |
+
"""Mean pooling"""
|
105 |
+
|
106 |
+
@staticmethod
|
107 |
+
def forward(x: BaseModelOutput, attention_mask: torch.Tensor):
|
108 |
+
masked_output = x.last_hidden_state * attention_mask.unsqueeze(-1)
|
109 |
+
return masked_output.sum(dim=1) / attention_mask.sum(-1, keepdim=True)
|
110 |
+
|
111 |
+
|
112 |
+
@register_pooler
|
113 |
+
class MaxPooler(nn.Module):
|
114 |
+
"""
|
115 |
+
Max pooling
|
116 |
+
"""
|
117 |
+
|
118 |
+
@staticmethod
|
119 |
+
def forward(x: BaseModelOutput, attention_mask: torch.Tensor):
|
120 |
+
masked_output = x.last_hidden_state.masked_fill(
|
121 |
+
attention_mask.unsqueeze(-1), -torch.inf
|
122 |
+
)
|
123 |
+
return masked_output.max(1).values
|
124 |
+
|
125 |
+
|
126 |
+
@register_pooler
|
127 |
+
class ClsPooler(nn.Module):
|
128 |
+
"""
|
129 |
+
CLS token pooling
|
130 |
+
"""
|
131 |
+
|
132 |
+
def __init__(self, use_pooler_output=True):
|
133 |
+
super().__init__()
|
134 |
+
self.cls_token_position = 0
|
135 |
+
self.use_pooler_output = use_pooler_output
|
136 |
+
|
137 |
+
def forward(self, x: BaseModelOutput, _: torch.Tensor):
|
138 |
+
if (
|
139 |
+
self.use_pooler_output
|
140 |
+
and isinstance(
|
141 |
+
x,
|
142 |
+
(
|
143 |
+
BaseModelOutputWithPooling,
|
144 |
+
BaseModelOutputWithPoolingAndCrossAttentions,
|
145 |
+
),
|
146 |
+
)
|
147 |
+
and (x.pooler_output is not None)
|
148 |
+
):
|
149 |
+
return x.pooler_output
|
150 |
+
|
151 |
+
return x.last_hidden_state[:, self.cls_token_position, :]
|
152 |
+
|
153 |
+
|
154 |
+
"""
|
155 |
+
HF text model
|
156 |
+
"""
|
157 |
+
|
158 |
+
|
159 |
+
class HFTextEncoder(nn.Module):
|
160 |
+
output_tokens: torch.jit.Final[bool]
|
161 |
+
|
162 |
+
def __init__(
|
163 |
+
self,
|
164 |
+
model_name_or_path: str,
|
165 |
+
output_dim: int,
|
166 |
+
config: PretrainedConfig = None,
|
167 |
+
pooler_type: str = None,
|
168 |
+
proj_type: str = None,
|
169 |
+
proj_bias: bool = False,
|
170 |
+
pretrained: bool = True,
|
171 |
+
output_tokens: bool = False,
|
172 |
+
trust_remote_code: bool = False,
|
173 |
+
revision: Optional[str] = None,
|
174 |
+
model_config_kwargs: Optional[Dict] = None,
|
175 |
+
):
|
176 |
+
super().__init__()
|
177 |
+
self.output_tokens = output_tokens
|
178 |
+
self.output_dim = output_dim
|
179 |
+
|
180 |
+
# TODO: find better way to get this information
|
181 |
+
uses_transformer_pooler = pooler_type == 'cls_pooler'
|
182 |
+
model_config_kwargs = model_config_kwargs or {}
|
183 |
+
|
184 |
+
if config is None:
|
185 |
+
self.config = AutoConfig.from_pretrained(
|
186 |
+
model_name_or_path,
|
187 |
+
trust_remote_code=trust_remote_code,
|
188 |
+
code_revision=revision,
|
189 |
+
)
|
190 |
+
self.config.update(model_config_kwargs)
|
191 |
+
create_func, model_args = (
|
192 |
+
(AutoModel.from_pretrained, model_name_or_path)
|
193 |
+
if pretrained
|
194 |
+
else (AutoModel.from_config, self.config)
|
195 |
+
)
|
196 |
+
# TODO: do all model configs have this attribute?
|
197 |
+
# PretrainedConfig does so yes??
|
198 |
+
if (
|
199 |
+
hasattr(self.config, 'is_encoder_decoder')
|
200 |
+
and self.config.is_encoder_decoder
|
201 |
+
):
|
202 |
+
self.transformer = create_func(model_args)
|
203 |
+
self.transformer = self.transformer.encoder
|
204 |
+
else:
|
205 |
+
self.transformer = create_func(
|
206 |
+
model_args,
|
207 |
+
trust_remote_code=trust_remote_code,
|
208 |
+
add_pooling_layer=uses_transformer_pooler,
|
209 |
+
code_revision=revision,
|
210 |
+
)
|
211 |
+
else:
|
212 |
+
self.config = config
|
213 |
+
self.config.update(model_config_kwargs)
|
214 |
+
self.transformer = AutoModel.from_config(self.config)
|
215 |
+
|
216 |
+
if pooler_type is None: # get default arch pooler
|
217 |
+
pooler_type = _HF_ARCH_DICT[self.config.model_type]['pooler']
|
218 |
+
|
219 |
+
# FIXME downstream users of OpenCLIP models use these attr,
|
220 |
+
# need to verify valid across all models
|
221 |
+
self.vocab_size = getattr(self.config, 'vocab_size', 0)
|
222 |
+
self.context_length = getattr(self.config, 'max_position_embeddings', 0)
|
223 |
+
|
224 |
+
self.pooler = _POOLERS[pooler_type]()
|
225 |
+
|
226 |
+
d_model = getattr(
|
227 |
+
self.config, _HF_ARCH_DICT[self.config.model_type]['config_names']['width']
|
228 |
+
)
|
229 |
+
if (d_model == output_dim) and (proj_type is None): # do we always need a proj?
