prune.py 14.9 KB
Newer Older
W
whs 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
# Copyright (c) 2019 PaddlePaddle Authors. All Rights Reserved.
#
# 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.

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

Q
qingqing01 已提交
19 20 21 22 23 24 25
import os, sys

# add python path of PadleDetection to sys.path
parent_path = os.path.abspath(os.path.join(__file__, *(['..'] * 3)))
if parent_path not in sys.path:
    sys.path.append(parent_path)

W
whs 已提交
26 27 28 29 30 31 32
import time
import numpy as np
import datetime
from collections import deque
from paddleslim.prune import Pruner
from paddleslim.analysis import flops
from paddle import fluid
Q
qingqing01 已提交
33

W
whs 已提交
34 35 36 37 38 39 40
from ppdet.experimental import mixed_precision_context
from ppdet.core.workspace import load_config, merge_config, create
from ppdet.data.reader import create_reader
from ppdet.utils import dist_utils
from ppdet.utils.eval_utils import parse_fetches, eval_run, eval_results
from ppdet.utils.stats import TrainingStats
from ppdet.utils.cli import ArgsParser
41
from ppdet.utils.check import check_gpu, check_version, check_config
W
whs 已提交
42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61
import ppdet.utils.checkpoint as checkpoint

import logging
FORMAT = '%(asctime)s-%(levelname)s: %(message)s'
logging.basicConfig(level=logging.INFO, format=FORMAT)
logger = logging.getLogger(__name__)


def main():
    env = os.environ
    FLAGS.dist = 'PADDLE_TRAINER_ID' in env and 'PADDLE_TRAINERS_NUM' in env
    if FLAGS.dist:
        trainer_id = int(env['PADDLE_TRAINER_ID'])
        import random
        local_seed = (99 + trainer_id)
        random.seed(local_seed)
        np.random.seed(local_seed)

    cfg = load_config(FLAGS.config)
    merge_config(FLAGS.opt)
62
    check_config(cfg)
W
whs 已提交
63 64 65 66
    # check if set use_gpu=True in paddlepaddle cpu version
    check_gpu(cfg.use_gpu)
    # check if paddlepaddle version is satisfied
    check_version()
67 68

    main_arch = cfg.architecture
W
whs 已提交
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

    if cfg.use_gpu:
        devices_num = fluid.core.get_cuda_device_count()
    else:
        devices_num = int(os.environ.get('CPU_NUM', 1))

    if 'FLAGS_selected_gpus' in env:
        device_id = int(env['FLAGS_selected_gpus'])
    else:
        device_id = 0
    place = fluid.CUDAPlace(device_id) if cfg.use_gpu else fluid.CPUPlace()
    exe = fluid.Executor(place)

    lr_builder = create('LearningRate')
    optim_builder = create('OptimizerBuilder')

    # build program
    startup_prog = fluid.Program()
    train_prog = fluid.Program()
    with fluid.program_guard(train_prog, startup_prog):
        with fluid.unique_name.guard():
            model = create(main_arch)
            if FLAGS.fp16:
                assert (getattr(model.backbone, 'norm_type', None)
                        != 'affine_channel'), \
                    '--fp16 currently does not support affine channel, ' \
                    ' please modify backbone settings to use batch norm'

            with mixed_precision_context(FLAGS.loss_scale, FLAGS.fp16) as ctx:
                inputs_def = cfg['TrainReader']['inputs_def']
                feed_vars, train_loader = model.build_inputs(**inputs_def)
                train_fetches = model.train(feed_vars)
                loss = train_fetches['loss']
                if FLAGS.fp16:
                    loss *= ctx.get_loss_scale_var()
                lr = lr_builder()
                optimizer = optim_builder(lr)
                optimizer.minimize(loss)
                if FLAGS.fp16:
                    loss /= ctx.get_loss_scale_var()

