File size: 4,657 Bytes
59bb219
e7d3e2d
 
 
59bb219
 
e7d3e2d
 
98b4762
 
 
 
 
 
 
 
 
 
 
 
e7d3e2d
59bb219
 
e7d3e2d
 
 
 
 
 
0800885
e7d3e2d
 
 
 
 
59bb219
 
 
 
0800885
 
 
59bb219
 
ba043a3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
aac4b76
 
e7d3e2d
aac4b76
 
 
 
 
 
59bb219
 
e7d3e2d
59bb219
 
 
 
 
 
 
 
 
 
 
0800885
 
 
 
 
 
 
 
 
 
59bb219
0800885
 
 
 
 
 
 
 
 
59bb219
 
aac4b76
 
 
 
 
 
 
 
 
 
 
 
59bb219
 
 
 
 
 
 
 
 
 
 
 
 
ba043a3
 
 
 
 
 
 
 
 
 
 
 
 
 
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
"""Module containing the SimpleShareGPTPromptTokenizingStrategy class"""
from typing import Any, Dict, Optional

from fastchat.conversation import Conversation, SeparatorStyle, register_conv_template

from axolotl.prompt_tokenizers import ShareGPTPromptTokenizingStrategy
from axolotl.prompters import ShareGPTPrompterV2


def register_chatml_template(system_message=None):
    system_message = system_message or "You are a helpful assistant."
    register_conv_template(
        Conversation(
            name="chatml",
            system_template="<|im_start|>system\n{system_message}",
            system_message=system_message,
            roles=["<|im_start|>user", "<|im_start|>assistant"],
            sep_style=SeparatorStyle.CHATML,
            sep="<|im_end|>",
        )
    )


def load(tokenizer, cfg, ds_cfg: Optional[Dict[str, Any]] = None):
    conversation = (
        ds_cfg["conversation"] if ds_cfg and "conversation" in ds_cfg else None
    )
    field_human = ds_cfg["field_human"] if ds_cfg and "field_human" in ds_cfg else None
    field_model = ds_cfg["field_model"] if ds_cfg and "field_model" in ds_cfg else None
    strategy = SimpleShareGPTPromptTokenizingStrategy(
        ShareGPTPrompterV2(
            conversation=conversation,
            role_key_model=field_model,
            role_key_human=field_human,
        ),
        tokenizer,
        cfg.train_on_inputs,
        cfg.sequence_len,
    )
    if ds_cfg and "strict" in ds_cfg:
        strategy.strict = ds_cfg["strict"]
    return strategy


def load_ultrachat(tokenizer, cfg, ds_cfg: Optional[Dict[str, Any]] = None):
    conversation = (
        ds_cfg["conversation"] if ds_cfg and "conversation" in ds_cfg else None
    )
    strategy = UltrachatShareGPTPromptTokenizingStrategy(
        ShareGPTPrompterV2(
            conversation=conversation,
        ),
        tokenizer,
        cfg.train_on_inputs,
        cfg.sequence_len,
    )
    if ds_cfg and "strict" in ds_cfg:
        strategy.strict = ds_cfg["strict"]
    return strategy


def load_role(tokenizer, cfg):
    return SimpleRoleShareGPTPromptTokenizingStrategy(
        ShareGPTPrompterV2(),
        tokenizer,
        cfg.train_on_inputs,
        cfg.sequence_len,
    )


def load_guanaco(tokenizer, cfg):
    return GuanacoShareGPTPromptTokenizingStrategy(
        ShareGPTPrompterV2(),
        tokenizer,
        cfg.train_on_inputs,
        cfg.sequence_len,
    )


class SimpleShareGPTPromptTokenizingStrategy(ShareGPTPromptTokenizingStrategy):
    """
    basic sharegpt strategy to grab conversations from the sample row
    """

    _strict = True

    @property
    def strict(self):
        return self._strict

    @strict.setter
    def strict(self, strict):
        self._strict = strict

    def get_conversation_thread(self, prompt):
        conversations = prompt["conversations"]
        if self.strict:
            return conversations
        # remap roles - allow for assistant turn
        role_map = {"human": "human", "assistant": "gpt", "gpt": "gpt"}
        turns = [
            {"from": role_map[t["from"]], "value": t["value"]} for t in conversations
        ]
        return turns


class SimpleRoleShareGPTPromptTokenizingStrategy(ShareGPTPromptTokenizingStrategy):
    """
    basic sharegpt strategy to grab conversations from the sample row, but uses role instead of from
    """

    def get_conversation_thread(self, prompt):
        conversations = prompt["conversations"]
        # remap role: prompter/assistant, text: ... => from: human/gpt, value: ...
        turns = [{"from": t["role"], "value": t["value"]} for t in conversations]
        return turns


class GuanacoShareGPTPromptTokenizingStrategy(ShareGPTPromptTokenizingStrategy):
    """
    sharegpt strategy that remaps oasst data to sharegpt format
    """

    def get_conversation_thread(self, prompt):
        conversations = prompt["conversations"]
        # remap role: prompter/assistant, text: ... => from: human/gpt, value: ...
        role_map = {"prompter": "human", "assistant": "gpt"}
        turns = [
            {"from": role_map[t["role"]], "value": t["text"]} for t in conversations
        ]
        return turns


class UltrachatShareGPTPromptTokenizingStrategy(SimpleShareGPTPromptTokenizingStrategy):
    """
    sharegpt strategy that remaps ultrachat data to sharegpt format
    """

    def get_conversation_thread(self, prompt):
        conversations = prompt["messages"]
        role_map = {"user": "human", "assistant": "gpt"}
        turns = [
            {"from": role_map[t["role"]], "value": t["content"]} for t in conversations
        ]
        return turns