engine.py 21.1 KB
Newer Older
D
dongshuilong 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
# Copyright (c) 2021 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

D
dongshuilong 已提交
17
import os
D
dongshuilong 已提交
18 19 20
import paddle
import paddle.distributed as dist
from visualdl import LogWriter
D
dongshuilong 已提交
21
from paddle import nn
D
dongshuilong 已提交
22 23
import numpy as np
import random
D
dongshuilong 已提交
24 25 26 27 28 29

from ppcls.utils.misc import AverageMeter
from ppcls.utils import logger
from ppcls.utils.logger import init_logger
from ppcls.utils.config import print_config
from ppcls.data import build_dataloader
W
dbg  
weishengyu 已提交
30
from ppcls.arch import build_model, RecModel, DistillationModel, TheseusLayer
D
dongshuilong 已提交
31 32 33
from ppcls.loss import build_loss
from ppcls.metric import build_metrics
from ppcls.optimizer import build_optimizer
F
flytocc 已提交
34
from ppcls.utils.ema import ExponentialMovingAverage
D
dongshuilong 已提交
35 36 37 38 39 40 41
from ppcls.utils.save_load import load_dygraph_pretrain, load_dygraph_pretrain_from_url
from ppcls.utils.save_load import init_model
from ppcls.utils import save_load

from ppcls.data.utils.get_image_list import get_image_list
from ppcls.data.postprocess import build_postprocess
from ppcls.data import create_operators
42 43
from .train import build_train_epoch_func
from .evaluation import build_eval_func
44
from ppcls.engine.train.utils import type_name
D
dongshuilong 已提交
45
from ppcls.engine import evaluation
D
dongshuilong 已提交
46 47 48
from ppcls.arch.gears.identity_head import IdentityHead


D
dongshuilong 已提交
49
class Engine(object):
D
dongshuilong 已提交
50
    def __init__(self, config, mode="train"):
D
dongshuilong 已提交
51
        assert mode in ["train", "eval", "infer", "export"]
D
dongshuilong 已提交
52 53
        self.mode = mode
        self.config = config
D
dongshuilong 已提交
54

D
dongshuilong 已提交
55
        # set seed
G
gaotingquan 已提交
56
        self._init_seed()
D
dongshuilong 已提交
57

D
dongshuilong 已提交
58
        # init logger
G
gaotingquan 已提交
59
        init_logger(self.config, mode=mode)
D
dongshuilong 已提交
60

G
gaotingquan 已提交
61 62 63
        # for visualdl
        self.vdl_writer = self._init_vdl()

D
dongshuilong 已提交
64
        # init train_func and eval_func
65
        self.train_epoch_func = build_train_epoch_func(self.config)
66
        self.eval_func = build_eval_func(self.config)
D
dongshuilong 已提交
67 68

        # set device
69
        self._init_device()
D
dongshuilong 已提交
70

71 72 73
        # gradient accumulation
        self.update_freq = self.config["Global"].get("update_freq", 1)

D
dongshuilong 已提交
74
        # build dataloader
G
gaotingquan 已提交
75
        self.dataloader_dict = build_dataloader(self)
D
dongshuilong 已提交
76 77

        # build loss
G
gaotingquan 已提交
78 79
        self.train_loss_func, self.unlabel_train_loss_func, self.eval_loss_func = build_loss(
            self.config, self.mode)
D
dongshuilong 已提交
80 81

        # build metric
G
gaotingquan 已提交
82
        self.train_metric_func, self.eval_metric_func = build_metrics(self)
D
dongshuilong 已提交
83 84

        # build model
littletomatodonkey's avatar
littletomatodonkey 已提交
85
        self.model = build_model(self.config, self.mode)
D
dongshuilong 已提交
86

D
dongshuilong 已提交
87
        # load_pretrain
G
gaotingquan 已提交
88
        self._init_pretrained()
D
dongshuilong 已提交
89 90