    # parse train fetches
    train_keys, train_values, _ = parse_fetches(train_fetches)
    train_values.append(lr)

    if FLAGS.print_params:
115 116
        param_delimit_str = '-' * 20 + "All parameters in current graph" + '-' * 20
        print(param_delimit_str)
W
whs 已提交
117 118
        for block in train_prog.blocks:
            for param in block.all_parameters():
119 120 121
                print("parameter name: {}\tshape: {}".format(param.name,
                                                             param.shape))
        print('-' * len(param_delimit_str))
W
whs 已提交
122 123 124 125 126 127 128 129 130 131 132 133 134
        return

    if FLAGS.eval:
        eval_prog = fluid.Program()
        with fluid.program_guard(eval_prog, startup_prog):
            with fluid.unique_name.guard():
                model = create(main_arch)
                inputs_def = cfg['EvalReader']['inputs_def']
                feed_vars, eval_loader = model.build_inputs(**inputs_def)
                fetches = model.eval(feed_vars)
        eval_prog = eval_prog.clone(True)

        eval_reader = create_reader(cfg.EvalReader)
135 136
        # When iterable mode, set set_sample_list_generator(eval_reader, place)
        eval_loader.set_sample_list_generator(eval_reader)
W
whs 已提交
137 138 139 140 141 142

        # parse eval fetches
        extra_keys = []
        if cfg.metric == 'COCO':
            extra_keys = ['im_info', 'im_id', 'im_shape']
        if cfg.metric == 'VOC':
K
Kaipeng Deng 已提交
143
            extra_keys = ['gt_bbox', 'gt_class', 'is_difficult']
W
whs 已提交
144
        if cfg.metric == 'WIDERFACE':
K
Kaipeng Deng 已提交
145
            extra_keys = ['im_id', 'im_shape', 'gt_bbox']
W
whs 已提交
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
        eval_keys, eval_values, eval_cls = parse_fetches(fetches, eval_prog,
                                                         extra_keys)

    # compile program for multi-devices
    build_strategy = fluid.BuildStrategy()
    build_strategy.fuse_all_optimizer_ops = False
    build_strategy.fuse_elewise_add_act_ops = True
    # only enable sync_bn in multi GPU devices
    sync_bn = getattr(model.backbone, 'norm_type', None) == 'sync_bn'
    build_strategy.sync_batch_norm = sync_bn and devices_num > 1 \
        and cfg.use_gpu

    exec_strategy = fluid.ExecutionStrategy()
    # iteration number when CompiledProgram tries to drop local execution scopes.
    # Set it to be 1 to save memory usages, so that unused variables in
    # local execution scopes can be deleted after each iteration.
    exec_strategy.num_iteration_per_drop_scope = 1
    if FLAGS.dist:
        dist_utils.prepare_for_multi_process(exe, build_strategy, startup_prog,
                                             train_prog)
        exec_strategy.num_threads = 1

    exe.run(startup_prog)

    fuse_bn = getattr(model.backbone, 'norm_type', None) == 'affine_channel'

    start_iter = 0
K
Kaipeng Deng 已提交
173
    if cfg.pretrain_weights:
174
        checkpoint.load_params(exe, train_prog, cfg.pretrain_weights)
W
whs 已提交
175 176

    pruned_params = FLAGS.pruned_params
177 178
    assert FLAGS.pruned_params is not None, \
        "FLAGS.pruned_params is empty!!! Please set it by '--pruned_params' option."
W
whs 已提交
179 180
    pruned_params = FLAGS.pruned_params.strip().split(",")
    logger.info("pruned params: {}".format(pruned_params))
181
    pruned_ratios = [float(n) for n in FLAGS.pruned_ratios.strip().split(",")]
W
whs 已提交
182
    logger.info("pruned ratios: {}".format(pruned_ratios))
183 184 185 186 187
    assert len(pruned_params) == len(pruned_ratios), \
        "The length of pruned params and pruned ratios should be equal."
    assert (pruned_ratios > [0] * len(pruned_ratios) and
            pruned_ratios < [1] * len(pruned_ratios)
            ), "The elements of pruned ratios should be in range (0, 1)."
W
whs 已提交
188

189 190 191
    assert FLAGS.prune_criterion in ['l1_norm', 'geometry_median'], \
            "unsupported prune criterion {}".format(FLAGS.prune_criterion)
    pruner = Pruner(criterion=FLAGS.prune_criterion)
W
whs 已提交
192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215
    train_prog = pruner.prune(
        train_prog,
        fluid.global_scope(),
        params=pruned_params,
        ratios=pruned_ratios,
        place=place,
        only_graph=False)[0]

    compiled_train_prog = fluid.CompiledProgram(train_prog).with_data_parallel(
        loss_name=loss.name,
        build_strategy=build_strategy,
        exec_strategy=exec_strategy)

    if FLAGS.eval:

        base_flops = flops(eval_prog)
        eval_prog = pruner.prune(
            eval_prog,
            fluid.global_scope(),
            params=pruned_params,
            ratios=pruned_ratios,
            place=place,
            only_graph=True)[0]
        pruned_flops = flops(eval_prog)
216 217 218
        logger.info("FLOPs -{}; total FLOPs: {}; pruned FLOPs: {}".format(
            float(base_flops - pruned_flops) / base_flops, base_flops,
            pruned_flops))
219
        compiled_eval_prog = fluid.CompiledProgram(eval_prog)
W
whs 已提交
220

