File size: 2,755 Bytes
8b26dd7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
from copy import deepcopy
from typing import Optional

import torch
from transformers import AutoConfig, VisionTextDualEncoderConfig
from transformers.utils import logging

logger = logging.get_logger(__name__)


class CustomCLIPPooler(torch.nn.Module):
    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
        first_token_tensor = hidden_states[:, 0, :]
        return first_token_tensor


def get_text_model_pooler(text_model_pooler: str) -> torch.nn.Module:
    if text_model_pooler == "CustomCLIPPooler":
        return CustomCLIPPooler
    else:
        raise ValueError(f"Unrecognized text model pooler type {text_model_pooler!r}.")


def is_valid_text_model_pooler(
    text_model_pooler: str, suppress_error: bool = False
) -> bool:
    try:
        get_text_model_pooler(text_model_pooler)
    except ValueError:
        if not suppress_error:
            raise
        return False
    else:
        return True


class CustomCLIPConfig(VisionTextDualEncoderConfig):
    model_type = "custom-clip-model"

    DEFAULT_TEXT_MODEL_POOLER_STR: str = "CustomCLIPPooler"
    DEFAULT_TEXT_MODEL_POOLER_KWARGS: dict = {}

    def __init__(
        self,
        *args,
        text_model_pooler: Optional[str] = None,
        text_model_pooler_kwargs: Optional[dict] = None,
        **kwargs,
    ):
        super().__init__(*args, **kwargs)

        self.text_model_pooler = (
            self.DEFAULT_TEXT_MODEL_POOLER_STR
            if text_model_pooler is None
            else text_model_pooler
        )
        is_valid_text_model_pooler(self.text_model_pooler, suppress_error=False)

        self.text_model_pooler_kwargs = (
            self.DEFAULT_TEXT_MODEL_POOLER_KWARGS
            if text_model_pooler_kwargs is None
            else text_model_pooler_kwargs
        )

    @classmethod
    def from_base(cls, obj: VisionTextDualEncoderConfig):
        if not isinstance(obj, cls):
            base = VisionTextDualEncoderConfig
            if not isinstance(obj, base):
                raise TypeError(f"obj must be of type {cls!r} or {base!r}.")
            obj = deepcopy(obj)
            logger.warning(f"Changing config class from {obj.__class__!r} to {cls!r}.")
            obj.__class__ = cls

            def setattr_with_warning(object, name, value):
                logger.warning(f"Setting {name!r} to {value!r}.")
                setattr(object, name, value)

            setattr_with_warning(
                obj, "text_model_pooler", cls.DEFAULT_TEXT_MODEL_POOLER_STR
            )
            setattr_with_warning(
                obj, "text_model_pooler_kwargs", cls.DEFAULT_TEXT_MODEL_POOLER_KWARGS
            )
        return obj


AutoConfig.register(CustomCLIPConfig.model_type, CustomCLIPConfig)