File size: 5,460 Bytes
e52682b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# Adopted from: https://github.com/haotian-liu/LLaVA. Below is the original copyright:
#    Copyright 2023 Haotian Liu
#
#    Licensed under the Apache License, Version 2.0 (the "License");
#    you may not use this file except in compliance with the License.
#    You may obtain a copy of the License at
#
#        http://www.apache.org/licenses/LICENSE-2.0
#
#    Unless required by applicable law or agreed to in writing, software
#    distributed under the License is distributed on an "AS IS" BASIS,
#    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
#    See the License for the specific language governing permissions and
#    limitations under the License.


from typing import List, Optional, Tuple, Union

import torch
import torch.nn as nn

from transformers import AutoConfig, AutoModelForCausalLM, \
                         LlamaConfig, LlamaModel, LlamaForCausalLM
from transformers.modeling_outputs import CausalLMOutputWithPast
from transformers.generation.utils import GenerateOutput

from .videollama2_arch import Videollama2MetaModel, Videollama2MetaForCausalLM


class Videollama2LlamaConfig(LlamaConfig):
    model_type = "videollama2_llama"

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
        self.model_type = "videollama2_llama"


class Videollama2LlamaModel(Videollama2MetaModel, LlamaModel):
    config_class = Videollama2LlamaConfig

    def __init__(self, config: LlamaConfig):
        super(Videollama2LlamaModel, self).__init__(config)


class Videollama2LlamaForCausalLM(LlamaForCausalLM, Videollama2MetaForCausalLM):
    config_class = Videollama2LlamaConfig

    def __init__(self, config, **kwargs):
        super(LlamaForCausalLM, self).__init__(config)
        self.model = Videollama2LlamaModel(config)
        self.pretraining_tp = config.pretraining_tp
        self.vocab_size = config.vocab_size
        self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False)

        # Initialize weights and apply final processing
        self.post_init()

    def get_model(self):
        return self.model

    def forward(
        self,
        input_ids: torch.LongTensor = None,
        attention_mask: Optional[torch.Tensor] = None,
        position_ids: Optional[torch.LongTensor] = None,
        past_key_values: Optional[List[torch.FloatTensor]] = None,
        inputs_embeds: Optional[torch.FloatTensor] = None,
        labels: Optional[torch.LongTensor] = None,
        use_cache: Optional[bool] = None,
        output_attentions: Optional[bool] = None,
        output_hidden_states: Optional[bool] = None,
        images: Optional[torch.FloatTensor] = None,
        return_dict: Optional[bool] = None,
        cache_position: Optional[torch.LongTensor] = None,
        **kwargs
    ) -> Union[Tuple, CausalLMOutputWithPast]:

        if inputs_embeds is None:
            (
                input_ids,
                attention_mask,
                past_key_values,
                inputs_embeds,
                labels
            ) = self.prepare_inputs_labels_for_multimodal(
                input_ids,
                attention_mask,
                past_key_values,
                labels,
                images
            )

        outputs = super().forward(
            input_ids=input_ids,
            attention_mask=attention_mask,
            past_key_values=past_key_values,
            inputs_embeds=inputs_embeds,
            labels=labels,
            use_cache=use_cache,
            output_attentions=output_attentions,
            output_hidden_states=output_hidden_states,
            return_dict=return_dict,
            cache_position=cache_position,
        )

        outputs.labels = labels

        return outputs

    @torch.no_grad()
    def generate(
        self,
        inputs: Optional[torch.Tensor] = None,
        images: Optional[torch.Tensor] = None,
        **kwargs,
    ) -> Union[GenerateOutput, torch.LongTensor]:
        position_ids = kwargs.pop("position_ids", None)
        attention_mask = kwargs.pop("attention_mask", None)
        if "inputs_embeds" in kwargs:
            raise NotImplementedError("`inputs_embeds` is not supported")

        if images is not None:
            (
                input_ids,
                attention_mask,
                past_key_values,
                inputs_embeds,
                _
            ) = self.prepare_inputs_labels_for_multimodal(
                input_ids=inputs,
                attention_mask=attention_mask,
                past_key_values=None,
                labels=None,
                images=images
            )
        else:
            inputs_embeds = self.get_model().embed_tokens(inputs)

        return super().generate(
            position_ids=position_ids,
            attention_mask=attention_mask,
            inputs_embeds=inputs_embeds,
            **kwargs
        )

    def prepare_inputs_for_generation(self, input_ids, past_key_values=None, inputs_embeds=None, **kwargs):
        images = kwargs.pop("images", None)
        _inputs = super().prepare_inputs_for_generation(
            input_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, **kwargs
        )
        if images is not None:
            _inputs['images'] = images
        return _inputs


AutoConfig.register("videollama2_llama", Videollama2LlamaConfig)
AutoModelForCausalLM.register(Videollama2LlamaConfig, Videollama2LlamaForCausalLM)