trainer.py 15.0 KB
Newer Older
Q
qingqing01 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14
#   Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
#
# 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.

L
LielinJiang 已提交
15 16
import os
import time
L
LielinJiang 已提交
17
import copy
L
LielinJiang 已提交
18

L
LielinJiang 已提交
19
import logging
L
LielinJiang 已提交
20
import datetime
L
LielinJiang 已提交
21

L
LielinJiang 已提交
22
import paddle
L
LielinJiang 已提交
23
from paddle.distributed import ParallelEnv
L
LielinJiang 已提交
24 25 26 27

from ..datasets.builder import build_dataloader
from ..models.builder import build_model
from ..utils.visual import tensor2img, save_image
L
LielinJiang 已提交
28
from ..utils.filesystem import makedirs, save, load
29
from ..utils.timer import TimeAverager
L
LielinJiang 已提交
30

L
fix nan  
LielinJiang 已提交
31

32 33 34 35 36
class IterLoader:
    def __init__(self, dataloader):
        self._dataloader = dataloader
        self.iter_loader = iter(self._dataloader)
        self._epoch = 1
L
LielinJiang 已提交
37

38 39 40 41 42 43 44 45 46 47 48
    @property
    def epoch(self):
        return self._epoch

    def __next__(self):
        try:
            data = next(self.iter_loader)
        except StopIteration:
            self._epoch += 1
            self.iter_loader = iter(self._dataloader)
            data = next(self.iter_loader)
L
LielinJiang 已提交
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
        return data

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


class Trainer:
    """
    # trainer calling logic:
    #
    #                build_model                               ||    model(BaseModel)
    #                     |                                    ||
    #               build_dataloader                           ||    dataloader
    #                     |                                    ||
    #               model.setup_lr_schedulers                  ||    lr_scheduler
    #                     |                                    ||
    #               model.setup_optimizers                     ||    optimizers
    #                     |                                    ||
    #     train loop (model.setup_input + model.train_iter)    ||    train loop
    #                     |                                    ||
    #         print log (model.get_current_losses)             ||
    #                     |                                    ||
    #         save checkpoint (model.nets)                     \/
    """
    def __init__(self, cfg):
L
LielinJiang 已提交
75

L
LielinJiang 已提交
76
        # build model
77
        self.model = build_model(cfg.model)
78 79 80
        # multiple gpus prepare
        if ParallelEnv().nranks > 1:
            self.distributed_data_parallel()
L
LielinJiang 已提交
81

82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
        # build train dataloader
        self.train_dataloader = build_dataloader(cfg.dataset.train)
        self.iters_per_epoch = len(self.train_dataloader)

        # build lr scheduler
        # TODO: has a better way?
        if 'lr_scheduler' in cfg and 'iters_per_epoch' in cfg.lr_scheduler:
            cfg.lr_scheduler.iters_per_epoch = self.iters_per_epoch
        self.lr_schedulers = self.model.setup_lr_schedulers(cfg.lr_scheduler)

        # build optimizers
        self.optimizers = self.model.setup_optimizers(self.lr_schedulers,
                                                      cfg.optimizer)

        # build metrics
        self.metrics = None
        validate_cfg = cfg.get('validate', None)
        if validate_cfg and 'metrics' in validate_cfg:
            self.metrics = self.model.setup_metrics(validate_cfg['metrics'])

L
LielinJiang 已提交
102
        self.logger = logging.getLogger(__name__)
郑启航 已提交
103 104 105 106
        self.enable_visualdl = cfg.get('enable_visualdl', False)
        if self.enable_visualdl:
            import visualdl
            self.vdl_logger = visualdl.LogWriter(logdir=cfg.output_dir)
107

L
LielinJiang 已提交
108 109
        # base config
        self.output_dir = cfg.output_dir
L
lijianshe02 已提交
110
        self.max_eval_steps = cfg.model.get('max_eval_steps', None)
111 112 113 114 115 116 117 118
        self.epochs = cfg.get('epochs', None)
        if self.epochs:
            self.total_iters = self.epochs * self.iters_per_epoch
            self.by_epoch = True
        else:
            self.by_epoch = False
            self.total_iters = cfg.total_iters