K
Kaipeng Deng 已提交
221 222 223 224
    if FLAGS.resume_checkpoint:
        checkpoint.load_checkpoint(exe, train_prog, FLAGS.resume_checkpoint)
        start_iter = checkpoint.global_step()

W
whs 已提交
225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247
    train_reader = create_reader(cfg.TrainReader, (cfg.max_iters - start_iter) *
                                 devices_num, cfg)
    train_loader.set_sample_list_generator(train_reader, place)

    # whether output bbox is normalized in model output layer
    is_bbox_normalized = False
    if hasattr(model, 'is_bbox_normalized') and \
            callable(model.is_bbox_normalized):
        is_bbox_normalized = model.is_bbox_normalized()

    # if map_type not set, use default 11point, only use in VOC eval
    map_type = cfg.map_type if 'map_type' in cfg else '11point'

    train_stats = TrainingStats(cfg.log_smooth_window, train_keys)
    train_loader.start()
    start_time = time.time()
    end_time = time.time()

    cfg_name = os.path.basename(FLAGS.config).split('.')[0]
    save_dir = os.path.join(cfg.save_dir, cfg_name)
    time_stat = deque(maxlen=cfg.log_smooth_window)
    best_box_ap_list = [0.0, 0]  #[map, iter]

走神的阿圆's avatar
走神的阿圆 已提交
248 249 250 251 252 253
    # use VisualDL to log data
    if FLAGS.use_vdl:
        from visualdl import LogWriter
        vdl_writer = LogWriter(FLAGS.vdl_log_dir)
        vdl_loss_step = 0
        vdl_mAP_step = 0
W
whs 已提交
254 255 256

    if FLAGS.eval:
        resolution = None
257
        if 'Mask' in cfg.architecture:
W
whs 已提交
258
            resolution = model.mask_head.resolution
259 260 261 262 263 264 265 266 267 268
        # evaluation
        results = eval_run(
            exe,
            compiled_eval_prog,
            eval_loader,
            eval_keys,
            eval_values,
            eval_cls,
            cfg,
            resolution=resolution)
W
whs 已提交
269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
        dataset = cfg['EvalReader']['dataset']
        box_ap_stats = eval_results(
            results,
            cfg.metric,
            cfg.num_classes,
            resolution,
            is_bbox_normalized,
            FLAGS.output_eval,
            map_type,
            dataset=dataset)

    for it in range(start_iter, cfg.max_iters):
        start_time = end_time
        end_time = time.time()
        time_stat.append(end_time - start_time)
        time_cost = np.mean(time_stat)
        eta_sec = (cfg.max_iters - it) * time_cost
        eta = str(datetime.timedelta(seconds=int(eta_sec)))
        outs = exe.run(compiled_train_prog, fetch_list=train_values)
        stats = {k: np.array(v).mean() for k, v in zip(train_keys, outs[:-1])}

走神的阿圆's avatar
走神的阿圆 已提交
290 291
        # use VisualDL to log loss
        if FLAGS.use_vdl:
W
whs 已提交
292 293
            if it % cfg.log_iter == 0:
                for loss_name, loss_value in stats.items():
走神的阿圆's avatar
走神的阿圆 已提交
294 295
                    vdl_writer.add_scalar(loss_name, loss_value, vdl_loss_step)
                vdl_loss_step += 1
W
whs 已提交
296 297 298 299 300 301 302 303 304 305 306 307 308 309 310

        train_stats.update(stats)
        logs = train_stats.log()
        if it % cfg.log_iter == 0 and (not FLAGS.dist or trainer_id == 0):
            strs = 'iter: {}, lr: {:.6f}, {}, time: {:.3f}, eta: {}'.format(
                it, np.mean(outs[-1]), logs, time_cost, eta)
            logger.info(strs)

