File size: 8,627 Bytes
5a7ab71
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
# Adopted from tatsu-lab@stanford_alpaca. Below is the original copyright:
#    Copyright 2023 Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li
#
#    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.

import copy
from dataclasses import dataclass, field
import json
import pathlib
from typing import Dict, Optional, Sequence

import numpy as np
import torch
from torch.utils.data import Dataset
import transformers
from transformers import Trainer
from transformers.trainer_pt_utils import LabelSmoother

from fastchat.conversation import get_default_conv_template, SeparatorStyle

IGNORE_TOKEN_ID = LabelSmoother.ignore_index


@dataclass
class ModelArguments:
    model_name_or_path: Optional[str] = field(default="facebook/opt-125m")


@dataclass
class DataArguments:
    data_path: str = field(
        default=None, metadata={"help": "Path to the training data."}
    )
    lazy_preprocess: bool = False


@dataclass
class TrainingArguments(transformers.TrainingArguments):
    cache_dir: Optional[str] = field(default=None)
    optim: str = field(default="adamw_torch")
    model_max_length: int = field(
        default=512,
        metadata={
            "help": "Maximum sequence length. Sequences will be right padded (and possibly truncated)."
        },
    )


local_rank = None


def rank0_print(*args):
    if local_rank == 0:
        print(*args)


def safe_save_model_for_hf_trainer(trainer: transformers.Trainer, output_dir: str):
    """Collects the state dict and dump to disk."""
    state_dict = trainer.model.state_dict()
    if trainer.args.should_save:
        cpu_state_dict = {key: value.cpu() for key, value in state_dict.items()}
        del state_dict
        trainer._save(output_dir, state_dict=cpu_state_dict)  # noqa


def preprocess(
    sources,
    tokenizer: transformers.PreTrainedTokenizer,
) -> Dict:
    conv = get_default_conv_template("vicuna").copy()
    roles = {"human": conv.roles[0], "gpt": conv.roles[1]}

    # Apply prompt templates
    conversations = []
    for i, source in enumerate(sources):
        if roles[source[0]["from"]] != conv.roles[0]:
            # Skip the first one if it is not from human
            source = source[1:]

        conv.messages = []
        for j, sentence in enumerate(source):
            role = roles[sentence["from"]]
            assert role == conv.roles[j % 2], f"{i}"
            conv.append_message(role, sentence["value"])
        conversations.append(conv.get_prompt())

    # Tokenize conversations
    input_ids = tokenizer(
        conversations,
        return_tensors="pt",
        padding="max_length",
        max_length=tokenizer.model_max_length,
        truncation=True,
    ).input_ids
    targets = input_ids.clone()

    assert conv.sep_style == SeparatorStyle.TWO

    # Mask targets
    sep = conv.sep + conv.roles[1] + ": "
    for conversation, target in zip(conversations, targets):
        total_len = int(target.ne(tokenizer.pad_token_id).sum())

        rounds = conversation.split(conv.sep2)
        cur_len = 1
        target[:cur_len] = IGNORE_TOKEN_ID
        for i, rou in enumerate(rounds):
            if rou == "":
                break

            parts = rou.split(sep)
            if len(parts) != 2:
                break
            parts[0] += sep
            round_len = len(tokenizer(rou).input_ids)
            instruction_len = len(tokenizer(parts[0]).input_ids) - 2

            target[cur_len : cur_len + instruction_len] = IGNORE_TOKEN_ID

            cur_len += round_len
        target[cur_len:] = IGNORE_TOKEN_ID

        if False:
            z = target.clone()
            z = torch.where(z == IGNORE_TOKEN_ID, tokenizer.unk_token_id, z)
            rank0_print(tokenizer.decode(z))

        if cur_len < tokenizer.model_max_length:
            if cur_len != total_len:
                target[:] = IGNORE_TOKEN_ID
                rank0_print(
                    f"WARNING: tokenization mismatch: {cur_len} vs. {total_len}."
                    f" (ignored)"
                )

    return dict(
        input_ids=input_ids,
        labels=targets,
        attention_mask=input_ids.ne(tokenizer.pad_token_id),
    )


class SupervisedDataset(Dataset):
    """Dataset for supervised fine-tuning."""