L
LielinJiang 已提交
119 120
        self.start_epoch = 1
        self.current_epoch = 1
121 122
        self.current_iter = 1
        self.inner_iter = 1
L
LielinJiang 已提交
123
        self.batch_id = 0
郑启航 已提交
124
        self.global_steps = 0
L
LielinJiang 已提交
125 126 127
        self.weight_interval = cfg.snapshot_config.interval
        self.log_interval = cfg.log_config.interval
        self.visual_interval = cfg.log_config.visiual_interval
L
LielinJiang 已提交
128 129 130
        if self.by_epoch:
            self.weight_interval *= self.iters_per_epoch

L
LielinJiang 已提交
131 132 133
        self.validate_interval = -1
        if cfg.get('validate', None) is not None:
            self.validate_interval = cfg.validate.get('interval', -1)
L
LielinJiang 已提交
134 135 136
        self.cfg = cfg

        self.local_rank = ParallelEnv().local_rank
137 138

        self.time_count = {}
L
LielinJiang 已提交
139 140
        self.best_metric = {}

141
    def distributed_data_parallel(self):
L
LielinJiang 已提交
142
        strategy = paddle.distributed.prepare_context()
143 144
        for net_name, net in self.model.nets.items():
            self.model.nets[net_name] = paddle.DataParallel(net, strategy)
145

L
LielinJiang 已提交
146
    def train(self):
147 148
        reader_cost_averager = TimeAverager()
        batch_cost_averager = TimeAverager()
L
LielinJiang 已提交
149

150
        iter_loader = IterLoader(self.train_dataloader)
L
LielinJiang 已提交
151

152 153 154
        while self.current_iter < (self.total_iters + 1):
            self.current_epoch = iter_loader.epoch
            self.inner_iter = self.current_iter % self.iters_per_epoch
L
LielinJiang 已提交
155

156 157 158 159 160 161 162 163 164 165 166
            start_time = step_start_time = time.time()
            data = next(iter_loader)
            reader_cost_averager.record(time.time() - step_start_time)
            # unpack data from dataset and apply preprocessing
            # data input should be dict
            self.model.setup_input(data)
            self.model.train_iter(self.optimizers)

            batch_cost_averager.record(time.time() - step_start_time,
                                       num_samples=self.cfg.get(
                                           'batch_size', 1))
167 168 169

            step_start_time = time.time()

170 171 172 173 174 175 176 177 178 179 180 181 182
            if self.current_iter % self.log_interval == 0:
                self.data_time = reader_cost_averager.get_average()
                self.step_time = batch_cost_averager.get_average()
                self.ips = batch_cost_averager.get_ips_average()
                self.print_log()

                reader_cost_averager.reset()
                batch_cost_averager.reset()

            if self.current_iter % self.visual_interval == 0:
                self.visual('visual_train')

            self.model.lr_scheduler.step()
L
LielinJiang 已提交
183

L
LielinJiang 已提交
184
            if self.validate_interval > -1 and self.current_iter % self.validate_interval == 0:
185
                self.test()
L
fix nan  
LielinJiang 已提交
186

L
LielinJiang 已提交
187 188 189
            if self.current_iter % self.weight_interval == 0:
                self.save(self.current_iter, 'weight', keep=-1)
                self.save(self.current_iter)
L
LielinJiang 已提交
190

191
            self.current_iter += 1
L
LielinJiang 已提交
192

L
LielinJiang 已提交
193 194
    def test(self):
        if not hasattr(self, 'test_dataloader'):
195
            self.test_dataloader = build_dataloader(self.cfg.dataset.test,
196 197
                                                    is_train=False,
                                                    distributed=False)
L
lijianshe02 已提交
198 199 200
        iter_loader = IterLoader(self.test_dataloader)
        if self.max_eval_steps is None:
            self.max_eval_steps = len(self.test_dataloader)
201 202 203 204

        if self.metrics:
            for metric in self.metrics.values():
                metric.reset()
L
LielinJiang 已提交
205

L
lijianshe02 已提交
206 207
        for i in range(self.max_eval_steps):
            data = next(iter_loader)
208 209
            self.model.setup_input(data)
            self.model.test_iter(metrics=self.metrics)
L
LielinJiang 已提交
210 211

            visual_results = {}
L
LielinJiang 已提交
212 213 214
            current_paths = self.model.get_image_paths()
            current_visuals = self.model.get_current_visuals()

