BAAI
/

File size: 7,085 Bytes
94953a3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
from functools import partial
from typing import List, Optional
from argparse import Namespace
import torch
from torch import nn
import torch.nn.functional as F
from transformers import PreTrainedModel, PreTrainedTokenizer

from .configuration_emu import EmuConfig
from .constants import *
from .modeling_llama import LlamaForCausalLM
from .visual import EVAVisionTransformer


class EmuPreTrainedModel(PreTrainedModel):
    config_class = EmuConfig
    base_model_prefix = "model"
    supports_gradient_checkpointing = False
    _no_split_modules = ["LlamaDecoderLayer", "Block"]
    _skip_keys_device_placement = "past_key_values"

    def _init_weights(self, module):
        std = self.config.initializer_range
        if isinstance(module, nn.Linear):
            module.weight.data.normal_(mean=0.0, std=std)
            if module.bias is not None:
                module.bias.data.zero_()
        elif isinstance(module, nn.Embedding):
            module.weight.data.normal_(mean=0.0, std=std)
            if module.padding_idx is not None:
                module.weight.data[module.padding_idx].zero_()

class EmuForClsAndRegression(EmuPreTrainedModel):

    def __init__(self, config):
        super(EmuForClsAndRegression, self).__init__(config)

        self.lm = LlamaForCausalLM(config=config)

        self.lm.model.embed_tokens.padding_idx = config.pad_token_id

    def get_num_layers(self):
        return len(self.lm.model.layers)

class EmuModel(EmuPreTrainedModel):

    def __init__(self, config):
        super().__init__(config)

        vision_config = Namespace(**config.vision_config)

        self.visual = EVAVisionTransformer(
            img_size=vision_config.image_size,
            patch_size=vision_config.patch_size,
            embed_dim=vision_config.width,
            depth=vision_config.layers,
            num_heads=vision_config.width // vision_config.head_width,
            mlp_ratio=vision_config.mlp_ratio,
            qkv_bias=vision_config.qkv_bias,
            drop_path_rate=vision_config.drop_path_rate,
            norm_layer=partial(nn.LayerNorm, eps=vision_config.layer_norm_eps),
            xattn=vision_config.xattn,
            postnorm=vision_config.postnorm,
        )

        self.decoder = EmuForClsAndRegression(config)

        self.gradient_checkpointing = False
        
        self.n_query = vision_config.n_query
        self.v_query = vision_config.v_query

    @property
    def device(self):
        return next(iter(self.parameters())).device

    @property
    def dtype(self):
        return next(iter(self.parameters())).dtype

    @torch.no_grad()
    def encode_image(self, image: torch.Tensor, *, n_query=None):
        n_query = n_query if n_query is not None else self.n_query

        image_embeds = self.visual(image)
        image_embeds = image_embeds[:, 1:, :]
        b, n, c = image_embeds.shape
        sqrt_n = int(n**0.5)
        image_embeds = image_embeds.permute(0, 2, 1).view(b, c, sqrt_n, sqrt_n)

        stride = int(sqrt_n // (n_query ** 0.5))
        image_embeds = F.avg_pool2d(image_embeds, kernel_size=(stride, stride), stride=stride)
        image_embeds = image_embeds.view(b, c, -1).permute(0, 2, 1).contiguous()
        return image_embeds


class EmuForCausalLM(EmuPreTrainedModel):
    _auto_class = "AutoModelForCausalLM"

    def __init__(self, config):
        super().__init__(config)

        self.config = config
        self.model = EmuModel(config)
        # LM to EVA
        self.project_down = nn.Linear(config.hidden_size, config.d_model, bias=False)
        # EVA to LM
        self.project_up = nn.Linear(config.d_model, config.hidden_size, bias=False)

        self.n_query = self.model.n_query
        self.image_placeholder = DEFAULT_IMG_TOKEN + DEFAULT_IMAGE_TOKEN * self.n_query + DEFAULT_IMG_END_TOKEN

    def device(self, module=None):
        if module is None:
            return next(self.parameters()).device
        return next(module.parameters()).device

    def dtype(self, module):
        if module is None:
            return next(self.parameters()).dtype
        return next(module.parameters()).dtype

    @torch.no_grad()
    def generate_image(
        self,
        text: List[str],
        tokenizer: PreTrainedTokenizer,
        image: Optional[torch.Tensor] = None,
        placeholder: str = DEFAULT_IMG_PLACEHOLDER,
    ):
        IMAGE, BOI = tokenizer.convert_tokens_to_ids([DEFAULT_IMAGE_TOKEN, DEFAULT_IMG_TOKEN])
        if image is not None:
            prompt_image_embeds = self.model.encode_image(image)
            _, _, c = prompt_image_embeds.shape
            prompt_image_embeds = prompt_image_embeds.view(-1, c)
            prompt_image_embeds = self.project_up(prompt_image_embeds)

        text = [t.replace(placeholder, self.image_placeholder) for t in text]

        target_image_embeds = None
        for num_img_token in range(self.n_query):
            if num_img_token == 0:
                text = [f"{t}{DEFAULT_IMG_TOKEN}" for t in text]
            else:
                text = [f"{t}{DEFAULT_IMAGE_TOKEN}" for t in text]

            inputs = tokenizer(text, padding="longest", return_tensors="pt")
            device = self.device(self.model.decoder.lm.model.embed_tokens)
            attention_mask = inputs.attention_mask.to(device)
            input_ids = inputs.input_ids.to(device) # B x N

            text_embeds = self.model.decoder.lm.model.embed_tokens(input_ids)

            image_idx = (input_ids == IMAGE)
            cumsum_idx = torch.flip(torch.cumsum(torch.flip(image_idx, dims=[1]), dim=1), dims=[1])
            if image is not None:
                prompt_idx = torch.logical_and(image_idx, cumsum_idx > num_img_token)
                text_embeds[prompt_idx] = prompt_image_embeds.to(text_embeds.device)

            if target_image_embeds is not None:
                target_idx = torch.logical_and(image_idx, torch.logical_and(cumsum_idx > 0, cumsum_idx <= num_img_token))
                text_embeds[target_idx] = self.project_up(target_image_embeds).to(text_embeds.device)

            outputs = self.model.decoder.lm.model(
                inputs_embeds=text_embeds,
                attention_mask=attention_mask,
                output_hidden_states=True,
                return_dict=True,
            )

            image_idx = (input_ids == IMAGE) + (input_ids == BOI)
            cumsum_idx = torch.flip(torch.cumsum(torch.flip(image_idx, dims=[1]), dim=1), dims=[1])
            target_idx = torch.logical_and(image_idx, torch.logical_and(cumsum_idx > 0, cumsum_idx <= num_img_token+1))

            hidden_states = outputs.hidden_states[-1]
            target_image_embeds = hidden_states[target_idx.to(hidden_states.device)]
            target_image_embeds = target_image_embeds.view(-1, target_image_embeds.shape[-1])
            target_image_embeds = self.project_down(target_image_embeds)

        _, C = target_image_embeds.shape
        B = hidden_states.shape[0]
        target_image_embeds = target_image_embeds.view(B, -1, C)

        return target_image_embeds