        # build optimizer
91
        self.optimizer, self.lr_sch = build_optimizer(self)
92

93
        # AMP training and evaluating
G
gaotingquan 已提交
94
        self._init_amp()
95 96

        # for distributed
G
gaotingquan 已提交
97
        self._init_dist()
D
dongshuilong 已提交
98

99 100
        print_config(config)

D
dongshuilong 已提交
101 102 103 104 105
    def train(self):
        assert self.mode == "train"
        print_batch_step = self.config['Global']['print_batch_step']
        save_interval = self.config["Global"]["save_interval"]
        best_metric = {
C
cuicheng01 已提交
106
            "metric": -1.0,
D
dongshuilong 已提交
107 108
            "epoch": 0,
        }
G
gaotingquan 已提交
109 110 111

        # build EMA model
        self.ema = "EMA" in self.config and self.mode == "train"
F
flytocc 已提交
112
        if self.ema:
G
gaotingquan 已提交
113 114
            self.model_ema = ExponentialMovingAverage(
                self.model, self.config['EMA'].get("decay", 0.9999))
F
flytocc 已提交
115 116
            best_metric_ema = 0.0
            ema_module = self.model_ema.module
G
gaotingquan 已提交
117 118 119
        else:
            ema_module = None

D
dongshuilong 已提交
120 121 122 123 124 125 126 127 128 129 130 131
        # key:
        # val: metrics list word
        self.output_info = dict()
        self.time_info = {
            "batch_cost": AverageMeter(
                "batch_cost", '.5f', postfix=" s,"),
            "reader_cost": AverageMeter(
                "reader_cost", ".5f", postfix=" s,"),
        }
        # global iter counter
        self.global_step = 0

132 133
        if self.config.Global.checkpoints is not None:
            metric_info = init_model(self.config.Global, self.model,
F
flytocc 已提交
134 135
                                     self.optimizer, self.train_loss_func,
                                     ema_module)
D
dongshuilong 已提交
136 137 138 139 140 141 142
            if metric_info is not None:
                best_metric.update(metric_info)

        for epoch_id in range(best_metric["epoch"] + 1,
                              self.config["Global"]["epochs"] + 1):
            acc = 0.0
            # for one epoch train
D
dongshuilong 已提交
143
            self.train_epoch_func(self, epoch_id, print_batch_step)
D
dongshuilong 已提交
144

littletomatodonkey's avatar
littletomatodonkey 已提交
145 146
            metric_msg = ", ".join(
                [self.output_info[key].avg_info for key in self.output_info])
D
dongshuilong 已提交
147 148 149 150 151
            logger.info("[Train][Epoch {}/{}][Avg]{}".format(
                epoch_id, self.config["Global"]["epochs"], metric_msg))
            self.output_info.clear()

            # eval model and save model if possible
littletomatodonkey's avatar
littletomatodonkey 已提交
152 153
            start_eval_epoch = self.config["Global"].get("start_eval_epoch",
                                                         0) - 1
D
dongshuilong 已提交
154 155
            if self.config["Global"][
                    "eval_during_train"] and epoch_id % self.config["Global"][
C
cuicheng01 已提交
156
                        "eval_interval"] == 0 and epoch_id > start_eval_epoch:
D
dongshuilong 已提交
157
                acc = self.eval(epoch_id)
H
add xbm  
HydrogenSulfate 已提交
158 159 160 161

                # step lr (by epoch) according to given metric, such as acc
                for i in range(len(self.lr_sch)):
                    if getattr(self.lr_sch[i], "by_epoch", False) and \
162
                            type_name(self.lr_sch[i]) == "ReduceOnPlateau":
H
add xbm  
HydrogenSulfate 已提交
163 164
                        self.lr_sch[i].step(acc)