        if (it > 0 and it % cfg.snapshot_iter == 0 or it == cfg.max_iters - 1) \
           and (not FLAGS.dist or trainer_id == 0):
            save_name = str(it) if it != cfg.max_iters - 1 else "model_final"
            checkpoint.save(exe, train_prog, os.path.join(save_dir, save_name))

            if FLAGS.eval:
                # evaluation
W
whs 已提交
311 312 313
                resolution = None
                if 'Mask' in cfg.architecture:
                    resolution = model.mask_head.resolution
314 315 316 317 318 319 320
                results = eval_run(
                    exe,
                    compiled_eval_prog,
                    eval_loader,
                    eval_keys,
                    eval_values,
                    eval_cls,
W
whs 已提交
321 322
                    cfg=cfg,
                    resolution=resolution)
W
whs 已提交
323
                box_ap_stats = eval_results(
K
Kaipeng Deng 已提交
324 325 326 327 328 329 330 331
                    results,
                    cfg.metric,
                    cfg.num_classes,
                    resolution,
                    is_bbox_normalized,
                    FLAGS.output_eval,
                    map_type,
                    dataset=dataset)
W
whs 已提交
332

走神的阿圆's avatar
走神的阿圆 已提交
333 334 335 336
                # use VisualDL to log mAP
                if FLAGS.use_vdl:
                    vdl_writer.add_scalar("mAP", box_ap_stats[0], vdl_mAP_step)
                    vdl_mAP_step += 1
W
whs 已提交
337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377

                if box_ap_stats[0] > best_box_ap_list[0]:
                    best_box_ap_list[0] = box_ap_stats[0]
                    best_box_ap_list[1] = it
                    checkpoint.save(exe, train_prog,
                                    os.path.join(save_dir, "best_model"))
                logger.info("Best test box ap: {}, in iter: {}".format(
                    best_box_ap_list[0], best_box_ap_list[1]))

    train_loader.reset()


if __name__ == '__main__':
    parser = ArgsParser()
    parser.add_argument(
        "-r",
        "--resume_checkpoint",
        default=None,
        type=str,
        help="Checkpoint path for resuming training.")
    parser.add_argument(
        "--fp16",
        action='store_true',
        default=False,
        help="Enable mixed precision training.")
    parser.add_argument(
        "--loss_scale",
        default=8.,
        type=float,
        help="Mixed precision training loss scale.")
    parser.add_argument(
        "--eval",
        action='store_true',
        default=False,
        help="Whether to perform evaluation in train")
    parser.add_argument(
        "--output_eval",
        default=None,
        type=str,
        help="Evaluation directory, default is current directory.")
    parser.add_argument(
走神的阿圆's avatar
走神的阿圆 已提交
378
        "--use_vdl",
W
whs 已提交
379 380
        type=bool,
        default=False,
走神的阿圆's avatar
走神的阿圆 已提交
381
        help="whether to record the data to VisualDL.")
W
whs 已提交
382
    parser.add_argument(
走神的阿圆's avatar
走神的阿圆 已提交
383
        '--vdl_log_dir',
W
whs 已提交
384
        type=str,
走神的阿圆's avatar
走神的阿圆 已提交
385 386
        default="vdl_log_dir/scalar",
        help='VisualDL logging directory for scalar.')
W
whs 已提交
387 388 389 390 391 392 393 394 395

    parser.add_argument(
        "-p",
        "--pruned_params",
        default=None,
        type=str,
        help="The parameters to be pruned when calculating sensitivities.")
    parser.add_argument(
        "--pruned_ratios",
396
        default=None,
W
whs 已提交
397
        type=str,
398 399
        help="The ratios pruned iteratively for each parameter when calculating sensitivities."
    )
W
whs 已提交
400 401 402 403 404 405
    parser.add_argument(
        "-P",
        "--print_params",
        default=False,
        action='store_true',
        help="Whether to only print the parameters' names and shapes.")
406 407 408 409 410 411
    parser.add_argument(
        "--prune_criterion",
        default='l1_norm',
        type=str,
        help="criterion function type for channels sorting in pruning, can be set " \
             "as 'l1_norm' or 'geometry_median' currently, default 'l1_norm'")
W
whs 已提交
412 413
    FLAGS = parser.parse_args()
    main()