|
230 |
+
self.proj = nn.Identity()
|
231 |
+
elif proj_type == 'linear':
|
232 |
+
self.proj = nn.Linear(d_model, output_dim, bias=proj_bias)
|
233 |
+
elif proj_type == 'mlp':
|
234 |
+
hidden_size = (d_model + output_dim) // 2
|
235 |
+
self.proj = nn.Sequential(
|
236 |
+
nn.Linear(d_model, hidden_size, bias=proj_bias),
|
237 |
+
nn.GELU(),
|
238 |
+
nn.Linear(hidden_size, output_dim, bias=proj_bias),
|
239 |
+
)
|
240 |
+
|
241 |
+
def forward(self, x: torch.Tensor):
|
242 |
+
attn_mask = (x != self.config.pad_token_id).long()
|
243 |
+
out = self.transformer(input_ids=x, attention_mask=attn_mask)
|
244 |
+
pooled_out = self.pooler(out, attn_mask)
|
245 |
+
projected = self.proj(pooled_out)
|
246 |
+
|
247 |
+
seq_len = out.last_hidden_state.shape[1]
|
248 |
+
tokens = (
|
249 |
+
out.last_hidden_state[
|
250 |
+
:, torch.arange(seq_len) != self.pooler.cls_token_position, :
|
251 |
+
]
|
252 |
+
if isinstance(self.pooler, ClsPooler)
|
253 |
+
else out.last_hidden_state
|
254 |
+
)
|
255 |
+
|
256 |
+
if self.output_tokens:
|
257 |
+
return projected, tokens
|
258 |
+
return projected
|
259 |
+
|
260 |
+
def lock(self, unlocked_layers: int = 0, freeze_layer_norm: bool = True):
|
261 |
+
if not unlocked_layers: # full freezing
|
262 |
+
for n, p in self.transformer.named_parameters():
|
263 |
+
p.requires_grad = (
|
264 |
+
(not freeze_layer_norm) if 'LayerNorm' in n.split('.') else False
|
265 |
+
)
|
266 |
+
return
|
267 |
+
|
268 |
+
encoder = (
|
269 |
+
self.transformer.encoder
|
270 |
+
if hasattr(self.transformer, 'encoder')
|
271 |
+
else self.transformer
|
272 |
+
)
|
273 |
+
layer_list = getattr(
|
274 |
+
encoder, _HF_ARCH_DICT[self.config.model_type]['config_names']['layer_attr']
|
275 |
+
)
|
276 |
+
print(f'Unlocking {unlocked_layers}/{len(layer_list) + 1} layers of hf model')
|
277 |
+
embeddings = getattr(
|
278 |
+
self.transformer,
|
279 |
+
_HF_ARCH_DICT[self.config.model_type]['config_names'][
|
280 |
+
'token_embeddings_attr'
|
281 |
+
],
|
282 |
+
)
|
283 |
+
modules = [embeddings, *layer_list][:-unlocked_layers]
|
284 |
+
# freeze layers
|
285 |
+
for module in modules:
|
286 |
+
for n, p in module.named_parameters():
|
287 |
+
p.requires_grad = (
|
288 |
+
(not freeze_layer_norm) if 'LayerNorm' in n.split('.') else False
|
289 |
+
)
|
290 |
+
|
291 |
+
@torch.jit.ignore
|
292 |
+
def set_grad_checkpointing(self, _=True):
|
293 |
+
self.transformer.gradient_checkpointing_enable()
|
294 |
+
|
295 |
+
def init_parameters(self):
|
296 |
+
pass
|
297 |
+
|
298 |
+
|
299 |
+
"""
|
300 |
+
HF vision model
|
301 |
+
"""
|
302 |
+
|
303 |
+
|
304 |
+
class HFVisionEncoder(nn.Module):
|
305 |
+
output_tokens: torch.jit.Final[bool]
|
306 |
+
|
307 |
+
def __init__(
|
308 |
+
self,
|
309 |
+
model_name_or_path: str,
|
310 |
+
image_size: int,
|
311 |
+
output_dim: int,
|
312 |
+
config: PretrainedConfig = None,
|
313 |
+
pool_type: str = 'tok',
|
314 |
+
proj_type: Optional[str] = None,
|
315 |
+
proj_bias: bool = False,
|
316 |
+
attn_drop: float = 0.0,
|
317 |
+
hidden_drop: float = 0.0,
|
318 |
+
drop_path: Optional[float] = None,
|
319 |
+
pretrained: bool = True,
|
320 |
+
output_tokens: bool = False,
|
321 |
+
trust_remote_code: bool = False,
|
322 |
+
):
|
323 |
+
super().__init__()
|
324 |
+
self.output_tokens = output_tokens
|
325 |
+
self.output_dim = output_dim
|
326 |
+
self.image_size = (image_size, image_size)
|
327 |
+
|
328 |
+
if config is None:
|
329 |
+
self.config = AutoConfig.from_pretrained(
|
330 |
+
model_name_or_path,
|
331 |
+
trust_remote_code=trust_remote_code,
|
332 |
+
hidden_dropout_prob=hidden_drop,
|
333 |
+
attention_probs_dropout_prob=attn_drop,
|
334 |
+
drop_path_rate=drop_path,
|
335 |
+
)
|
336 |
+
create_func, model_args = (
|
337 |
+
(AutoModel.from_pretrained, model_name_or_path)
|
338 |
+
if pretrained
|
339 |
+
else (AutoModel.from_config, self.config)
|
340 |
+
)
|
341 |
+
self.transformer = create_func(
|
342 |
+
model_args,
|
343 |
+
trust_remote_code=trust_remote_code,
|
344 |
+
hidden_dropout_prob=hidden_drop,
|
345 |
+
attention_probs_dropout_prob=attn_drop,
|
346 |
+
)
|
347 |
+
else:
|
348 |
+
self.config = config
|
349 |
+
self.transformer = AutoModel.from_config(config)
|
350 |
+
|
351 |
+
if 'dinov2' in model_name_or_path:
|
352 |
+
self.transformer.embeddings.mask_token.requires_grad = False
|
353 |
+
|
354 |
+
assert pool_type in ('tok', 'avg', 'none')
|
355 |
+
self.pool_type = pool_type
|
356 |
+
|
357 |
+
d_model = self.config.hidden_size
|
358 |
+
if (d_model == output_dim) and (proj_type is None): # do we always need a proj?
|
359 |
+
self.proj = nn.Identity()
|
360 |
+
elif proj_type == 'linear':
|
361 |
+
self.proj = nn.Linear(d_model, output_dim, bias=proj_bias)
|
362 |
+
elif proj_type == 'mlp':
|
363 |
+
hidden_size = (d_model + output_dim) // 2
|
364 |
+
self.proj = nn.Sequential(
|
365 |
+
nn.Linear(d_model, hidden_size, bias=proj_bias),
|
366 |
+
nn.GELU(),
|
367 |
+
nn.Linear(hidden_size, output_dim, bias=proj_bias),
|
368 |
+
)
|
369 |
+
|
370 |
+
def _global_pool(self, x: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
|
371 |
+
if self.pool_type == 'avg':
|
372 |
+
pooled, tokens = x[:, 1:].mean(dim=1), x[:, 1:]
|
373 |
+
elif self.pool_type == 'tok':
|
374 |
+
pooled, tokens = x[:, 0], x[:, 1:]
|
375 |
+
else:
|
376 |
+
pooled = tokens = x
|
377 |
+
|
378 |
+
return pooled, tokens
|
379 |
+
|
380 |
+
def forward(self, x: torch.Tensor):
|
381 |
+
# returns a tuple of (final hidden states, token pooled outputs)
|
382 |
+
x = self.transformer(x)[0]
|
383 |
+
pooled, tokens = self._global_pool(x)
|
384 |
+
projected = self.proj(pooled)
|
385 |
+
|
386 |
+
return projected
|
387 |
+
|
388 |
+
def lock(self, unlocked_layers: int = 0, freeze_bn_stats: bool = True):
|
389 |
+
if not unlocked_layers: # full freezing
|
390 |
+
for n, p in self.transformer.named_parameters():
|
391 |
+
p.requires_grad = (
|
392 |
+
(not freeze_bn_stats) if 'LayerNorm' in n.split('.') else False
|
393 |
+
)
|
394 |
+
return
|
395 |
+
|
396 |
+
# TODO: make it work if unlocked_layers !=0
|
397 |
+
encoder = (
|
398 |
+
self.transformer.encoder
|
399 |
+
if hasattr(self.transformer, 'encoder')
|
400 |
+
else self.transformer
|
401 |
+
)
|
402 |
+
layer_list = getattr(
|
403 |
+
encoder, _HF_ARCH_DICT[self.config.model_type]['config_names']['layer_attr']
|
404 |
+
)
|
405 |
+
print(f'Unlocking {unlocked_layers}/{len(layer_list) + 1} layers of hf model')
|
406 |
+
embeddings = getattr(
|
407 |
+
self.transformer,
|
408 |
+
_HF_ARCH_DICT[self.config.model_type]['config_names'][
|
409 |
+
'token_embeddings_attr'
|
410 |
+
],
|
411 |
+
)
|
412 |
+
modules = [embeddings, *layer_list][:-unlocked_layers]
|
413 |
+
# freeze layers
|
414 |
+
for module in modules:
|
415 |
+
for n, p in module.named_parameters():
|
416 |
+
p.requires_grad = (
|
417 |
+
(not freeze_bn_stats) if 'LayerNorm' in n.split('.') else False
|
418 |
+
)
|
419 |
+
|
420 |
+
@torch.jit.ignore
|
421 |
+
def set_grad_checkpointing(self, *_, **__):
|
422 |
+
self.transformer.gradient_checkpointing_enable()
|
423 |
+
|
424 |
+
def init_parameters(self):
|
425 |
+
pass
|
modeling_clip.py
ADDED
@@ -0,0 +1,317 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
#
|
3 |
+
# Code mainly copied from:
|
4 |
+
# https://github.com/huggingface/transformers/blob/main/src/transformers/models/clip/modeling_clip.py
|
5 |
+
# and adjusted for Jina CLIP
|
6 |
+
|
7 |
+
from functools import partial
|
8 |
+
from typing import Optional, Tuple, Union
|
9 |
+
|
10 |
+
import torch
|
11 |
+
import torch.nn.functional as f
|
12 |
+
import torch.utils.checkpoint
|
13 |
+
from torch import nn
|
14 |
+
from transformers import BatchEncoding, BatchFeature, PreTrainedModel, logging
|
15 |
+
from transformers.models.clip.modeling_clip import (
|
16 |
+
CLIPOutput,
|
17 |
+
CLIPTextModelOutput,
|
18 |
+
CLIPVisionModelOutput,
|
19 |
+
clip_loss,
|
20 |
+
)
|
21 |
+
|
22 |
+
from .configuration_clip import JinaCLIPConfig, JinaCLIPTextConfig, JinaCLIPVisionConfig
|
23 |
+
from .eva_model import EVAVisionTransformer
|
24 |
+
from .hf_model import HFTextEncoder
|
25 |
+
|
26 |
+
logger = logging.get_logger(__name__)
|
27 |
+
|
28 |
+
|
29 |
+
""" Jina CLIP model implementation """
|
30 |
+
|
31 |
+
|
32 |
+
class LayerNorm(nn.LayerNorm):
|
33 |
+
"""Subclass torch's LayerNorm (with cast back to input dtype)."""
