File size: 4,921 Bytes
94c4ed7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# coding=utf-8
# Copyright 2024 The Qwen team, Alibaba Group and the HuggingFace Inc. team. All rights reserved.
#
# 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.
"""Qwen2 model configuration"""

import torch
from dataclasses import dataclass, asdict
from enum import Enum

from transformers.configuration_utils import PretrainedConfig, layer_type_validation
from transformers.modeling_rope_utils import rope_config_validation
from transformers.utils import logging

from transformers.models.qwen2.configuration_qwen2 import Qwen2Config

from quasar.kernel.configs import QuantType

logger = logging.get_logger(__name__)


@dataclass
class FP8Config:
    """
    Configuration for FP8 quantization.
    """

    float8_dtype: torch.dtype = torch.float8_e4m3fn
    quant_type: QuantType = QuantType.DIV
    layer_name: str = ""

    act_block_size: int = 16
    mm_block_size: int = 128

    training_mode: bool = True
    """
    If True, the linear layer will use high-precision weight.
    If False, the linear layer will use per-block quantized weight.
    """


class FP8Qwen2Config(Qwen2Config):
    model_type = "fp8_qwen2"
    fp8_config: FP8Config = FP8Config()
    model_name_orig: str = ""
    """Pass the name of the BF16 model"""

    def __init__(
        self,
        vocab_size=151936,
        hidden_size=4096,
        intermediate_size=22016,
        num_hidden_layers=32,
        num_attention_heads=32,
        num_key_value_heads=32,
        hidden_act="silu",
        max_position_embeddings=32768,
        initializer_range=0.02,
        rms_norm_eps=1e-6,
        use_cache=True,
        tie_word_embeddings=False,
        rope_theta=10000.0,
        rope_scaling=None,
        use_sliding_window=False,
        sliding_window=4096,
        max_window_layers=28,
        layer_types=None,
        attention_dropout=0.0,
        # Customized configs begins here
        fp8_config=None,
        model_name_orig="",
        **kwargs,
    ):
        super().__init__(
            vocab_size=vocab_size,
            hidden_size=hidden_size,
            intermediate_size=intermediate_size,
            num_hidden_layers=num_hidden_layers,
            num_attention_heads=num_attention_heads,
            num_key_value_heads=num_key_value_heads,
            hidden_act=hidden_act,
            max_position_embeddings=max_position_embeddings,
            initializer_range=initializer_range,
            rms_norm_eps=rms_norm_eps,
            use_cache=use_cache,
            tie_word_embeddings=tie_word_embeddings,
            rope_theta=rope_theta,
            rope_scaling=rope_scaling,
            use_sliding_window=use_sliding_window,
            sliding_window=sliding_window,
            max_window_layers=max_window_layers,
            layer_types=layer_types,
            attention_dropout=attention_dropout,
            **kwargs,
        )


        # Convert it from dict to FP8Config (dataclass)
        if fp8_config is not None:
            self.fp8_config = fp8_config if isinstance(fp8_config, FP8Config) else FP8Config(**fp8_config)
        else:
            self.fp8_config = FP8Config()

        self.model_name_orig = model_name_orig
        

    def to_dict(self):
        output = super().to_dict()
        if hasattr(self.fp8_config, "__dataclass_fields__"):
            cfg_dict = asdict(self.fp8_config)
            for k, v in cfg_dict.items():
                if isinstance(v, torch.dtype): # float8_dtype
                    cfg_dict[k] = str(v)  # save as 'torch.float8_e4m3fn'
                elif isinstance(v, Enum): # quant_type
                    cfg_dict[k] = v.name   # save as 'DIV'
            output["fp8_config"] = cfg_dict
        else:
            output["fp8_config"] = self.fp8_config
        return output


    @classmethod
    def from_dict(cls, config_dict, **kwargs):
        config = super().from_dict(config_dict, **kwargs)
        
        fp8_config = config_dict.get("fp8_config", {})
        for k, v in fp8_config.items():
            if k == "float8_dtype":
                assert v.startswith("torch."), f"Invalid float8_dtype: {v}"
                fp8_config[k] = getattr(torch, v[len("torch."):]) #
            elif k == "quant_type":
                fp8_config[k] = getattr(QuantType, v)
        config.fp8_config = FP8Config(**fp8_config)
        return config
    
    
__all__ = ["FP8Qwen2Config"]

FP8Qwen2Config.register_for_auto_class()