ExpGuard/train.py at main · brightjade/ExpGuard · GitHub

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
import os
import os.path as osp
import glob
import argparse

import torch
from transformers import set_seed
from trl import SFTConfig, SFTTrainer
from accelerate import Accelerator

from dataset import load_data, custom_train_collate_fn
from model import load_model, load_tokenizer

import wandb


def main(args):
    tokenizer = load_tokenizer(args)            # Load tokenizer
    train_dataset = load_data(args, tokenizer)  # Load data
    model, peft_config = load_model(args)       # Load model

    # Configure trainer
    trainer = SFTTrainer(
        args=SFTConfig(
            output_dir=args.output_dir,
            dataloader_num_workers=args.num_workers,
            per_device_train_batch_size=args.per_device_train_batch_size,
            per_device_eval_batch_size=args.per_device_eval_batch_size,
            gradient_accumulation_steps=args.gradient_accumulation_steps,
            num_train_epochs=args.epochs,
            fp16=args.fp16,
            bf16=args.bf16,
            gradient_checkpointing=args.use_gradient_checkpointing,
            gradient_checkpointing_kwargs={"use_reentrant": False},
            optim="adamw_torch",
            learning_rate=args.learning_rate,
            warmup_ratio=args.warmup_ratio,
            weight_decay=args.weight_decay,
            lr_scheduler_type=args.lr_scheduler_type,
            logging_steps=args.logging_steps,
            save_strategy="epoch",
            save_only_model=True,
            torch_compile=args.torch_compile,
            report_to="wandb" if args.wandb_mode == "online" else "none",
            # * SFT arguments
            max_seq_length=args.max_seq_len,
            dataset_text_field="text",
            packing=args.packing,
            dataset_kwargs={"add_special_tokens": False, "append_concat_token": False},
        ),
        model=model,
        train_dataset=train_dataset,
        peft_config=peft_config,
        processing_class=tokenizer,
        data_collator=custom_train_collate_fn if not args.packing else None,
    )
    trainer.train()


if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="Guardrails training script")
    # Model arguments
    parser.add_argument("--model_type", type=str, default="qwen2.5-7b")
    parser.add_argument("--model_name_or_path", type=str, default="Qwen/Qwen2.5-7B")
    parser.add_argument("--cache_dir", type=str, default=None)
    # Data arguments
    parser.add_argument("--data_dir", type=str, default="data/")
    parser.add_argument("--dataset_name", type=str, default="expguardmix")
    parser.add_argument("--num_workers", type=int, default=4)
    parser.add_argument("--max_seq_len", type=int, default=4096)
    parser.add_argument("--packing", action="store_true")
    # Training arguments
    parser.add_argument("--output_dir", type=str, default="")
    parser.add_argument("--per_device_train_batch_size", type=int, default=4)
    parser.add_argument("--per_device_eval_batch_size", type=int, default=8)
    parser.add_argument("--gradient_accumulation_steps", type=int, default=2)
    parser.add_argument("--lr_scheduler_type", type=str, default="cosine")
    parser.add_argument("--learning_rate", type=float, default=5e-6)
    parser.add_argument("--warmup_ratio", type=float, default=0.03)
    parser.add_argument("--weight_decay", type=float, default=0.0)
    parser.add_argument("--max_grad_norm", type=float, default=1.0)
    parser.add_argument("--epochs", type=int, default=3)
    parser.add_argument("--logging_steps", type=float, default=0.1)
    parser.add_argument("--seed", type=int, default=42)
    parser.add_argument("--use_4bit_quantization", action="store_true")
    parser.add_argument("--use_8bit_quantization", action="store_true")
    parser.add_argument("--fp16", action="store_true")
    parser.add_argument("--bf16", action="store_true")
    parser.add_argument("--deterministic", action="store_true")
    parser.add_argument("--torch_compile", action="store_true")
    parser.add_argument("--use_peft_lora", action="store_true")
    parser.add_argument("--use_gradient_checkpointing", action="store_true")
    parser.add_argument("--disable_checkpointing", action="store_true")
    parser.add_argument("--do_train", action="store_true")
    parser.add_argument("--do_eval", action="store_true")
    parser.add_argument("--do_test", action="store_true")
    parser.add_argument("--local_files_only", action="store_true")
    parser.add_argument("--attn_implementation", type=str, default="flash_attention_2")
    parser.add_argument("--wandb_mode", type=str, default="disabled")
    args = parser.parse_args()

    # Set seed
    set_seed(args.seed, deterministic=args.deterministic)

    # Set number of threads for CPU computation
    torch.set_num_threads(1)

    # Set batch size
    world_size = torch.cuda.device_count()
    args.distributed = world_size != 1
    args.train_batch_size = args.per_device_train_batch_size * args.gradient_accumulation_steps * world_size
    args.eval_batch_size = args.per_device_eval_batch_size * world_size
    args.data_dir = osp.join(args.data_dir, args.dataset_name)

    # Set data type
    if args.bf16:
        args.dtype = torch.bfloat16
    elif args.fp16:
        args.dtype = torch.float16
    else:
        args.dtype = torch.float32

    # Set output directory
    args.group_name = osp.join(args.dataset_name, args.model_type)
    args.run_name = f"BS{args.train_batch_size}_LR{args.learning_rate}_W{args.warmup_ratio}_D{args.weight_decay}_E{args.epochs}_S{args.seed}"
    args.output_dir = osp.join(".checkpoints", args.group_name, args.run_name)

    # Do not overwrite checkpoint files
    if args.do_train and (
        glob.glob(osp.join(args.output_dir, "*.safetensors")) or    # if one checkpoint is saved
        glob.glob(osp.join(args.output_dir, "checkpoint-*"))        # if multiple checkpoints are saved
    ):
        raise FileExistsError(f"Output directory {args.output_dir} already exists.")

    # Set up wandb
    if args.wandb_mode == "online":
        accelerator = Accelerator()
        if accelerator.is_main_process:
            wandb.init(
                project="guardrails",
                group=args.group_name,
                name=args.run_name,
                mode=args.wandb_mode,
            )

    if args.cache_dir:
        os.makedirs(args.cache_dir, exist_ok=True)
    os.makedirs(args.output_dir, exist_ok=True)

    main(args)