L
LielinJiang 已提交
215 216 217 218 219 220 221 222 223 224 225 226
            if len(current_visuals) > 0 and list(
                    current_visuals.values())[0].shape == 4:
                num_samples = list(current_visuals.values())[0].shape[0]
            else:
                num_samples = 1

            for j in range(num_samples):
                if j < len(current_paths):
                    short_path = os.path.basename(current_paths[j])
                    basename = os.path.splitext(short_path)[0]
                else:
                    basename = '{:04d}_{:04d}'.format(i, j)
L
LielinJiang 已提交
227 228
                for k, img_tensor in current_visuals.items():
                    name = '%s_%s' % (basename, k)
L
LielinJiang 已提交
229 230 231 232
                    if len(img_tensor.shape) == 4:
                        visual_results.update({name: img_tensor[j]})
                    else:
                        visual_results.update({name: img_tensor})
L
LielinJiang 已提交
233

郑启航 已提交
234 235 236 237
            self.visual('visual_test',
                        visual_results=visual_results,
                        step=self.batch_id,
                        is_save_image=True)
L
LielinJiang 已提交
238

L
LielinJiang 已提交
239
            if i % self.log_interval == 0:
240
                self.logger.info('Test iter: [%d/%d]' %
L
lijianshe02 已提交
241
                                 (i, self.max_eval_steps))
L
LielinJiang 已提交
242

243 244 245 246 247
        if self.metrics:
            for metric_name, metric in self.metrics.items():
                self.logger.info("Metric {}: {:.4f}".format(
                    metric_name, metric.accumulate()))

L
LielinJiang 已提交
248 249
    def print_log(self):
        losses = self.model.get_current_losses()
L
LielinJiang 已提交
250

251 252 253 254 255 256 257 258 259
        message = ''
        if self.by_epoch:
            message += 'Epoch: %d/%d, iter: %d/%d ' % (
                self.current_epoch, self.epochs, self.inner_iter,
                self.iters_per_epoch)
        else:
            message += 'Iter: %d/%d ' % (self.current_iter, self.total_iters)

        message += f'lr: {self.current_learning_rate:.3e} '
L
LielinJiang 已提交
260 261 262

        for k, v in losses.items():
            message += '%s: %.3f ' % (k, v)
郑启航 已提交
263 264
            if self.enable_visualdl:
                self.vdl_logger.add_scalar(k, v, step=self.global_steps)
L
LielinJiang 已提交
265

266 267 268
        if hasattr(self, 'step_time'):
            message += 'batch_cost: %.5f sec ' % self.step_time

269
        if hasattr(self, 'data_time'):
270
            message += 'reader_cost: %.5f sec ' % self.data_time
271

272
        if hasattr(self, 'ips'):
L
LielinJiang 已提交
273 274 275
            message += 'ips: %.5f images/s ' % self.ips

        if hasattr(self, 'step_time'):
276
            eta = self.step_time * (self.total_iters - self.current_iter - 1)
L
LielinJiang 已提交
277 278
            eta_str = str(datetime.timedelta(seconds=int(eta)))
            message += f'eta: {eta_str}'
279

L
LielinJiang 已提交
280 281 282 283 284
        # print the message
        self.logger.info(message)

    @property
    def current_learning_rate(self):
L
LielinJiang 已提交
285 286
        for optimizer in self.model.optimizers.values():
            return optimizer.get_lr()
L
LielinJiang 已提交
287

郑启航 已提交
288 289 290 291 292 293 294 295 296 297 298 299 300 301
    def visual(self,
               results_dir,
               visual_results=None,
               step=None,
               is_save_image=False):
        """
        visual the images, use visualdl or directly write to the directory

        Parameters:
            results_dir (str)     --  directory name which contains saved images
            visual_results (dict) --  the results images dict
            step (int)            --  global steps, used in visualdl
            is_save_image (bool)  --  weather write to the directory or visualdl
        """
L
LielinJiang 已提交
302 303 304 305 306
        self.model.compute_visuals()

        if visual_results is None:
            visual_results = self.model.get_current_visuals()

L
LielinJiang 已提交
307 308 309
        min_max = self.cfg.get('min_max', None)
        if min_max is None:
            min_max = (-1., 1.)
310