D
dongshuilong 已提交
165 166 167 168 169 170 171 172
                if acc > best_metric["metric"]:
                    best_metric["metric"] = acc
                    best_metric["epoch"] = epoch_id
                    save_load.save_model(
                        self.model,
                        self.optimizer,
                        best_metric,
                        self.output_dir,
F
flytocc 已提交
173
                        ema=ema_module,
D
dongshuilong 已提交
174
                        model_name=self.config["Arch"]["name"],
175
                        prefix="best_model",
littletomatodonkey's avatar
littletomatodonkey 已提交
176 177
                        loss=self.train_loss_func,
                        save_student_model=True)
D
dongshuilong 已提交
178 179 180 181 182 183 184 185 186 187
                logger.info("[Eval][Epoch {}][best metric: {}]".format(
                    epoch_id, best_metric["metric"]))
                logger.scaler(
                    name="eval_acc",
                    value=acc,
                    step=epoch_id,
                    writer=self.vdl_writer)

                self.model.train()

F
flytocc 已提交
188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213
                if self.ema:
                    ori_model, self.model = self.model, ema_module
                    acc_ema = self.eval(epoch_id)
                    self.model = ori_model
                    ema_module.eval()

                    if acc_ema > best_metric_ema:
                        best_metric_ema = acc_ema
                        save_load.save_model(
                            self.model,
                            self.optimizer,
                            {"metric": acc_ema,
                             "epoch": epoch_id},
                            self.output_dir,
                            ema=ema_module,
                            model_name=self.config["Arch"]["name"],
                            prefix="best_model_ema",
                            loss=self.train_loss_func)
                    logger.info("[Eval][Epoch {}][best metric ema: {}]".format(
                        epoch_id, best_metric_ema))
                    logger.scaler(
                        name="eval_acc_ema",
                        value=acc_ema,
                        step=epoch_id,
                        writer=self.vdl_writer)

D
dongshuilong 已提交
214
            # save model
D
dongshuilong 已提交
215
            if save_interval > 0 and epoch_id % save_interval == 0:
D
dongshuilong 已提交
216 217 218 219 220
                save_load.save_model(
                    self.model,
                    self.optimizer, {"metric": acc,
                                     "epoch": epoch_id},
                    self.output_dir,
F
flytocc 已提交
221
                    ema=ema_module,
D
dongshuilong 已提交
222
                    model_name=self.config["Arch"]["name"],
223 224
                    prefix="epoch_{}".format(epoch_id),
                    loss=self.train_loss_func)
G
gaotingquan 已提交
225 226 227 228 229 230
            # save the latest model
            save_load.save_model(
                self.model,
                self.optimizer, {"metric": acc,
                                 "epoch": epoch_id},
                self.output_dir,
F
flytocc 已提交
231
                ema=ema_module,
G
gaotingquan 已提交
232
                model_name=self.config["Arch"]["name"],
233 234
                prefix="latest",
                loss=self.train_loss_func)
D
dongshuilong 已提交
235 236 237 238 239 240 241 242

        if self.vdl_writer is not None:
            self.vdl_writer.close()

    @paddle.no_grad()
    def eval(self, epoch_id=0):
        assert self.mode in ["train", "eval"]
        self.model.eval()
D
dongshuilong 已提交
243
        eval_result = self.eval_func(self, epoch_id)
D
dongshuilong 已提交
244 245 246 247 248 249
        self.model.train()
        return eval_result

    @paddle.no_grad()
    def infer(self):
        assert self.mode == "infer" and self.eval_mode == "classification"
G
gaotingquan 已提交
250 251 252 253 254 255

        self.preprocess_func = create_operators(self.config["Infer"][
            "transforms"])
        self.postprocess_func = build_postprocess(self.config["Infer"][
            "PostProcess"])