|
34 |
+
|
35 |
+
def forward(self, x: torch.Tensor):
|
36 |
+
origtype = x.dtype
|
37 |
+
x = f.layer_norm(x, self.normalized_shape, self.weight, self.bias, self.eps)
|
38 |
+
return x.to(origtype)
|
39 |
+
|
40 |
+
|
41 |
+
def _build_text_tower(config: JinaCLIPTextConfig) -> HFTextEncoder:
|
42 |
+
return HFTextEncoder(
|
43 |
+
model_name_or_path=config.hf_model_name_or_path,
|
44 |
+
output_dim=config.embed_dim,
|
45 |
+
pooler_type=config.pooler_type,
|
46 |
+
proj_type=config.proj_type,
|
47 |
+
proj_bias=config.proj_bias,
|
48 |
+
pretrained=False,
|
49 |
+
output_tokens=False,
|
50 |
+
trust_remote_code=True,
|
51 |
+
revision=None,
|
52 |
+
model_config_kwargs=config.hf_model_config_kwargs,
|
53 |
+
)
|
54 |
+
|
55 |
+
|
56 |
+
def _build_vision_tower(config: JinaCLIPVisionConfig) -> EVAVisionTransformer:
|
57 |
+
norm_layer = partial(LayerNorm, eps=1e-6)
|
58 |
+
|
59 |
+
if config.fused_layer_norm:
|
60 |
+
try:
|
61 |
+
from apex.normalization import FusedLayerNorm
|
62 |
+
|
63 |
+
norm_layer = partial(FusedLayerNorm, eps=1e-6)
|
64 |
+
except (ModuleNotFoundError, ImportError):
|
65 |
+
logger.warning('Please install apex to use fused layer norm, ignoring')
|
66 |
+
|
67 |
+
return EVAVisionTransformer(
|
68 |
+
img_size=config.image_size,
|
69 |
+
patch_size=config.patch_size,
|
70 |
+
num_classes=config.embed_dim,
|
71 |
+
use_mean_pooling=False,
|
72 |
+
init_values=config.ls_init_value,
|
73 |
+
patch_dropout=config.patch_dropout,
|
74 |
+
embed_dim=config.width,
|
75 |
+
depth=config.layers,
|
76 |
+
num_heads=config.width // config.head_width,
|
77 |
+
mlp_ratio=config.mlp_ratio,
|
78 |
+
qkv_bias=config.qkv_bias,
|
79 |
+
drop_path_rate=config.drop_path_rate,
|
80 |
+
norm_layer=norm_layer,
|
81 |
+
xattn=config.x_attention,
|
82 |
+
rope=config.rope_embeddings,
|
83 |
+
postnorm=config.post_norm,
|
84 |
+
pt_hw_seq_len=config.pt_hw_seq_len,
|
85 |
+
intp_freq=config.intp_freq,
|
86 |
+
naiveswiglu=config.naive_swiglu,
|
87 |
+
subln=config.subln,
|
88 |
+
proj_type=config.proj_type,
|
89 |
+
)
|
90 |
+
|
91 |
+
|
92 |
+
class JinaCLIPPreTrainedModel(PreTrainedModel):
|
93 |
+
"""
|
94 |
+
An abstract class to handle weights initialization and a simple interface for
|
95 |
+
downloading and loading pretrained models.
|
96 |
+
"""
|
97 |
+
|
98 |
+
config_class = JinaCLIPConfig
|
99 |
+
base_model_prefix = 'clip'
|
100 |
+
supports_gradient_checkpointing = True
|
101 |
+
|
102 |
+
def _init_weights(self, module):
|
103 |
+
"""Initialize the weights"""
|
104 |
+
if isinstance(module, JinaCLIPModel):
|
105 |
+
if isinstance(module.text_projection, nn.Linear):
|
106 |
+
nn.init.normal_(
|
107 |
+
module.text_projection.weight,
|
108 |
+
std=module.text_embed_dim**-0.5 * self.config.initializer_factor,
|
109 |
+
)
|
110 |
+
if isinstance(module.text_projection, nn.Linear):
|
111 |
+
nn.init.normal_(
|
112 |
+
module.visual_projection.weight,
|
113 |
+
std=module.vision_embed_dim**-0.5 * self.config.initializer_factor,
|
114 |
+
)
|
115 |
+
if isinstance(module, nn.LayerNorm):
|
116 |
+
module.bias.data.zero_()
|
117 |
+
module.weight.data.fill_(1.0)
|
118 |
+
if isinstance(module, nn.Linear) and module.bias is not None:
|
119 |
+
module.bias.data.zero_()
|
120 |
+
|
121 |
+
|
122 |
+
class JinaCLIPTextModel(JinaCLIPPreTrainedModel):
|
123 |
+
config_class = JinaCLIPTextConfig
|
124 |
+
|
125 |
+
def __init__(self, config: JinaCLIPTextConfig):
|
126 |
+
super().__init__(config)
|
127 |
+
self.text_model = _build_text_tower(config)
|
128 |
+
self.post_init()
|
129 |
+
|
130 |
+
def forward(
|
131 |
+
self,
|
132 |
+
input_ids: Union[None, torch.Tensor, BatchEncoding] = None,
|
133 |
+
return_dict: Optional[bool] = None,
|
134 |
+
*_,
|
135 |
+
**__,
|
136 |
+
) -> Union[Tuple[Optional[torch.FloatTensor], ...], CLIPTextModelOutput]:
|
137 |
+
return_dict = (
|
138 |
+
return_dict if return_dict is not None else self.config.use_return_dict
|
139 |
+
)
|
140 |
+
x = input_ids.input_ids if isinstance(input_ids, BatchEncoding) else input_ids
|
141 |
+
feats = self.text_model(x=x)
|
142 |
+
out = CLIPTextModelOutput(text_embeds=feats)
|
143 |
+
return out if return_dict else out.to_tuple()
|
144 |
+
|
145 |
+
|
146 |
+
class JinaCLIPVisionModel(JinaCLIPPreTrainedModel):
|
147 |
+
config_class = JinaCLIPVisionConfig
|
148 |
+
main_input_name = 'pixel_values'
|
149 |
+
|
150 |
+
def __init__(self, config: JinaCLIPVisionConfig):
|
151 |
+
super().__init__(config)
|
152 |
+
self.vision_model = _build_vision_tower(config)
|
153 |
+
self.post_init()
|
154 |
+
|
155 |
+
def forward(
|
156 |
+
self,
|
157 |
+
pixel_values: Union[None, torch.FloatTensor, BatchFeature] = None,
|
158 |
+
return_dict: Optional[bool] = None,
|
159 |
+
*_,
|
160 |
+
**__,
|
161 |
+
) -> Union[Tuple[Optional[torch.FloatTensor], ...], CLIPVisionModelOutput]:
|
162 |
+
return_dict = (
|
163 |
+
return_dict if return_dict is not None else self.config.use_return_dict
|
164 |
+
)
|
165 |
+
x = (
|
166 |
+
pixel_values.pixel_values
|
167 |
+
if isinstance(pixel_values, BatchFeature)
|
168 |
+
else pixel_values
|
169 |
+
)
|
170 |
+
feats = self.vision_model(x=x)
|
171 |
+
out = CLIPVisionModelOutput(image_embeds=feats)
|
172 |
+
return out if return_dict else out.to_tuple()
|
173 |
+
|
174 |
+
|
175 |
+
class JinaCLIPModel(JinaCLIPPreTrainedModel):
|
176 |
+
config_class = JinaCLIPConfig
|
177 |
+
|
178 |
+
def __init__(self, config: JinaCLIPConfig):
|
179 |
+
super().__init__(config)
|
180 |
+
|
181 |
+
if not isinstance(config.text_config, JinaCLIPTextConfig):
|
182 |
+
raise ValueError(
|
183 |
+
'Attribute config.text_config is expected to be of type '
|
184 |
+
f'JinaCLIPTextConfig but is of type {type(config.text_config)}.'