    def __init__(self, raw_data, tokenizer: transformers.PreTrainedTokenizer):
        super(SupervisedDataset, self).__init__()

        rank0_print("Formatting inputs...")
        sources = [example["conversations"] for example in raw_data]
        data_dict = preprocess(sources, tokenizer)

        self.input_ids = data_dict["input_ids"]
        self.labels = data_dict["labels"]
        self.attention_mask = data_dict["attention_mask"]

    def __len__(self):
        return len(self.input_ids)

    def __getitem__(self, i) -> Dict[str, torch.Tensor]:
        return dict(
            input_ids=self.input_ids[i],
            labels=self.labels[i],
            attention_mask=self.attention_mask[i],
        )


class LazySupervisedDataset(Dataset):
    """Dataset for supervised fine-tuning."""

    def __init__(self, raw_data, tokenizer: transformers.PreTrainedTokenizer):
        super(LazySupervisedDataset, self).__init__()
        self.tokenizer = tokenizer

        rank0_print("Formatting inputs...Skip in lazy mode")
        self.tokenizer = tokenizer
        self.raw_data = raw_data
        self.cached_data_dict = {}

    def __len__(self):
        return len(self.raw_data)

    def __getitem__(self, i) -> Dict[str, torch.Tensor]:
        if i in self.cached_data_dict:
            return self.cached_data_dict[i]

        ret = preprocess([self.raw_data[i]["conversations"]], self.tokenizer)
        ret = dict(
            input_ids=ret["input_ids"][0],
            labels=ret["labels"][0],
            attention_mask=ret["attention_mask"][0],
        )
        self.cached_data_dict[i] = ret

        return ret


def make_supervised_data_module(
    tokenizer: transformers.PreTrainedTokenizer, data_args
) -> Dict:
    """Make dataset and collator for supervised fine-tuning."""
    dataset_cls = (
        LazySupervisedDataset if data_args.lazy_preprocess else SupervisedDataset
    )
    rank0_print("Loading data...")
    raw_data = json.load(open(data_args.data_path, "r"))

    # Split train/test
    perm = np.random.permutation(len(raw_data))
    split = int(len(perm) * 0.98)
    train_indices = perm[:split]
    eval_indices = perm[split:]
    train_raw_data = [raw_data[i] for i in train_indices]
    eval_raw_data = [raw_data[i] for i in eval_indices]
    rank0_print(f"#train {len(train_raw_data)}, #eval {len(eval_raw_data)}")

    train_dataset = dataset_cls(train_raw_data, tokenizer=tokenizer)
    eval_dataset = dataset_cls(eval_raw_data, tokenizer=tokenizer)
    return dict(train_dataset=train_dataset, eval_dataset=eval_dataset)


def train():
    global local_rank

    parser = transformers.HfArgumentParser(
        (ModelArguments, DataArguments, TrainingArguments)
    )
    model_args, data_args, training_args = parser.parse_args_into_dataclasses()
    local_rank = training_args.local_rank
    model = transformers.AutoModelForCausalLM.from_pretrained(
        model_args.model_name_or_path,
        cache_dir=training_args.cache_dir,
    )
    model.config.use_cache = False
    tokenizer = transformers.AutoTokenizer.from_pretrained(
        model_args.model_name_or_path,
        cache_dir=training_args.cache_dir,
        model_max_length=training_args.model_max_length,
        padding_side="right",
        use_fast=False,
    )
    tokenizer.pad_token = tokenizer.unk_token

    data_module = make_supervised_data_module(tokenizer=tokenizer, data_args=data_args)
    trainer = Trainer(
        model=model, tokenizer=tokenizer, args=training_args, **data_module
    )

    if list(pathlib.Path(training_args.output_dir).glob("checkpoint-*")):
        trainer.train(resume_from_checkpoint=True)
    else:
        trainer.train()
    trainer.save_state()
    safe_save_model_for_hf_trainer(trainer=trainer, output_dir=training_args.output_dir)


if __name__ == "__main__":
    train()