train.py 4.1 KB
Newer Older
W
WuHaobo 已提交
1
# copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
W
WuHaobo 已提交
2
#
W
WuHaobo 已提交
3 4 5
# 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
W
WuHaobo 已提交
6 7 8
#
#    http://www.apache.org/licenses/LICENSE-2.0
#
W
WuHaobo 已提交
9 10 11 12 13
# 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.
W
add ad  
WuHaobo 已提交
14 15 16 17 18

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

W
WuHaobo 已提交
19 20 21 22 23 24 25
import argparse
import os

import paddle.fluid as fluid

from ppcls.data import Reader
from ppcls.utils.config import get_config
W
add ad  
WuHaobo 已提交
26
from ppcls.utils.save_load import init_model, save_model
littletomatodonkey's avatar
littletomatodonkey 已提交
27
from ppcls.utils import logger
W
add ad  
WuHaobo 已提交
28
import program
W
WuHaobo 已提交
29 30 31 32 33 34 35 36


def parse_args():
    parser = argparse.ArgumentParser("PaddleClas train script")
    parser.add_argument(
        '-c',
        '--config',
        type=str,
W
WuHaobo 已提交
37
        default='configs/ResNet/ResNet50.yaml',
W
WuHaobo 已提交
38 39 40 41 42 43 44 45 46 47 48 49 50
        help='config file path')
    parser.add_argument(
        '-o',
        '--override',
        action='append',
        default=[],
        help='config options to be overridden')
    args = parser.parse_args()
    return args


def main(args):
    config = get_config(args.config, overrides=args.override, show=True)
W
WuHaobo 已提交
51
    # assign the place
W
WuHaobo 已提交
52
    gpu_id = fluid.dygraph.parallel.Env().dev_id
W
WuHaobo 已提交
53
    place = fluid.CUDAPlace(gpu_id)
W
WuHaobo 已提交
54

W
WuHaobo 已提交
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
    with fluid.dygraph.guard(place):
        strategy = fluid.dygraph.parallel.prepare_context()
        net = program.create_model(config.ARCHITECTURE, config.classes_num)
        net = fluid.dygraph.parallel.DataParallel(net, strategy)

        optimizer = program.create_optimizer(
            config, parameter_list=net.parameters())

        # load model from checkpoint or pretrained model
        init_model(config, net, optimizer)

        train_dataloader = program.create_dataloader()
        train_reader = Reader(config, 'train')()
        train_dataloader.set_sample_list_generator(train_reader, place)

        if config.validate:
            valid_dataloader = program.create_dataloader()
            valid_reader = Reader(config, 'valid')()
            valid_dataloader.set_sample_list_generator(valid_reader, place)

        best_top1_acc = 0.0  # best top1 acc record
        for epoch_id in range(config.epochs):
            net.train()
            # 1. train with train dataset
            program.run(train_dataloader, config, net, optimizer, epoch_id,
                        'train')

            if fluid.dygraph.parallel.Env().local_rank == 0:
                # 2. validate with validate dataset
                if config.validate and epoch_id % config.valid_interval == 0:
                    net.eval()
                    top1_acc = program.run(valid_dataloader, config, net, None,
                                           epoch_id, 'valid')
                    if top1_acc > best_top1_acc:
                        best_top1_acc = top1_acc
                        message = "The best top1 acc {:.5f}, in epoch: {:d}".format(
                            best_top1_acc, epoch_id)
                        logger.info("{:s}".format(
                            logger.coloring(message, "RED")))
                        if epoch_id % config.save_interval == 0:

                            model_path = os.path.join(
                                config.model_save_dir,
                                config.ARCHITECTURE["name"])
                            save_model(net, optimizer, model_path,
                                       "best_model_in_epoch_" + str(epoch_id))

                # 3. save the persistable model
                if epoch_id % config.save_interval == 0:
                    model_path = os.path.join(config.model_save_dir,
W
WuHaobo 已提交
105
                                              config.ARCHITECTURE["name"])
W
WuHaobo 已提交
106
                    save_model(net, optimizer, model_path, epoch_id)
W
WuHaobo 已提交
107 108 109 110


if __name__ == '__main__':
    args = parse_args()
W
WuHaobo 已提交
111
    main(args)