|
185 |
+
)
|
186 |
+
|
187 |
+
if not isinstance(config.vision_config, JinaCLIPVisionConfig):
|
188 |
+
raise ValueError(
|
189 |
+
'Attribute config.vision_config is expected to be of type '
|
190 |
+
f'JinaCLIPVisionConfig but is of type {type(config.vision_config)}.'
|
191 |
+
)
|
192 |
+
|
193 |
+
text_config = config.text_config
|
194 |
+
vision_config = config.vision_config
|
195 |
+
|
196 |
+
self.add_projections = config.add_projections
|
197 |
+
self.projection_dim = config.projection_dim
|
198 |
+
self.text_embed_dim = text_config.embed_dim
|
199 |
+
self.vision_embed_dim = vision_config.embed_dim
|
200 |
+
|
201 |
+
self.text_model = _build_text_tower(text_config)
|
202 |
+
self.vision_model = _build_vision_tower(vision_config)
|
203 |
+
self.logit_scale = nn.Parameter(
|
204 |
+
torch.tensor(self.config.logit_scale_init_value)
|
205 |
+
)
|
206 |
+
|
207 |
+
if self.add_projections:
|
208 |
+
self.visual_projection = nn.Linear(
|
209 |
+
self.vision_embed_dim, self.projection_dim, bias=False
|
210 |
+
)
|
211 |
+
self.text_projection = nn.Linear(
|
212 |
+
self.text_embed_dim, self.projection_dim, bias=False
|
213 |
+
)
|
214 |
+
else:
|
215 |
+
self.visual_projection = nn.Identity()
|
216 |
+
self.text_projection = nn.Identity()
|
217 |
+
|
218 |
+
self.post_init()
|
219 |
+
|
220 |
+
def get_text_features(
|
221 |
+
self,
|
222 |
+
input_ids: Union[None, torch.Tensor, BatchEncoding] = None,
|
223 |
+
*_,
|
224 |
+
**__,
|
225 |
+
) -> torch.FloatTensor:
|
226 |
+
x = input_ids.input_ids if isinstance(input_ids, BatchEncoding) else input_ids
|
227 |
+
return self.text_projection(self.text_model(x=x))
|
228 |
+
|
229 |
+
def get_image_features(
|
230 |
+
self,
|
231 |
+
pixel_values: Union[None, torch.FloatTensor, BatchFeature] = None,
|
232 |
+
*_,
|
233 |
+
**__,
|
234 |
+
) -> torch.FloatTensor:
|
235 |
+
x = (
|
236 |
+
pixel_values.pixel_values
|
237 |
+
if isinstance(pixel_values, BatchFeature)
|
238 |
+
else pixel_values
|
239 |
+
)
|
240 |
+
return self.visual_projection(self.vision_model(x=x))
|
241 |
+
|
242 |
+
def encode_text(
|
243 |
+
self,
|
244 |
+
input_ids: Union[None, torch.Tensor, BatchEncoding] = None,
|
245 |
+
return_dict: Optional[bool] = None,
|
246 |
+
*_,
|
247 |
+
**__,
|
248 |
+
) -> Union[Tuple[Optional[torch.FloatTensor], ...], CLIPTextModelOutput]:
|
249 |
+
return_dict = (
|
250 |
+
return_dict if return_dict is not None else self.config.use_return_dict
|
251 |
+
)
|
252 |
+
feats = self.get_text_features(input_ids=input_ids)
|
253 |
+
out = CLIPTextModelOutput(text_embeds=feats)
|
254 |
+
return out if return_dict else out.to_tuple()
|
255 |
+
|
256 |
+
def encode_image(
|
257 |
+
self,
|
258 |
+
pixel_values: Union[None, torch.FloatTensor, BatchFeature] = None,
|
259 |
+
return_dict: Optional[bool] = None,
|
260 |
+
*_,
|
261 |
+
**__,
|
262 |
+
) -> Union[Tuple[Optional[torch.FloatTensor], ...], CLIPVisionModelOutput]:
|
263 |
+
return_dict = (
|
264 |
+
return_dict if return_dict is not None else self.config.use_return_dict
|
265 |
+
)
|
266 |
+
feats = self.get_image_features(pixel_values=pixel_values)
|
267 |
+
out = CLIPVisionModelOutput(image_embeds=feats)
|
268 |
+
return out if return_dict else out.to_tuple()
|
269 |
+
|
270 |
+
def forward(
|
271 |
+
self,
|
272 |
+
input_ids: Union[None, torch.Tensor, BatchEncoding] = None,
|
273 |
+
pixel_values: Union[None, torch.FloatTensor, BatchFeature] = None,
|
274 |
+
return_dict: Optional[bool] = None,
|
275 |
+
return_loss: Optional[bool] = None,
|
276 |
+
*_,
|
277 |
+
**__,
|
278 |
+
) -> Union[Tuple[Optional[torch.FloatTensor], ...], CLIPOutput]:
|
279 |
+
return_dict = (
|
280 |
+
return_dict if return_dict is not None else self.config.use_return_dict
|
281 |
+
)
|
282 |
+
image_embeds = self.get_image_features(pixel_values=pixel_values)
|
283 |
+
text_embeds = self.get_text_features(input_ids=input_ids)
|
284 |
+
|
285 |
+
# normalized features
|
286 |
+
image_embeds = image_embeds / image_embeds.norm(p=2, dim=-1, keepdim=True)
|
287 |
+
text_embeds = text_embeds / text_embeds.norm(p=2, dim=-1, keepdim=True)
|
288 |
+
|
289 |
+
# cosine similarity as logits
|
290 |
+
logit_scale = self.logit_scale.exp()
|
291 |
+
logits_per_text = torch.matmul(text_embeds, image_embeds.t()) * logit_scale
|
292 |
+
logits_per_image = logits_per_text.t()
|
293 |
+
|
294 |
+
loss = None
|
295 |
+
if return_loss:
|
296 |
+
loss = clip_loss(logits_per_text)
|
297 |
+
|
298 |
+
if not return_dict:
|
299 |
+
output = (
|
300 |
+
logits_per_image,
|
301 |
+
logits_per_text,
|
302 |
+
text_embeds,
|
303 |
+
image_embeds,
|
304 |
+
None,
|
305 |
+
None,
|
306 |
+
)
|
307 |
+
return ((loss,) + output) if loss is not None else output
|
308 |
+
|
309 |
+
return CLIPOutput(
|
310 |
+
loss=loss,
|
311 |
+
logits_per_image=logits_per_image,
|
312 |
+
logits_per_text=logits_per_text,
|
313 |
+
text_embeds=text_embeds,
|
314 |
+
image_embeds=image_embeds,
|
315 |
+
text_model_output=None,
|
316 |
+
vision_model_output=None,
|
317 |
+
)
|
processing_clip.py
ADDED
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
#
|
3 |
+
# Code mainly copied from:
|
4 |
+
# https://github.com/huggingface/transformers/blob/main/src/transformers/models/clip/image_processing_clip.py
|
5 |
+
# and adjusted for Jina CLIP
|
6 |
+
|
7 |
+
from typing import Tuple, Union
|
8 |
+
|
9 |
+
import torch
|
10 |
+
from transformers.image_processing_utils import BaseImageProcessor, BatchFeature
|
11 |
+
from transformers.image_utils import ImageInput, make_list_of_images
|
12 |
+
from transformers.models.clip import CLIPProcessor
|
13 |
+
|
14 |
+
from .transform import OPENAI_DATASET_MEAN, OPENAI_DATASET_STD, image_transform
|
15 |
+
|
16 |
+
""" Jina CLIP processor implementation """
|
17 |
+
|
18 |
+
|
19 |
+
class JinaCLIPProcessor(CLIPProcessor):
|
20 |
+
image_processor_class = 'JinaCLIPImageProcessor'
|
21 |
+
tokenizer_class = 'CLIPTokenizer'
|
22 |
+
|
23 |
+
|
24 |
+
""" Jina CLIP image processor implementation """
|
25 |
+
|
26 |
+
|
27 |
+
class JinaCLIPImageProcessor(BaseImageProcessor):
|
28 |
+
model_input_names = ['pixel_values']
|
29 |
+
|
30 |
+
def __init__(
|
31 |
+
self,
|
32 |
+
size: Union[int, Tuple[int, int]] = 224,
|
33 |
+
mean: Union[float, Tuple[float]] = OPENAI_DATASET_MEAN,
|
34 |
+
std: Union[float, Tuple[float]] = OPENAI_DATASET_STD,
|
35 |
+
resize_mode: str = 'shortest',
|
36 |
+
interpolation: str = 'bicubic',
|
37 |
+
fill_color: int = 0,
|
38 |
+
**kwargs,
|