256 257
        total_trainer = dist.get_world_size()
        local_rank = dist.get_rank()
D
dongshuilong 已提交
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274
        image_list = get_image_list(self.config["Infer"]["infer_imgs"])
        # data split
        image_list = image_list[local_rank::total_trainer]

        batch_size = self.config["Infer"]["batch_size"]
        self.model.eval()
        batch_data = []
        image_file_list = []
        for idx, image_file in enumerate(image_list):
            with open(image_file, 'rb') as f:
                x = f.read()
            for process in self.preprocess_func:
                x = process(x)
            batch_data.append(x)
            image_file_list.append(image_file)
            if len(batch_data) >= batch_size or idx == len(image_list) - 1:
                batch_tensor = paddle.to_tensor(batch_data)
G
gaotingquan 已提交
275 276 277 278 279 280 281 282 283 284 285

                if self.amp and self.amp_eval:
                    with paddle.amp.auto_cast(
                            custom_black_list={
                                "flatten_contiguous_range", "greater_than"
                            },
                            level=self.amp_level):
                        out = self.model(batch_tensor)
                else:
                    out = self.model(batch_tensor)

D
dongshuilong 已提交
286 287
                if isinstance(out, list):
                    out = out[0]
littletomatodonkey's avatar
littletomatodonkey 已提交
288 289
                if isinstance(out, dict) and "Student" in out:
                    out = out["Student"]
290 291 292
                if isinstance(out, dict) and "logits" in out:
                    out = out["logits"]
                if isinstance(out, dict) and "output" in out:
W
dbg  
weishengyu 已提交
293
                    out = out["output"]
D
dongshuilong 已提交
294 295 296 297 298 299 300
                result = self.postprocess_func(out, image_file_list)
                print(result)
                batch_data.clear()
                image_file_list.clear()

    def export(self):
        assert self.mode == "export"
Z
zhiboniu 已提交
301 302
        use_multilabel = self.config["Global"].get(
            "use_multilabel",
C
cuicheng01 已提交
303
            False) or "ATTRMetric" in self.config["Metric"]["Eval"][0]
C
cuicheng01 已提交
304
        model = ExportModel(self.config["Arch"], self.model, use_multilabel)
305 306 307 308 309 310 311 312 313
        if self.config["Global"]["pretrained_model"] is not None:
            if self.config["Global"]["pretrained_model"].startswith("http"):
                load_dygraph_pretrain_from_url(
                    model.base_model,
                    self.config["Global"]["pretrained_model"])
            else:
                load_dygraph_pretrain(
                    model.base_model,
                    self.config["Global"]["pretrained_model"])
D
dongshuilong 已提交
314 315

        model.eval()
G
gaotingquan 已提交
316

317
        # for re-parameterization nets
H
HydrogenSulfate 已提交
318
        for layer in self.model.sublayers():
319 320 321
            if hasattr(layer, "re_parameterize") and not getattr(layer,
                                                                 "is_repped"):
                layer.re_parameterize()
G
gaotingquan 已提交
322

D
dongshuilong 已提交
323 324
        save_path = os.path.join(self.config["Global"]["save_inference_dir"],
                                 "inference")
littletomatodonkey's avatar
littletomatodonkey 已提交
325 326 327 328 329 330 331 332 333 334 335 336

        model = paddle.jit.to_static(
            model,
            input_spec=[
                paddle.static.InputSpec(
                    shape=[None] + self.config["Global"]["image_shape"],
                    dtype='float32')
            ])
        if hasattr(model.base_model,
                   "quanter") and model.base_model.quanter is not None:
            model.base_model.quanter.save_quantized_model(model,
                                                          save_path + "_int8")
D
dongshuilong 已提交
337 338
        else:
            paddle.jit.save(model, save_path)
G
gaotingquan 已提交
339 340 341
        logger.info(
            f"Export succeeded! The inference model exported has been saved in \"{self.config['Global']['save_inference_dir']}\"."
        )
D
dongshuilong 已提交
342