郑启航 已提交
311 312 313
        image_num = self.cfg.get('image_num', None)
        if (image_num is None) or (not self.enable_visualdl):
            image_num = 1
L
LielinJiang 已提交
314
        for label, image in visual_results.items():
郑启航 已提交
315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
            image_numpy = tensor2img(image, min_max, image_num)
            if (not is_save_image) and self.enable_visualdl:
                self.vdl_logger.add_image(
                    results_dir + '/' + label,
                    image_numpy,
                    step=step if step else self.global_steps,
                    dataformats="HWC" if image_num == 1 else "NCHW")
            else:
                if self.cfg.is_train:
                    msg = 'epoch%.3d_' % self.current_epoch
                else:
                    msg = ''
                makedirs(os.path.join(self.output_dir, results_dir))
                img_path = os.path.join(self.output_dir, results_dir,
                                        msg + '%s.png' % (label))
                save_image(image_numpy, img_path)
L
LielinJiang 已提交
331 332 333 334

    def save(self, epoch, name='checkpoint', keep=1):
        if self.local_rank != 0:
            return
L
LielinJiang 已提交
335

L
LielinJiang 已提交
336 337 338
        assert name in ['checkpoint', 'weight']

        state_dicts = {}
L
LielinJiang 已提交
339 340 341 342 343 344
        if self.by_epoch:
            save_filename = 'epoch_%s_%s.pdparams' % (
                epoch // self.iters_per_epoch, name)
        else:
            save_filename = 'iter_%s_%s.pdparams' % (epoch, name)

L
lijianshe02 已提交
345
        os.makedirs(self.output_dir, exist_ok=True)
L
LielinJiang 已提交
346
        save_path = os.path.join(self.output_dir, save_filename)
L
LielinJiang 已提交
347 348
        for net_name, net in self.model.nets.items():
            state_dicts[net_name] = net.state_dict()
L
LielinJiang 已提交
349 350 351 352 353 354 355

        if name == 'weight':
            save(state_dicts, save_path)
            return

        state_dicts['epoch'] = epoch

L
LielinJiang 已提交
356 357
        for opt_name, opt in self.model.optimizers.items():
            state_dicts[opt_name] = opt.state_dict()
L
LielinJiang 已提交
358 359 360 361 362

        save(state_dicts, save_path)

        if keep > 0:
            try:
L
LielinJiang 已提交
363 364 365 366 367 368 369 370 371 372
                if self.by_epoch:
                    checkpoint_name_to_be_removed = os.path.join(
                        self.output_dir, 'epoch_%s_%s.pdparams' %
                        ((epoch - keep * self.weight_interval) //
                         self.iters_per_epoch, name))
                else:
                    checkpoint_name_to_be_removed = os.path.join(
                        self.output_dir, 'iter_%s_%s.pdparams' %
                        (epoch - keep * self.weight_interval, name))

L
LielinJiang 已提交
373 374 375 376 377 378 379 380 381 382
                if os.path.exists(checkpoint_name_to_be_removed):
                    os.remove(checkpoint_name_to_be_removed)

            except Exception as e:
                self.logger.info('remove old checkpoints error: {}'.format(e))

    def resume(self, checkpoint_path):
        state_dicts = load(checkpoint_path)
        if state_dicts.get('epoch', None) is not None:
            self.start_epoch = state_dicts['epoch'] + 1
L
LielinJiang 已提交
383
            self.global_steps = self.iters_per_epoch * state_dicts['epoch']
L
LielinJiang 已提交
384

L
lijianshe02 已提交
385 386
            self.current_iter = state_dicts['epoch'] + 1

L
LielinJiang 已提交
387
        for net_name, net in self.model.nets.items():
388
            net.set_state_dict(state_dicts[net_name])
L
LielinJiang 已提交
389

L
LielinJiang 已提交
390
        for opt_name, opt in self.model.optimizers.items():
391
            opt.set_state_dict(state_dicts[opt_name])
L
LielinJiang 已提交
392 393 394

    def load(self, weight_path):
        state_dicts = load(weight_path)
L
LielinJiang 已提交
395

L
LielinJiang 已提交
396
        for net_name, net in self.model.nets.items():
397 398 399 400 401 402 403 404
            if net_name in state_dicts:
                net.set_state_dict(state_dicts[net_name])
                self.logger.info(
                    'Loaded pretrained weight for net {}'.format(net_name))
            else:
                self.logger.warning(
                    'Can not find state dict of net {}. Skip load pretrained weight for net {}'
                    .format(net_name, net_name))
郑启航 已提交
405 406 407 408 409 410 411 412

    def close(self):
        """
        when finish the training need close file handler or other.

        """
        if self.enable_visualdl:
            self.vdl_logger.close()