39 |
+
) -> None:
|
40 |
+
super().__init__(**kwargs)
|
41 |
+
self.size = size
|
42 |
+
self.mean = mean
|
43 |
+
self.std = std
|
44 |
+
self.resize_mode = resize_mode
|
45 |
+
self.interpolation = interpolation
|
46 |
+
self.fill_color = fill_color
|
47 |
+
self.transform = image_transform(
|
48 |
+
image_size=size,
|
49 |
+
is_train=False,
|
50 |
+
mean=mean,
|
51 |
+
std=std,
|
52 |
+
resize_mode=resize_mode,
|
53 |
+
interpolation=interpolation,
|
54 |
+
fill_color=fill_color,
|
55 |
+
aug_cfg=None,
|
56 |
+
)
|
57 |
+
|
58 |
+
def to_dict(self):
|
59 |
+
output = super().to_dict()
|
60 |
+
output.pop('transform')
|
61 |
+
return output
|
62 |
+
|
63 |
+
def preprocess(self, images: ImageInput, **kwargs) -> BatchFeature:
|
64 |
+
images = make_list_of_images(images)
|
65 |
+
out = torch.stack([self.transform(image) for image in images], dim=0)
|
66 |
+
return BatchFeature(data={'pixel_values': out})
|
rope_embeddings.py
ADDED
@@ -0,0 +1,165 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# --------------------------------------------------------
|
2 |
+
# Adapted from EVA CLIP
|
3 |
+
# https://github.com/baaivision/EVA/tree/master/EVA-CLIP/rei/eva_clip
|
4 |
+
# --------------------------------------------------------
|
5 |
+
|
6 |
+
import logging
|
7 |
+
from math import pi
|
8 |
+
|
9 |
+
import torch
|
10 |
+
from einops import rearrange, repeat
|
11 |
+
from torch import nn
|
12 |
+
|
13 |
+
|
14 |
+
def broadcast(tensors, dim=-1):
|
15 |
+
num_tensors = len(tensors)
|
16 |
+
shape_lens = set(list(map(lambda t: len(t.shape), tensors)))
|
17 |
+
assert len(shape_lens) == 1, 'tensors must all have the same number of dimensions'
|
18 |
+
shape_len = list(shape_lens)[0]
|
19 |
+
dim = (dim + shape_len) if dim < 0 else dim
|
20 |
+
dims = list(zip(*map(lambda t: list(t.shape), tensors)))
|
21 |
+
expandable_dims = [(i, val) for i, val in enumerate(dims) if i != dim]
|
22 |
+
assert all(
|
23 |
+
[*map(lambda t: len(set(t[1])) <= 2, expandable_dims)]
|
24 |
+
), 'invalid dimensions for broadcastable concatentation'
|
25 |
+
max_dims = list(map(lambda t: (t[0], max(t[1])), expandable_dims))
|
26 |
+
expanded_dims = list(map(lambda t: (t[0], (t[1],) * num_tensors), max_dims))
|
27 |
+
expanded_dims.insert(dim, (dim, dims[dim]))
|
28 |
+
expandable_shapes = list(zip(*map(lambda t: t[1], expanded_dims)))
|
29 |
+
tensors = list(map(lambda t: t[0].expand(*t[1]), zip(tensors, expandable_shapes)))
|
30 |
+
return torch.cat(tensors, dim=dim)
|
31 |
+
|
32 |
+
|
33 |
+
def rotate_half(x):
|
34 |
+
x = rearrange(x, '... (d r) -> ... d r', r=2)
|
35 |
+
x1, x2 = x.unbind(dim=-1)
|
36 |
+
x = torch.stack((-x2, x1), dim=-1)
|
37 |
+
return rearrange(x, '... d r -> ... (d r)')
|
38 |
+
|
39 |
+
|
40 |
+
class VisionRotaryEmbedding(nn.Module):
|
41 |
+
def __init__(
|
42 |
+
self,
|
43 |
+
dim,
|
44 |
+
pt_seq_len,
|
45 |
+
ft_seq_len=None,
|
46 |
+
custom_freqs=None,
|
47 |
+
freqs_for='lang',
|
48 |
+
theta=10000,
|
49 |
+
max_freq=10,
|
50 |
+
num_freqs=1,
|
51 |
+
):
|
52 |
+
super().__init__()
|
53 |
+
if custom_freqs:
|
54 |
+
freqs = custom_freqs
|
55 |
+
elif freqs_for == 'lang':
|
56 |
+
freqs = 1.0 / (
|
57 |
+
theta ** (torch.arange(0, dim, 2)[: (dim // 2)].float() / dim)
|
58 |
+
)
|
59 |
+
elif freqs_for == 'pixel':
|
60 |
+
freqs = torch.linspace(1.0, max_freq / 2, dim // 2) * pi
|
61 |
+
elif freqs_for == 'constant':
|
62 |
+
freqs = torch.ones(num_freqs).float()
|
63 |
+
else:
|
64 |
+
raise ValueError(f'unknown modality {freqs_for}')
|
65 |
+
|
66 |
+
if ft_seq_len is None:
|
67 |
+
ft_seq_len = pt_seq_len
|
68 |
+
t = torch.arange(ft_seq_len) / ft_seq_len * pt_seq_len
|
69 |
+
|
70 |
+
freqs_h = torch.einsum('..., f -> ... f', t, freqs)
|
71 |
+
freqs_h = repeat(freqs_h, '... n -> ... (n r)', r=2)
|
72 |
+
|
73 |
+
freqs_w = torch.einsum('..., f -> ... f', t, freqs)
|
74 |
+
freqs_w = repeat(freqs_w, '... n -> ... (n r)', r=2)
|
75 |
+
|
76 |
+
freqs = broadcast((freqs_h[:, None, :], freqs_w[None, :, :]), dim=-1)
|
77 |
+
|
78 |
+
self.register_buffer('freqs_cos', freqs.cos())
|
79 |
+
self.register_buffer('freqs_sin', freqs.sin())
|
80 |
+
|
81 |
+
logging.info(f'Shape of rope freq: {self.freqs_cos.shape}')
|
82 |
+
|
83 |
+
def forward(self, t, start_index=0):
|
84 |
+
rot_dim = self.freqs_cos.shape[-1]
|
85 |
+
end_index = start_index + rot_dim
|
86 |
+
assert rot_dim <= t.shape[-1], (
|
87 |
+
f'feature dimension {t.shape[-1]} is not of sufficient size to rotate in '
|
88 |
+
f'all the positions {rot_dim}'
|
89 |
+
)
|
90 |
+
t_left, t, t_right = (
|
91 |
+
t[..., :start_index],
|
92 |
+
t[..., start_index:end_index],
|
93 |
+
t[..., end_index:],
|
94 |
+
)
|
95 |
+
t = (t * self.freqs_cos) + (rotate_half(t) * self.freqs_sin)
|
96 |
+
|
97 |
+
return torch.cat((t_left, t, t_right), dim=-1)
|
98 |
+
|
99 |
+
|
100 |
+
class VisionRotaryEmbeddingFast(nn.Module):
|
101 |
+
def __init__(
|
102 |
+
self,
|
103 |
+
dim,
|
104 |
+
pt_seq_len,
|
105 |
+
ft_seq_len=None,
|
106 |
+
custom_freqs=None,
|
107 |
+
freqs_for='lang',
|
108 |
+
theta=10000,
|
109 |
+
max_freq=10,
|
110 |
+
num_freqs=1,
|
111 |
+
patch_dropout=0.0,
|
112 |
+
):
|
113 |
+
super().__init__()
|
114 |
+
if custom_freqs:
|
115 |
+
freqs = custom_freqs
|
116 |
+
elif freqs_for == 'lang':
|
117 |
+
freqs = 1.0 / (
|
118 |
+
theta ** (torch.arange(0, dim, 2)[: (dim // 2)].float() / dim)
|
119 |
+
)
|
120 |
+
elif freqs_for == 'pixel':
|
121 |
+
freqs = torch.linspace(1.0, max_freq / 2, dim // 2) * pi
|
122 |
+
elif freqs_for == 'constant':
|
123 |
+
freqs = torch.ones(num_freqs).float()
|
124 |
+
else:
|
125 |
+
raise ValueError(f'unknown modality {freqs_for}')
|
126 |
+
|
127 |
+
if ft_seq_len is None:
|
128 |
+
ft_seq_len = pt_seq_len
|
129 |
+
t = torch.arange(ft_seq_len) / ft_seq_len * pt_seq_len
|
130 |
+
|
131 |
+
freqs = torch.einsum('..., f -> ... f', t, freqs)
|
132 |
+
freqs = repeat(freqs, '... n -> ... (n r)', r=2)
|
133 |
+
freqs = broadcast((freqs[:, None, :], freqs[None, :, :]), dim=-1)
|
134 |
+
|
135 |
+
freqs_cos = freqs.cos().view(-1, freqs.shape[-1])
|
136 |
+
freqs_sin = freqs.sin().view(-1, freqs.shape[-1])
|
137 |
+
|
138 |
+
self.patch_dropout = patch_dropout
|
139 |
+
|
140 |
+
self.register_buffer('freqs_cos', freqs_cos)
|
141 |
+
self.register_buffer('freqs_sin', freqs_sin)
|
142 |
+
|
143 |
+
logging.info(f'Shape of rope freq: {self.freqs_cos.shape}')
|
144 |
+
|
145 |
+
def forward(self, t, patch_indices_keep=None):
|
146 |