G
gaotingquan 已提交
343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379
    def _init_vdl(self):
        if self.config['Global'][
                'use_visualdl'] and mode == "train" and dist.get_rank() == 0:
            vdl_writer_path = os.path.join(self.output_dir, "vdl")
            if not os.path.exists(vdl_writer_path):
                os.makedirs(vdl_writer_path)
            return LogWriter(logdir=vdl_writer_path)
        return None

    def _init_seed(self):
        seed = self.config["Global"].get("seed", False)
        if dist.get_world_size() != 1:
            # if self.config["Global"]["distributed"]:
            # set different seed in different GPU manually in distributed environment
            if not seed:
                logger.warning(
                    "The random seed cannot be None in a distributed environment. Global.seed has been set to 42 by default"
                )
                self.config["Global"]["seed"] = seed = 42
            logger.info(
                f"Set random seed to ({int(seed)} + $PADDLE_TRAINER_ID) for different trainer"
            )
            dist_seed = int(seed) + dist.get_rank()
            paddle.seed(dist_seed)
            np.random.seed(dist_seed)
            random.seed(dist_seed)
        elif seed or seed == 0:
            assert isinstance(seed, int), "The 'seed' must be a integer!"
            paddle.seed(seed)
            np.random.seed(seed)
            random.seed(seed)

    def _init_device(self):
        device = self.config["Global"]["device"]
        assert device in ["cpu", "gpu", "xpu", "npu", "mlu", "ascend"]
        logger.info('train with paddle {} and device {}'.format(
            paddle.__version__, device))
380
        paddle.set_device(device)
G
gaotingquan 已提交
381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484

    def _init_pretrained(self):
        if self.config["Global"]["pretrained_model"] is not None:
            if self.config["Global"]["pretrained_model"].startswith("http"):
                load_dygraph_pretrain_from_url(
                    [self.model, getattr(self, 'train_loss_func', None)],
                    self.config["Global"]["pretrained_model"])
            else:
                load_dygraph_pretrain(
                    [self.model, getattr(self, 'train_loss_func', None)],
                    self.config["Global"]["pretrained_model"])

    def _init_amp(self):
        self.amp = "AMP" in self.config and self.config["AMP"] is not None
        self.amp_eval = False
        # for amp
        if self.amp:
            AMP_RELATED_FLAGS_SETTING = {'FLAGS_max_inplace_grad_add': 8, }
            if paddle.is_compiled_with_cuda():
                AMP_RELATED_FLAGS_SETTING.update({
                    'FLAGS_cudnn_batchnorm_spatial_persistent': 1
                })
            paddle.set_flags(AMP_RELATED_FLAGS_SETTING)

            self.scale_loss = self.config["AMP"].get("scale_loss", 1.0)
            self.use_dynamic_loss_scaling = self.config["AMP"].get(
                "use_dynamic_loss_scaling", False)
            self.scaler = paddle.amp.GradScaler(
                init_loss_scaling=self.scale_loss,
                use_dynamic_loss_scaling=self.use_dynamic_loss_scaling)

            self.amp_level = self.config['AMP'].get("level", "O1")
            if self.amp_level not in ["O1", "O2"]:
                msg = "[Parameter Error]: The optimize level of AMP only support 'O1' and 'O2'. The level has been set 'O1'."
                logger.warning(msg)
                self.config['AMP']["level"] = "O1"
                self.amp_level = "O1"

            self.amp_eval = self.config["AMP"].get("use_fp16_test", False)
            # TODO(gaotingquan): Paddle not yet support FP32 evaluation when training with AMPO2
            if self.mode == "train" and self.config["Global"].get(
                    "eval_during_train",
                    True) and self.amp_level == "O2" and self.amp_eval == False:
                msg = "PaddlePaddle only support FP16 evaluation when training with AMP O2 now. "
                logger.warning(msg)
                self.config["AMP"]["use_fp16_test"] = True
                self.amp_eval = True

