RecPilot/save_stf_emb.py at main · RUCAIBox/RecPilot · 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
import argparse
import os
from logging import getLogger

import torch
import math
import numpy as np
import yaml
import importlib
from accelerate import Accelerator

from collator import Collator
from trainer import RLTrainer
from utils import *
from data_utils import *
import warnings
warnings.filterwarnings("ignore")

def parse_args():
    parser = argparse.ArgumentParser()
    parser.add_argument('--dataset', type=str, default='Musical_Instruments', help='Dataset name: Sports_and_Outdoors')
    parser.add_argument('--config_file', type=str, default='./config/ftconfig.yaml', help='Config file')
    return parser.parse_known_args()


def main(config):
    init_seed(config['rand_seed'], config['reproducibility'])
    init_logger(config)

    logger = getLogger()
    accelerator = config['accelerator']
    log(f'Device: {config["device"]}', accelerator, logger)
    log(f'Config: {str(config)}', accelerator, logger)

    # Tokenizer and Dataset
    tokenizer = get_tokenizer(config)
    train_dataset, valid_dataset, test_dataset = get_datasets(config)
    train_collate_fn = Collator(config, tokenizer,split="train")
    test_collate_fn = Collator(config, tokenizer,split="valid")

    model_name = config['model_name']
    model_class = getattr(importlib.import_module(f'models.{model_name}.model'), model_name)

    with accelerator.main_process_first():
        model = model_class(config, train_dataset, tokenizer)

        assert os.path.exists(config['pretrain_model']), "Pretrained model path does not exist."
        model_states = torch.load(config['pretrain_model'], map_location=accelerator.device)['model']
        model.load_state_dict(model_states)
        log(f"Loaded pretrained model from {config['pretrain_model']}", accelerator, logger)

    log(model, accelerator, logger)
    log(model.n_parameters, accelerator, logger)

    sft_embedding_table = model.t5.shared.weight.detach().cpu()

    torch.save(sft_embedding_table, config["save_path"])

    print(f"SFT Embedding Table saved. Shape: {sft_embedding_table.shape}")


    train_data = get_dataloader(config, train_dataset, train_collate_fn, 'train')
    valid_data = get_dataloader(config, valid_dataset, test_collate_fn,'valid')
    test_data = get_dataloader(config, test_dataset, test_collate_fn, 'test')

    if config['val_delay'] >= config['epochs']:
        config['val_delay'] = config['epochs'] - 1

    trainer = RLTrainer(config, model, tokenizer, train_data)

    trainer.model, test_data = accelerator.prepare(
        model, test_data
    )
    test_results, _ = trainer.evaluate(test_data, split='test', store=True)

    log(f'Test Results: {test_results}', accelerator, logger)
    trainer.end()

if __name__=="__main__":
    args, unparsed_args = parse_args()
    command_line_configs = parse_command_line_args(unparsed_args)

    # Config
    config = {}
    config.update(yaml.safe_load(open(args.config_file, 'r')))
    config.update(command_line_configs)

    config['run_local_time'] = get_local_time()

    ckpt_name = get_file_name(config)
    config['ckpt_name'] = ckpt_name
    config['dataset'] = args.dataset
    config['data_dir'] = os.path.join(config['data_dir'], config['dataset'])
    config['ckpt_dir'] = os.path.join(config['ckpt_dir'], config['dataset'], ckpt_name)

    config = convert_config_dict(config)

    config['device'], config['use_ddp'] = init_device()
    config['accelerator'] = Accelerator()
    torch.distributed.barrier(device_ids=[int(os.environ['LOCAL_RANK'])])

    main(config)