+
if patch_indices_keep is not None:
|
147 |
+
batch = t.size()[0]
|
148 |
+
batch_indices = torch.arange(batch)
|
149 |
+
batch_indices = batch_indices[..., None]
|
150 |
+
|
151 |
+
freqs_cos = repeat(
|
152 |
+
self.freqs_cos, 'i j -> n i m j', n=t.shape[0], m=t.shape[1]
|
153 |
+
)
|
154 |
+
freqs_sin = repeat(
|
155 |
+
self.freqs_sin, 'i j -> n i m j', n=t.shape[0], m=t.shape[1]
|
156 |
+
)
|
157 |
+
|
158 |
+
freqs_cos = freqs_cos[batch_indices, patch_indices_keep]
|
159 |
+
freqs_cos = rearrange(freqs_cos, 'n i m j -> n m i j')
|
160 |
+
freqs_sin = freqs_sin[batch_indices, patch_indices_keep]
|
161 |
+
freqs_sin = rearrange(freqs_sin, 'n i m j -> n m i j')
|
162 |
+
|
163 |
+
return t * freqs_cos + rotate_half(t) * freqs_sin
|
164 |
+
|
165 |
+
return t * self.freqs_cos + rotate_half(t) * self.freqs_sin
|
transform.py
ADDED
@@ -0,0 +1,458 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import numbers
|
2 |
+
import random
|
3 |
+
import warnings
|
4 |
+
from dataclasses import asdict, dataclass
|
5 |
+
from typing import Any, Dict, List, Optional, Sequence, Tuple, Union
|
6 |
+
|
7 |
+
import torch
|
8 |
+
import torchvision.transforms.functional as F
|
9 |
+
from torchvision.transforms import (
|
10 |
+
CenterCrop,
|
11 |
+
ColorJitter,
|
12 |
+
Compose,
|
13 |
+
Grayscale,
|
14 |
+
InterpolationMode,
|
15 |
+
Normalize,
|
16 |
+
RandomResizedCrop,
|
17 |
+
Resize,
|
18 |
+
ToTensor,
|
19 |
+
)
|
20 |
+
from transformers.image_utils import OPENAI_CLIP_MEAN, OPENAI_CLIP_STD
|
21 |
+
|
22 |
+
OPENAI_DATASET_MEAN = tuple(OPENAI_CLIP_MEAN)
|
23 |
+
OPENAI_DATASET_STD = tuple(OPENAI_CLIP_STD)
|
24 |
+
|
25 |
+
|
26 |
+
@dataclass
|
27 |
+
class PreprocessCfg:
|
28 |
+
size: Union[int, Tuple[int, int]] = 224
|
29 |
+
mode: str = 'RGB'
|
30 |
+
mean: Tuple[float, ...] = OPENAI_DATASET_MEAN
|
31 |
+
std: Tuple[float, ...] = OPENAI_DATASET_STD
|
32 |
+
interpolation: str = 'bicubic'
|
33 |
+
resize_mode: str = 'shortest'
|
34 |
+
fill_color: int = 0
|
35 |
+
|
36 |
+
def __post_init__(self):
|
37 |
+
assert self.mode in ('RGB',)
|
38 |
+
|
39 |
+
@property
|
40 |
+
def num_channels(self):
|
41 |
+
return 3
|
42 |
+
|
43 |
+
@property
|
44 |
+
def input_size(self):
|
45 |
+
return (self.num_channels,) + (self.size, self.size)
|
46 |
+
|
47 |
+
|
48 |
+
_PREPROCESS_KEYS = set(asdict(PreprocessCfg()).keys())
|
49 |
+
|
50 |
+
|
51 |
+
def merge_preprocess_dict(
|
52 |
+
base: Union[PreprocessCfg, Dict],
|
53 |
+
overlay: Dict,
|
54 |
+
):
|
55 |
+
"""Merge overlay key-value pairs on top of base preprocess cfg or dict.
|
56 |
+
Input dicts are filtered based on PreprocessCfg fields.
|
57 |
+
"""
|
58 |
+
if isinstance(base, PreprocessCfg):
|
59 |
+
base_clean = asdict(base)
|
60 |
+
else:
|
61 |
+
base_clean = {k: v for k, v in base.items() if k in _PREPROCESS_KEYS}
|
62 |
+
if overlay:
|
63 |
+
overlay_clean = {
|
64 |
+
k: v for k, v in overlay.items() if k in _PREPROCESS_KEYS and v is not None
|
65 |
+
}
|
66 |
+
base_clean.update(overlay_clean)
|
67 |
+
return base_clean
|
68 |
+
|
69 |
+
|
70 |
+
def merge_preprocess_kwargs(base: Union[PreprocessCfg, Dict], **kwargs):
|
71 |
+
return merge_preprocess_dict(base, kwargs)
|
72 |
+
|
73 |
+
|
74 |
+
@dataclass
|
75 |
+
class AugmentationCfg:
|
76 |
+
scale: Tuple[float, float] = (0.9, 1.0)
|
77 |
+
ratio: Optional[Tuple[float, float]] = None
|
78 |
+
color_jitter: Optional[
|
79 |
+
Union[float, Tuple[float, float, float], Tuple[float, float, float, float]]
|
80 |
+
] = None
|
81 |
+
re_prob: Optional[float] = None
|
82 |
+
re_count: Optional[int] = None
|
83 |
+
use_timm: bool = False
|
84 |
+
|
85 |
+
# params for simclr_jitter_gray
|
86 |
+
color_jitter_prob: float = None
|
87 |
+
gray_scale_prob: float = None
|
88 |
+
|
89 |
+
|
90 |
+
def _setup_size(size, error_msg):
|
91 |
+
if isinstance(size, numbers.Number):
|
92 |
+
return int(size), int(size)
|
93 |
+
|
94 |
+
if isinstance(size, Sequence) and len(size) == 1:
|
95 |
+
return size[0], size[0]
|
96 |
+
|
97 |
+
if len(size) != 2:
|
98 |
+
raise ValueError(error_msg)
|
99 |
+
|
100 |
+
return size
|
101 |
+
|
102 |
+
|
103 |
+
class ResizeKeepRatio:
|
104 |
+
"""Resize and Keep Ratio
|
105 |
+
|
106 |
+
Copy & paste from `timm`
|
107 |
+
"""
|
108 |
+
|
109 |
+
def __init__(
|
110 |
+
self,
|
111 |
+
size,
|
112 |
+
longest=0.0,
|
113 |
+
interpolation=InterpolationMode.BICUBIC,
|
114 |
+
random_scale_prob=0.0,
|
115 |
+
random_scale_range=(0.85, 1.05),
|
116 |
+
random_aspect_prob=0.0,
|
117 |
+
random_aspect_range=(0.9, 1.11),
|
118 |
+
):
|
119 |
+
if isinstance(size, (list, tuple)):
|
120 |
+
self.size = tuple(size)
|
121 |
+
else:
|
122 |
+
self.size = (size, size)
|
123 |
+
self.interpolation = interpolation
|
124 |
+
self.longest = float(longest) # [0, 1] where 0 == shortest edge, 1 == longest
|
125 |
+
self.random_scale_prob = random_scale_prob
|
126 |
+
self.random_scale_range = random_scale_range
|
127 |
+
self.random_aspect_prob = random_aspect_prob
|
128 |
+
self.random_aspect_range = random_aspect_range
|
129 |
+
|
130 |
+
@staticmethod
|
131 |
+
def get_params(
|
132 |
+
img,
|
133 |
+
target_size,
|
134 |
+
longest,
|
135 |
+
random_scale_prob=0.0,
|
136 |
+
random_scale_range=(0.85, 1.05),
|
137 |
+
random_aspect_prob=0.0,
|
138 |
+
random_aspect_range=(0.9, 1.11),
|
139 |
+
):
|
140 |
+
"""Get parameters"""
|
141 |
+
source_size = img.size[::-1] # h, w
|
142 |
+
h, w = source_size
|
143 |
+
target_h, target_w = target_size
|
144 |
+
ratio_h = h / target_h
|
145 |
+
ratio_w = w / target_w
|
146 |
+
ratio = max(ratio_h, ratio_w) * longest + min(ratio_h, ratio_w) * (
|
147 |
+
1.0 - longest
|
148 |
+
)
|
149 |
+
if random_scale_prob > 0 and random.random() < random_scale_prob:
|
150 |
+
ratio_factor = random.uniform(random_scale_range[0], random_scale_range[1])
|
151 |
+
ratio_factor = (ratio_factor, ratio_factor)
|
152 |
+
else:
|
153 |
+
ratio_factor = (1.0, 1.0)
|
154 |
+
if random_aspect_prob > 0 and random.random() < random_aspect_prob:
|
155 |
+
aspect_factor = random.uniform(
|
156 |
+
random_aspect_range[0], random_aspect_range[1]
|
157 |
+
)
|
158 |
+
ratio_factor = (
|
159 |
+
ratio_factor[0] / aspect_factor,
|
160 |
+
ratio_factor[1] * aspect_factor,
|
161 |
+
)
|
162 |
+
size = [round(x * f / ratio) for x, f in zip(source_size, ratio_factor)]
|
163 |
+
return size
|
164 |
+
|
165 |
+
def __call__(self, img):
|
166 |
+
"""
|
167 |
+
Args:
|
168 |
+
img (PIL Image): Image to be cropped and resized.
