File size: 6,614 Bytes
924daa7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
# AUTOGENERATED! DO NOT EDIT! File to edit: ../notebooks/12_modelling.ipynb.

# %% auto 0
__all__ = ['VTDEConfig', 'VTDEModel']

# %% ../notebooks/12_modelling.ipynb 1
from transformers.models.clip.modeling_clip import CLIPOutput, clip_loss
from typing import Optional, Tuple, Union
from transformers import VisionTextDualEncoderConfig, AutoModel, PreTrainedModel, VisionTextDualEncoderModel
import torch

class VTDEConfig(VisionTextDualEncoderConfig):
    model_type = "vtde"
    
    def __init__(self, projection_dim=512, logit_scale_init_value=2.6592, 
    text_pooling_mode='mean',
    vision_pooling_mode='max',
    **kwargs):
        """
        pooling_mode in ['mean', 'max', 'cls']
        https://arxiv.org/pdf/2210.09996.pdf
        https://github.com/kahnchana/clippy/blob/3c102c29c32f7c66c6e52e09b795fe9c061bbb03/src/open_clip/hf_model.py#L56
        """
        self.text_pooling_mode = text_pooling_mode
        self.vision_pooling_mode = vision_pooling_mode
        super().__init__(projection_dim, logit_scale_init_value, **kwargs)

class VTDEModel(VisionTextDualEncoderModel):
    config_class = VTDEConfig
    base_model_prefix = "vtde"

    def __init__(
        self,
        config: Optional[VTDEConfig] = None,
        vision_model: Optional[PreTrainedModel] = None,
        text_model: Optional[PreTrainedModel] = None,
    ):
        # You can customize the constructor if needed
        super().__init__(config, vision_model, text_model)
        self.text_pooling_mode = config.text_pooling_mode
        self.vision_pooling_mode = config.vision_pooling_mode

    def get_text_features(
        self,
        input_ids=None,
        attention_mask=None,
        position_ids=None,
        token_type_ids=None,
        output_attentions=None,
        output_hidden_states=None,
        return_dict=None,
    ):
        text_outputs = self.text_model(
            input_ids=input_ids,
            attention_mask=attention_mask,
            token_type_ids=token_type_ids,
            position_ids=position_ids,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )
        if self.text_pooling_mode == 'cls':
            pooled_output = text_outputs[1]
        elif self.text_pooling_mode == 'mean':
            pooled_output = torch.mean(text_outputs[0], dim=1)
        elif self.text_pooling_mode == 'max':
            pooled_output = torch.max(text_outputs[0], dim=1)[0]
        elif self.text_pooling_mode == 'norm':
            """we select the patch with the largest norm"""
            last_hidden_states = text_outputs[0]
            patch_norms = torch.norm(last_hidden_states[:, 1:, :], dim=-1)
            max_norm_idx = torch.argmax(patch_norms, dim=1)
            pooled_output = last_hidden_states[:, max_norm_idx, :][:, 0, :]
        else:
            "We want to raise the name of the pooling mode"
            raise NotImplementedError

        text_features = self.text_projection(pooled_output)

        return text_features

    def get_image_features(
        self,
        pixel_values=None,
        output_attentions=None,
        output_hidden_states=None,
        return_dict=None,
    ):
        vision_outputs = self.vision_model(
            pixel_values=pixel_values,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        if self.vision_pooling_mode == 'cls':
            pooled_output = vision_outputs[1]
        elif self.vision_pooling_mode == 'mean':
            pooled_output = torch.mean(vision_outputs[0], dim=1)
        elif self.vision_pooling_mode == 'max':
            pooled_output = torch.max(vision_outputs[0], dim=1)[0]
        elif self.vision_pooling_mode == 'norm':
            """we select the patch with the largest norm"""
            last_hidden_states = vision_outputs[0]
            patch_norms = torch.norm(last_hidden_states[:, 1:, :], dim=-1)
            max_norm_idx = torch.argmax(patch_norms, dim=1)
            pooled_output = last_hidden_states[:, max_norm_idx, :][:, 0, :]
        else:
            raise NotImplementedError

        image_features = self.visual_projection(pooled_output)

        return image_features

    def forward(
        self,
        input_ids: Optional[torch.LongTensor] = None,
        pixel_values: Optional[torch.FloatTensor] = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        return_loss: Optional[bool] = None,
        token_type_ids: Optional[torch.LongTensor] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        return_dict: Optional[bool] = None,
    ) -> Union[Tuple[torch.Tensor], CLIPOutput]:

        return_dict = return_dict if return_dict is not None else self.config.return_dict

        image_embeds = self.get_image_features(
            pixel_values=pixel_values,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        text_embeds = self.get_text_features(
            input_ids=input_ids,
            attention_mask=attention_mask,
            position_ids=position_ids,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
        )

        # normalized features
        image_embeds = image_embeds / image_embeds.norm(dim=-1, keepdim=True)
        text_embeds = text_embeds / text_embeds.norm(dim=-1, keepdim=True)

        # cosine similarity as logits
        logit_scale = self.logit_scale.exp()
        logits_per_text = torch.matmul(text_embeds, image_embeds.t()) * logit_scale
        logits_per_image = logits_per_text.T

        loss = None
        if return_loss:
            loss = clip_loss(logits_per_text)

        if not return_dict:
            output = (logits_per_image, logits_per_text, text_embeds, image_embeds, text_outputs, vision_outputs)
            return ((loss,) + output) if loss is not None else output

        return CLIPOutput(
            loss=loss,
            logits_per_image=logits_per_image,
            logits_per_text=logits_per_text,
            text_embeds=text_embeds,
            image_embeds=image_embeds,
            text_model_output=text_embeds,
            vision_model_output=image_embeds,
        )
    
VTDEConfig.register_for_auto_class()
VTDEModel.register_for_auto_class("AutoModel")