File size: 9,656 Bytes
7311cdd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
# %%
# 
from typing import List, Union
import torch
import os
from torch import nn
from typing import Optional, Tuple

from functools import partial

MODEL_DICT = {}
LAYER_DICT = {}

class Llama(nn.Module):
    def __init__(self, model_id="meta-llama/Meta-Llama-3.1-8B"):
        super().__init__()
        
        import transformers
        
        access_token = os.getenv("HF_ACCESS_TOKEN")
        if access_token is None:
            raise ValueError("HF_ACCESS_TOKEN environment variable must be set")
        
        pipeline = transformers.pipeline(
            "text-generation",
            model=model_id,
            model_kwargs={"torch_dtype": torch.bfloat16},
            token=access_token,
            device='cpu',
        )
        
        tokenizer = pipeline.tokenizer
        model = pipeline.model
        
        def new_forward(
            self,
            hidden_states: torch.Tensor,
            attention_mask: Optional[torch.Tensor] = None,
            position_ids: Optional[torch.LongTensor] = None,
            past_key_value = None,
            output_attentions: Optional[bool] = False,
            use_cache: Optional[bool] = False,
            cache_position: Optional[torch.LongTensor] = None,
            position_embeddings: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,  # will become mandatory in v4.45
            **kwargs,
        ) -> Tuple[torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]]:
            residual = hidden_states

            hidden_states = self.input_layernorm(hidden_states)

            # Self Attention
            hidden_states, self_attn_weights, present_key_value = self.self_attn(
                hidden_states=hidden_states,
                attention_mask=attention_mask,
                position_ids=position_ids,
                past_key_value=past_key_value,
                output_attentions=output_attentions,
                use_cache=use_cache,
                cache_position=cache_position,
                position_embeddings=position_embeddings,
                **kwargs,
            )
            
            self.attn_output = hidden_states.clone()
            
            hidden_states = residual + hidden_states

            # Fully Connected
            residual = hidden_states
            hidden_states = self.post_attention_layernorm(hidden_states)
            hidden_states = self.mlp(hidden_states)
            
            self.mlp_output = hidden_states.clone()
            
            hidden_states = residual + hidden_states
            
            self.block_output = hidden_states.clone()

            outputs = (hidden_states,)

            if output_attentions:
                outputs += (self_attn_weights,)

            if use_cache:
                outputs += (present_key_value,)

            return outputs
        
        # for layer in model.model.layers:
            # setattr(layer.__class__, "forward", new_forward)
            # setattr(layer.__class__, "__call__", new_forward)
        setattr(model.model.layers[0].__class__, "forward", new_forward)
        setattr(model.model.layers[0].__class__, "__call__", new_forward)
        
        self.model = model
        self.tokenizer = tokenizer
    
    @torch.no_grad()
    def forward(self, text: str):
        encoded_input = self.tokenizer(text, return_tensors='pt')
        device = next(self.model.parameters()).device
        encoded_input = {k: v.to(device) for k, v in encoded_input.items()}
        output = self.model(**encoded_input, output_hidden_states=True)
    
        attn_outputs, mlp_outputs, block_outputs = [], [], []
        for i, blk in enumerate(self.model.model.layers):
            attn_outputs.append(blk.attn_output)
            mlp_outputs.append(blk.mlp_output)
            block_outputs.append(blk.block_output)
        
        token_ids = encoded_input['input_ids']
        token_texts = [self.tokenizer.decode([token_id]) for token_id in token_ids[0]]
        
        return {"attn": attn_outputs, "mlp": mlp_outputs, "block": block_outputs, "token_texts": token_texts}

MODEL_DICT["meta-llama/Meta-Llama-3.1-8B"] = partial(Llama, model_id="meta-llama/Meta-Llama-3.1-8B")
LAYER_DICT["meta-llama/Meta-Llama-3.1-8B"] = 32
MODEL_DICT["meta-llama/Meta-Llama-3-8B"] = partial(Llama, model_id="meta-llama/Meta-Llama-3-8B")
LAYER_DICT["meta-llama/Meta-Llama-3-8B"] = 32

class GPT2(nn.Module):
    def __init__(self):
        super().__init__()
        from transformers import GPT2Tokenizer, GPT2Model
        tokenizer = GPT2Tokenizer.from_pretrained('gpt2')
        model = GPT2Model.from_pretrained('gpt2')
        