|
169 |
+
|
170 |
+
Returns:
|
171 |
+
PIL Image: Resized, padded to at least target size, possibly
|
172 |
+
cropped to exactly target size
|
173 |
+
"""
|
174 |
+
size = self.get_params(
|
175 |
+
img,
|
176 |
+
self.size,
|
177 |
+
self.longest,
|
178 |
+
self.random_scale_prob,
|
179 |
+
self.random_scale_range,
|
180 |
+
self.random_aspect_prob,
|
181 |
+
self.random_aspect_range,
|
182 |
+
)
|
183 |
+
img = F.resize(img, size, self.interpolation)
|
184 |
+
return img
|
185 |
+
|
186 |
+
def __repr__(self):
|
187 |
+
format_string = self.__class__.__name__ + '(size={0}'.format(self.size)
|
188 |
+
format_string += f', interpolation={self.interpolation})'
|
189 |
+
format_string += f', longest={self.longest:.3f})'
|
190 |
+
return format_string
|
191 |
+
|
192 |
+
|
193 |
+
def center_crop_or_pad(
|
194 |
+
img: torch.Tensor, output_size: List[int], fill=0
|
195 |
+
) -> torch.Tensor:
|
196 |
+
"""Center crops and/or pads the given image.
|
197 |
+
If the image is torch Tensor, it is expected
|
198 |
+
to have [..., H, W] shape, where ... means an arbitrary number of leading
|
199 |
+
dimensions. If image size is smaller than output size along any edge, image is
|
200 |
+
padded with 0 and then center cropped.
|
201 |
+
|
202 |
+
Args:
|
203 |
+
img (PIL Image or Tensor): Image to be cropped.
|
204 |
+
output_size (sequence or int): (height, width) of the crop box. If int or
|
205 |
+
sequence with single int, it is used for both directions.
|
206 |
+
fill (int, Tuple[int]): Padding color
|
207 |
+
|
208 |
+
Returns:
|
209 |
+
PIL Image or Tensor: Cropped image.
|
210 |
+
"""
|
211 |
+
if isinstance(output_size, numbers.Number):
|
212 |
+
output_size = (int(output_size), int(output_size))
|
213 |
+
elif isinstance(output_size, (tuple, list)) and len(output_size) == 1:
|
214 |
+
output_size = (output_size[0], output_size[0])
|
215 |
+
|
216 |
+
_, image_height, image_width = F.get_dimensions(img)
|
217 |
+
crop_height, crop_width = output_size
|
218 |
+
|
219 |
+
if crop_width > image_width or crop_height > image_height:
|
220 |
+
padding_ltrb = [
|
221 |
+
(crop_width - image_width) // 2 if crop_width > image_width else 0,
|
222 |
+
(crop_height - image_height) // 2 if crop_height > image_height else 0,
|
223 |
+
(crop_width - image_width + 1) // 2 if crop_width > image_width else 0,
|
224 |
+
(crop_height - image_height + 1) // 2 if crop_height > image_height else 0,
|
225 |
+
]
|
226 |
+
img = F.pad(img, padding_ltrb, fill=fill)
|
227 |
+
_, image_height, image_width = F.get_dimensions(img)
|
228 |
+
if crop_width == image_width and crop_height == image_height:
|
229 |
+
return img
|
230 |
+
|
231 |
+
crop_top = int(round((image_height - crop_height) / 2.0))
|
232 |
+
crop_left = int(round((image_width - crop_width) / 2.0))
|
233 |
+
return F.crop(img, crop_top, crop_left, crop_height, crop_width)
|
234 |
+
|
235 |
+
|
236 |
+
class CenterCropOrPad(torch.nn.Module):
|
237 |
+
"""Crops the given image at the center.
|
238 |
+
If the image is torch Tensor, it is expected
|
239 |
+
to have [..., H, W] shape, where ... means an arbitrary number of leading
|
240 |
+
dimensions. If image size is smaller than output size along any edge, image is
|
241 |
+
padded with 0 and then center cropped.
|
242 |
+
|
243 |
+
Args:
|
244 |
+
size (sequence or int): Desired output size of the crop. If size is an
|
245 |
+
int instead of sequence like (h, w), a square crop (size, size) is
|
246 |
+
made. If provided a sequence of length 1, it will be interpreted as
|
247 |
+
(size[0], size[0]).
|
248 |
+
"""
|
249 |
+
|
250 |
+
def __init__(self, size, fill=0):
|
251 |
+
super().__init__()
|
252 |
+
self.size = _setup_size(
|
253 |
+
size, error_msg='Please provide only two dimensions (h, w) for size.'
|
254 |
+
)
|
255 |
+
self.fill = fill
|
256 |
+
|
257 |
+
def forward(self, img):
|
258 |
+
"""
|
259 |
+
Args:
|
260 |
+
img (PIL Image or Tensor): Image to be cropped.
|
261 |
+
|
262 |
+
Returns:
|
263 |
+
PIL Image or Tensor: Cropped image.
|
264 |
+
"""
|
265 |
+
return center_crop_or_pad(img, self.size, fill=self.fill)
|
266 |
+
|
267 |
+
def __repr__(self) -> str:
|
268 |
+
return f'{self.__class__.__name__}(size={self.size})'
|
269 |
+
|
270 |
+
|
271 |
+
def _convert_to_rgb(image):
|
272 |
+
return image.convert('RGB')
|
273 |
+
|
274 |
+
|
275 |
+
class _ColorJitter(object):
|
276 |
+
"""
|
277 |
+
Apply Color Jitter to the PIL image with a specified probability.