            # TODO(gaotingquan): to compatible with different versions of Paddle
            paddle_version = paddle.__version__[:3]
            # paddle version < 2.3.0 and not develop
            if paddle_version not in ["2.3", "0.0"]:
                if self.mode == "train":
                    self.model, self.optimizer = paddle.amp.decorate(
                        models=self.model,
                        optimizers=self.optimizer,
                        level=self.amp_level,
                        save_dtype='float32')
                elif self.amp_eval:
                    if self.amp_level == "O2":
                        msg = "The PaddlePaddle that installed not support FP16 evaluation in AMP O2. Please use PaddlePaddle version >= 2.3.0. Use FP32 evaluation instead and please notice the Eval Dataset output_fp16 should be 'False'."
                        logger.warning(msg)
                        self.amp_eval = False
                    else:
                        self.model, self.optimizer = paddle.amp.decorate(
                            models=self.model,
                            level=self.amp_level,
                            save_dtype='float32')
            # paddle version >= 2.3.0 or develop
            else:
                if self.mode == "train" or self.amp_eval:
                    self.model = paddle.amp.decorate(
                        models=self.model,
                        level=self.amp_level,
                        save_dtype='float32')

            if self.mode == "train" and len(self.train_loss_func.parameters(
            )) > 0:
                self.train_loss_func = paddle.amp.decorate(
                    models=self.train_loss_func,
                    level=self.amp_level,
                    save_dtype='float32')

    def _init_dist(self):
        # check the gpu num
        world_size = dist.get_world_size()
        self.config["Global"]["distributed"] = world_size != 1
        # TODO(gaotingquan):
        if self.mode == "train":
            std_gpu_num = 8 if isinstance(
                self.config["Optimizer"],
                dict) and self.config["Optimizer"]["name"] == "AdamW" else 4
            if world_size != std_gpu_num:
                msg = f"The training strategy provided by PaddleClas is based on {std_gpu_num} gpus. But the number of gpu is {world_size} in current training. Please modify the stategy (learning rate, batch size and so on) if use this config to train."
                logger.warning(msg)

        if self.config["Global"]["distributed"]:
            dist.init_parallel_env()
            self.model = paddle.DataParallel(self.model)
            if self.mode == 'train' and len(self.train_loss_func.parameters(
            )) > 0:
                self.train_loss_func = paddle.DataParallel(
                    self.train_loss_func)

D
dongshuilong 已提交
485

W
dbg  
weishengyu 已提交
486
class ExportModel(TheseusLayer):
D
dongshuilong 已提交
487 488 489 490
    """
    ExportModel: add softmax onto the model
    """

C
cuicheng01 已提交
491
    def __init__(self, config, model, use_multilabel):
D
dongshuilong 已提交
492 493 494 495 496 497 498 499 500 501 502 503
        super().__init__()
        self.base_model = model
        # we should choose a final model to export
        if isinstance(self.base_model, DistillationModel):
            self.infer_model_name = config["infer_model_name"]
        else:
            self.infer_model_name = None

        self.infer_output_key = config.get("infer_output_key", None)
        if self.infer_output_key == "features" and isinstance(self.base_model,
                                                              RecModel):
            self.base_model.head = IdentityHead()
C
cuicheng01 已提交
504 505
        if use_multilabel:
            self.out_act = nn.Sigmoid()
D
dongshuilong 已提交
506
        else:
C
cuicheng01 已提交
507 508 509 510
            if config.get("infer_add_softmax", True):
                self.out_act = nn.Softmax(axis=-1)
            else:
                self.out_act = None
D
dongshuilong 已提交
511 512 513 514 515 516 517 518 519 520 521 522 523 524 525

    def eval(self):
        self.training = False
        for layer in self.sublayers():
            layer.training = False
            layer.eval()

    def forward(self, x):
        x = self.base_model(x)
        if isinstance(x, list):
            x = x[0]
        if self.infer_model_name is not None:
            x = x[self.infer_model_name]
        if self.infer_output_key is not None:
            x = x[self.infer_output_key]
C
cuicheng01 已提交
526
        if self.out_act is not None:
wc晨曦's avatar
wc晨曦 已提交
527 528
            if isinstance(x, dict):
                x = x["logits"]
C
cuicheng01 已提交
529
            x = self.out_act(x)
D
dongshuilong 已提交
530
        return x