        def new_forward(
            self,
            hidden_states: Optional[Tuple[torch.FloatTensor]],
            layer_past: Optional[Tuple[torch.Tensor]] = None,
            attention_mask: Optional[torch.FloatTensor] = None,
            head_mask: Optional[torch.FloatTensor] = None,
            encoder_hidden_states: Optional[torch.Tensor] = None,
            encoder_attention_mask: Optional[torch.FloatTensor] = None,
            use_cache: Optional[bool] = False,
            output_attentions: Optional[bool] = False,
        ) -> Union[Tuple[torch.Tensor], Optional[Tuple[torch.Tensor, Tuple[torch.FloatTensor, ...]]]]:
            residual = hidden_states
            hidden_states = self.ln_1(hidden_states)
            attn_outputs = self.attn(
                hidden_states,
                layer_past=layer_past,
                attention_mask=attention_mask,
                head_mask=head_mask,
                use_cache=use_cache,
                output_attentions=output_attentions,
            )
            attn_output = attn_outputs[0]  # output_attn: a, present, (attentions)
            outputs = attn_outputs[1:]
            # residual connection
            self.attn_output = attn_output.clone()
            hidden_states = attn_output + residual

            if encoder_hidden_states is not None:
                # add one self-attention block for cross-attention
                if not hasattr(self, "crossattention"):
                    raise ValueError(
                        f"If `encoder_hidden_states` are passed, {self} has to be instantiated with "
                        "cross-attention layers by setting `config.add_cross_attention=True`"
                    )
                residual = hidden_states
                hidden_states = self.ln_cross_attn(hidden_states)
                cross_attn_outputs = self.crossattention(
                    hidden_states,
                    attention_mask=attention_mask,
                    head_mask=head_mask,
                    encoder_hidden_states=encoder_hidden_states,
                    encoder_attention_mask=encoder_attention_mask,
                    output_attentions=output_attentions,
                )
                attn_output = cross_attn_outputs[0]
                # residual connection
                hidden_states = residual + attn_output
                outputs = outputs + cross_attn_outputs[2:]  # add cross attentions if we output attention weights

            residual = hidden_states
            hidden_states = self.ln_2(hidden_states)
            feed_forward_hidden_states = self.mlp(hidden_states)
            # residual connection
            self.mlp_output = feed_forward_hidden_states.clone()
            hidden_states = residual + feed_forward_hidden_states

            if use_cache:
                outputs = (hidden_states,) + outputs
            else:
                outputs = (hidden_states,) + outputs[1:]

            self.block_output = hidden_states.clone()
            return outputs  # hidden_states, present, (attentions, cross_attentions)
        
        setattr(model.h[0].__class__, "forward", new_forward)
        
        self.model = model
        self.tokenizer = tokenizer

    @torch.no_grad()
    def forward(self, text: str):
        encoded_input = self.tokenizer(text, return_tensors='pt')
        device = next(self.model.parameters()).device
        encoded_input = {k: v.to(device) for k, v in encoded_input.items()}
        output = self.model(**encoded_input, output_hidden_states=True)
    
        attn_outputs, mlp_outputs, block_outputs = [], [], []
        for i, blk in enumerate(self.model.h):
            attn_outputs.append(blk.attn_output)
            mlp_outputs.append(blk.mlp_output)
            block_outputs.append(blk.block_output)
            
        token_ids = encoded_input['input_ids']
        token_texts = [self.tokenizer.decode([token_id]) for token_id in token_ids[0]]
        
        return {"attn": attn_outputs, "mlp": mlp_outputs, "block": block_outputs, "token_texts": token_texts}
    
MODEL_DICT["gpt2"] = GPT2
LAYER_DICT["gpt2"] = 12


def download_all_models():
    for model_name in MODEL_DICT:
        print(f"Downloading {model_name}")
        try:
            model = MODEL_DICT[model_name]()
        except Exception as e:
            print(f"Error downloading {model_name}: {e}")
            continue


if __name__ == '__main__':

    model = MODEL_DICT["meta-llama/Meta-Llama-3-8B"]()
    # model = MODEL_DICT["gpt2"]()
    text = """
    1. The majestic giraffe, with its towering height and distinctive long neck, roams the savannas of Africa. These gentle giants use their elongated tongues to pluck leaves from the tallest trees, making them well-adapted to their environment. Their unique coat patterns, much like human fingerprints, are unique to each individual.
    """
    model = model.cuda()
    output = model(text)
    print(output["block"][1].shape)
    print(output["token_texts"])