|
278 |
+
"""
|
279 |
+
|
280 |
+
def __init__(self, brightness=0.0, contrast=0.0, saturation=0.0, hue=0.0, p=0.8):
|
281 |
+
assert 0.0 <= p <= 1.0
|
282 |
+
self.p = p
|
283 |
+
self.transf = ColorJitter(
|
284 |
+
brightness=brightness, contrast=contrast, saturation=saturation, hue=hue
|
285 |
+
)
|
286 |
+
|
287 |
+
def __call__(self, img):
|
288 |
+
if random.random() < self.p:
|
289 |
+
return self.transf(img)
|
290 |
+
else:
|
291 |
+
return img
|
292 |
+
|
293 |
+
|
294 |
+
class _GrayScale(object):
|
295 |
+
"""
|
296 |
+
Apply Gray Scale to the PIL image with a specified probability.
|
297 |
+
"""
|
298 |
+
|
299 |
+
def __init__(self, p=0.2):
|
300 |
+
assert 0.0 <= p <= 1.0
|
301 |
+
self.p = p
|
302 |
+
self.transf = Grayscale(num_output_channels=3)
|
303 |
+
|
304 |
+
def __call__(self, img):
|
305 |
+
if random.random() < self.p:
|
306 |
+
return self.transf(img)
|
307 |
+
else:
|
308 |
+
return img
|
309 |
+
|
310 |
+
|
311 |
+
def image_transform(
|
312 |
+
image_size: Union[int, Tuple[int, int]],
|
313 |
+
is_train: bool,
|
314 |
+
mean: Optional[Tuple[float, ...]] = None,
|
315 |
+
std: Optional[Tuple[float, ...]] = None,
|
316 |
+
resize_mode: Optional[str] = None,
|
317 |
+
interpolation: Optional[str] = None,
|
318 |
+
fill_color: int = 0,
|
319 |
+
aug_cfg: Optional[Union[Dict[str, Any], AugmentationCfg]] = None,
|
320 |
+
):
|
321 |
+
mean = mean or OPENAI_DATASET_MEAN
|
322 |
+
if not isinstance(mean, (list, tuple)):
|
323 |
+
mean = (mean,) * 3
|
324 |
+
|
325 |
+
std = std or OPENAI_DATASET_STD
|
326 |
+
if not isinstance(std, (list, tuple)):
|
327 |
+
std = (std,) * 3
|
328 |
+
|
329 |
+
interpolation = interpolation or 'bicubic'
|
330 |
+
assert interpolation in ['bicubic', 'bilinear', 'random']
|
331 |
+
# NOTE random is ignored for interpolation_mode, so defaults to BICUBIC for
|
332 |
+
# inference if set
|
333 |
+
interpolation_mode = (
|
334 |
+
InterpolationMode.BILINEAR
|
335 |
+
if interpolation == 'bilinear'
|
336 |
+
else InterpolationMode.BICUBIC
|
337 |
+
)
|
338 |
+
|
339 |
+
resize_mode = resize_mode or 'shortest'
|
340 |
+
assert resize_mode in ('shortest', 'longest', 'squash')
|
341 |
+
|
342 |
+
if isinstance(aug_cfg, dict):
|
343 |
+
aug_cfg = AugmentationCfg(**aug_cfg)
|
344 |
+
else:
|
345 |
+
aug_cfg = aug_cfg or AugmentationCfg()
|
346 |
+
|
347 |
+
normalize = Normalize(mean=mean, std=std)
|
348 |
+
|
349 |
+
if is_train:
|
350 |
+
aug_cfg_dict = {k: v for k, v in asdict(aug_cfg).items() if v is not None}
|
351 |
+
use_timm = aug_cfg_dict.pop('use_timm', False)
|
352 |
+
if use_timm:
|
353 |
+
from timm.data import create_transform # timm can still be optional
|
354 |
+
|
355 |
+
if isinstance(image_size, (tuple, list)):
|
356 |
+
assert len(image_size) >= 2
|
357 |
+
input_size = (3,) + image_size[-2:]
|
358 |
+
else:
|
359 |
+
input_size = (3, image_size, image_size)
|
360 |
+
|
361 |
+
aug_cfg_dict.setdefault('color_jitter', None) # disable by default
|
362 |
+
# drop extra non-timm items
|
363 |
+
aug_cfg_dict.pop('color_jitter_prob', None)
|
364 |
+
aug_cfg_dict.pop('gray_scale_prob', None)
|
365 |
+
|
366 |
+
train_transform = create_transform(
|
367 |
+
input_size=input_size,
|
368 |
+
is_training=True,
|
369 |
+
hflip=0.0,
|
370 |
+
mean=mean,
|
371 |
+
std=std,
|
372 |
+
re_mode='pixel',
|
373 |
+
interpolation=interpolation,
|
374 |
+
**aug_cfg_dict,
|
375 |
+
)
|
376 |
+
else:
|
377 |
+
train_transform = [
|
378 |
+
RandomResizedCrop(
|
379 |
+
image_size,
|
380 |
+
scale=aug_cfg_dict.pop('scale'),
|
381 |
+
interpolation=InterpolationMode.BICUBIC,
|
382 |
+
),
|
383 |
+
_convert_to_rgb,
|
384 |
+
]
|
385 |
+
if aug_cfg.color_jitter_prob:
|
386 |
+
assert (
|
387 |
+
aug_cfg.color_jitter is not None and len(aug_cfg.color_jitter) == 4
|
388 |
+
)
|
389 |
+
train_transform.extend(
|
390 |
+
[_ColorJitter(*aug_cfg.color_jitter, p=aug_cfg.color_jitter_prob)]
|
391 |
+
)
|
392 |
+
if aug_cfg.gray_scale_prob:
|
393 |
+
train_transform.extend([_GrayScale(aug_cfg.gray_scale_prob)])
|
394 |
+
train_transform.extend(
|
395 |
+
[
|
396 |
+
ToTensor(),
|
397 |
+
normalize,
|
398 |
+
]
|
399 |
+
)
|
400 |
+
train_transform = Compose(train_transform)
|
401 |
+
if aug_cfg_dict:
|
402 |
+
warnings.warn(
|
403 |
+
f'Unused augmentation cfg items, specify `use_timm` to use '
|
404 |
+
f'({list(aug_cfg_dict.keys())}).'
|
405 |
+
)
|
406 |
+
return train_transform
|
407 |
+
else:
|
408 |
+
if resize_mode == 'longest':
|
409 |
+
transforms = [
|
410 |
+
ResizeKeepRatio(
|
411 |
+
image_size, interpolation=interpolation_mode, longest=1
|
412 |
+
),
|
413 |
+
CenterCropOrPad(image_size, fill=fill_color),
|
414 |
+
]
|
415 |
+
elif resize_mode == 'squash':
|
416 |
+
if isinstance(image_size, int):
|
417 |
+
image_size = (image_size, image_size)
|
418 |
+
transforms = [
|
419 |
+
Resize(image_size, interpolation=interpolation_mode),
|
420 |
+
]
|
421 |
+
else:
|
422 |
+
assert resize_mode == 'shortest'
|
423 |
+
if not isinstance(image_size, (tuple, list)):
|
424 |
+
image_size = (image_size, image_size)
|
425 |
+
if image_size[0] == image_size[1]:
|
426 |
+
# simple case, use torchvision built-in Resize w/ shortest edge mode
|
427 |
+
# (scalar size arg)
|
428 |
+
transforms = [Resize(image_size[0], interpolation=interpolation_mode)]
|
429 |
+
else:
|
430 |
+
# resize shortest edge to matching target dim for non-square target
|
431 |
+
transforms = [ResizeKeepRatio(image_size)]
|
432 |
+
transforms += [CenterCrop(image_size)]
|
433 |
+
|
434 |
+
transforms.extend(
|
435 |
+
[
|
436 |
+
_convert_to_rgb,
|
437 |
+
ToTensor(),
|
438 |
+
normalize,
|
439 |
+
]
|
440 |
+
)
|
441 |
+
return Compose(transforms)
|
442 |
+
|
443 |
+
|
444 |
+
def image_transform_v2(
|
445 |
+
cfg: PreprocessCfg,
|
446 |
+
is_train: bool,
|
447 |
+
aug_cfg: Optional[Union[Dict[str, Any], AugmentationCfg]] = None,
|
448 |
+
):
|
449 |
+
return image_transform(
|
450 |
+
image_size=cfg.size,
|
451 |
+
is_train=is_train,
|
452 |
+
mean=cfg.mean,
|
453 |
+
std=cfg.std,
|
454 |
+
interpolation=cfg.interpolation,
|
455 |
+
resize_mode=cfg.resize_mode,
|
456 |
+
fill_color=cfg.fill_color,
|
457 |
+
aug_cfg=aug_cfg,
|
458 |
+
)
|