# Copyright (c) 2022 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 import os import sys import copy import time import typing import math import numpy as np import paddle import paddle.nn as nn import paddle.distributed as dist from paddle.distributed import fleet from ppdet.optimizer import ModelEMA, SimpleModelEMA from ppdet.core.workspace import create from ppdet.utils.checkpoint import load_weight, load_pretrain_weight import ppdet.utils.stats as stats from ppdet.utils import profiler from ppdet.modeling.ssod_utils import align_weak_strong_shape from .trainer import Trainer from ppdet.utils.logger import setup_logger logger = setup_logger('ppdet.engine') __all__ = ['Trainer_DenseTeacher'] class Trainer_DenseTeacher(Trainer): def __init__(self, cfg, mode='train'): self.cfg = cfg assert mode.lower() in ['train', 'eval', 'test'], \ "mode should be 'train', 'eval' or 'test'" self.mode = mode.lower() self.optimizer = None self.is_loaded_weights = False self.use_amp = self.cfg.get('amp', False) self.amp_level = self.cfg.get('amp_level', 'O1') self.custom_white_list = self.cfg.get('custom_white_list', None) self.custom_black_list = self.cfg.get('custom_black_list', None) # build data loader capital_mode = self.mode.capitalize() self.dataset = self.cfg['{}Dataset'.format(capital_mode)] = create( '{}Dataset'.format(capital_mode))() if self.mode == 'train': self.dataset_unlabel = self.cfg['UnsupTrainDataset'] = create( 'UnsupTrainDataset') self.loader = create('SemiTrainReader')( self.dataset, self.dataset_unlabel, cfg.worker_num) # build model if 'model' not in self.cfg: self.model = create(cfg.architecture) else: self.model = self.cfg.model self.is_loaded_weights = True # EvalDataset build with BatchSampler to evaluate in single device # TODO: multi-device evaluate if self.mode == 'eval': self._eval_batch_sampler = paddle.io.BatchSampler( self.dataset, batch_size=self.cfg.EvalReader['batch_size']) # If metric is VOC, need to be set collate_batch=False. if cfg.metric == 'VOC': cfg['EvalReader']['collate_batch'] = False self.loader = create('EvalReader')(self.dataset, cfg.worker_num, self._eval_batch_sampler) # TestDataset build after user set images, skip loader creation here # build optimizer in train mode if self.mode == 'train': steps_per_epoch = len(self.loader) if steps_per_epoch < 1: logger.warning( "Samples in dataset are less than batch_size, please set smaller batch_size in TrainReader." ) self.lr = create('LearningRate')(steps_per_epoch) self.optimizer = create('OptimizerBuilder')(self.lr, self.model) # Unstructured pruner is only enabled in the train mode. if self.cfg.get('unstructured_prune'): self.pruner = create('UnstructuredPruner')(self.model, steps_per_epoch) if self.use_amp and self.amp_level == 'O2': self.model, self.optimizer = paddle.amp.decorate( models=self.model, optimizers=self.optimizer, level=self.amp_level) self.use_ema = ('use_ema' in cfg and cfg['use_ema']) if self.use_ema: ema_decay = self.cfg.get('ema_decay', 0.9998) ema_decay_type = self.cfg.get('ema_decay_type', 'threshold') cycle_epoch = self.cfg.get('cycle_epoch', -1) ema_black_list = self.cfg.get('ema_black_list', None) self.ema = ModelEMA( self.model, decay=ema_decay, ema_decay_type=ema_decay_type, cycle_epoch=cycle_epoch, ema_black_list=ema_black_list) self.ema_start_iters = self.cfg.get('ema_start_iters', 0) # simple_ema for SSOD self.use_simple_ema = ('use_simple_ema' in cfg and cfg['use_simple_ema']) if self.use_simple_ema: self.use_ema = True ema_decay = self.cfg.get('ema_decay', 0.9996) self.ema = SimpleModelEMA(self.model, decay=ema_decay) self.ema_start_iters = self.cfg.get('ema_start_iters', 0) self._nranks = dist.get_world_size() self._local_rank = dist.get_rank() self.status = {} self.start_epoch = 0 self.end_epoch = 0 if 'epoch' not in cfg else cfg.epoch # initial default callbacks self._init_callbacks() # initial default metrics self._init_metrics() self._reset_metrics() def load_weights(self, weights): if self.is_loaded_weights: return self.start_epoch = 0 load_pretrain_weight(self.model, weights) load_pretrain_weight(self.ema.model, weights) logger.info("Load weights {} to start training for teacher and student". format(weights)) def resume_weights(self, weights, exchange=True): # support Distill resume weights if hasattr(self.model, 'student_model'): self.start_epoch = load_weight(self.model.student_model, weights, self.optimizer, exchange) else: self.start_epoch = load_weight(self.model, weights, self.optimizer, self.ema if self.use_ema else None, exchange) logger.debug("Resume weights of epoch {}".format(self.start_epoch)) def train(self, validate=False): self.semi_start_iters = self.cfg.get('semi_start_iters', 5000) Init_mark = False if validate: self.cfg['EvalDataset'] = self.cfg.EvalDataset = create( "EvalDataset")() sync_bn = (getattr(self.cfg, 'norm_type', None) == 'sync_bn' and self.cfg.use_gpu and self._nranks > 1) if sync_bn: self.model = paddle.nn.SyncBatchNorm.convert_sync_batchnorm( self.model) if self.cfg.get('fleet', False): self.model = fleet.distributed_model(self.model) self.optimizer = fleet.distributed_optimizer(self.optimizer) elif self._nranks > 1: find_unused_parameters = self.cfg[ 'find_unused_parameters'] if 'find_unused_parameters' in self.cfg else False self.model = paddle.DataParallel( self.model, find_unused_parameters=find_unused_parameters) self.ema.model = paddle.DataParallel( self.ema.model, find_unused_parameters=find_unused_parameters) self.status.update({ 'epoch_id': self.start_epoch, 'step_id': 0, 'steps_per_epoch': len(self.loader) }) self.status['batch_time'] = stats.SmoothedValue( self.cfg.log_iter, fmt='{avg:.4f}') self.status['data_time'] = stats.SmoothedValue( self.cfg.log_iter, fmt='{avg:.4f}') self.status['training_staus'] = stats.TrainingStats(self.cfg.log_iter) if self.cfg.get('print_flops', False): flops_loader = create('{}Reader'.format(self.mode.capitalize()))( self.dataset, self.cfg.worker_num) self._flops(flops_loader) profiler_options = self.cfg.get('profiler_options', None) self._compose_callback.on_train_begin(self.status) train_cfg = self.cfg.DenseTeacher['train_cfg'] concat_sup_data = train_cfg.get('concat_sup_data', True) for param in self.ema.model.parameters(): param.stop_gradient = True for epoch_id in range(self.start_epoch, self.cfg.epoch): self.status['mode'] = 'train' self.status['epoch_id'] = epoch_id self._compose_callback.on_epoch_begin(self.status) self.loader.dataset_label.set_epoch(epoch_id) self.loader.dataset_unlabel.set_epoch(epoch_id) iter_tic = time.time() loss_dict = { 'loss': paddle.to_tensor([0]), 'loss_sup_sum': paddle.to_tensor([0]), 'loss_unsup_sum': paddle.to_tensor([0]), 'fg_sum': paddle.to_tensor([0]), } if self._nranks > 1: for k in self.model._layers.get_loss_keys(): loss_dict.update({k: paddle.to_tensor([0.])}) for k in self.model._layers.get_loss_keys(): loss_dict.update({'distill_' + k: paddle.to_tensor([0.])}) else: for k in self.model.get_loss_keys(): loss_dict.update({k: paddle.to_tensor([0.])}) for k in self.model.get_loss_keys(): loss_dict.update({'distill_' + k: paddle.to_tensor([0.])}) # Note: for step_id, data in enumerate(self.loader): # enumerate bug for step_id in range(len(self.loader)): data = next(self.loader) self.model.train() self.ema.model.eval() data_sup_w, data_sup_s, data_unsup_w, data_unsup_s = data self.status['data_time'].update(time.time() - iter_tic) self.status['step_id'] = step_id profiler.add_profiler_step(profiler_options) self._compose_callback.on_step_begin(self.status) if data_sup_w['image'].shape != data_sup_s['image'].shape: data_sup_w, data_sup_s = align_weak_strong_shape(data_sup_w, data_sup_s) if concat_sup_data: for k, v in data_sup_s.items(): data_sup_s[k] = paddle.concat([v, data_sup_w[k]]) loss_dict_sup = self.model(data_sup_s) else: loss_dict_sup_w = self.model(data_sup_w) loss_dict_sup = self.model(data_sup_s) for k, v in loss_dict_sup_w.items(): loss_dict_sup[k] = (loss_dict_sup[k] + v) * 0.5 data_sup_w['epoch_id'] = epoch_id data_sup_s['epoch_id'] = epoch_id losses_sup = loss_dict_sup['loss'] * train_cfg['sup_weight'] losses_sup.backward() losses = losses_sup.detach() loss_dict.update(loss_dict_sup) loss_dict.update({'loss_sup_sum': loss_dict['loss']}) curr_iter = len(self.loader) * epoch_id + step_id st_iter = self.semi_start_iters if curr_iter == st_iter: logger.info("***" * 30) logger.info('Semi starting ...') logger.info("***" * 30) if curr_iter > st_iter: unsup_weight = train_cfg['unsup_weight'] if train_cfg['suppress'] == 'linear': tar_iter = st_iter * 2 if curr_iter <= tar_iter: unsup_weight *= (curr_iter - st_iter) / st_iter elif train_cfg['suppress'] == 'exp': tar_iter = st_iter + 2000 if curr_iter <= tar_iter: scale = np.exp((curr_iter - tar_iter) / 1000) unsup_weight *= scale elif train_cfg['suppress'] == 'step': tar_iter = st_iter * 2 if curr_iter <= tar_iter: unsup_weight *= 0.25 else: raise ValueError if data_unsup_w['image'].shape != data_unsup_s[ 'image'].shape: data_unsup_w, data_unsup_s = align_weak_strong_shape( data_unsup_w, data_unsup_s) data_unsup_w['epoch_id'] = epoch_id data_unsup_s['epoch_id'] = epoch_id data_unsup_s['get_data'] = True student_preds = self.model(data_unsup_s) with paddle.no_grad(): data_unsup_w['is_teacher'] = True teacher_preds = self.ema.model(data_unsup_w) if self._nranks > 1: loss_dict_unsup = self.model._layers.get_distill_loss( student_preds, teacher_preds, ratio=train_cfg['ratio']) else: loss_dict_unsup = self.model.get_distill_loss( student_preds, teacher_preds, ratio=train_cfg['ratio']) fg_num = loss_dict_unsup["fg_sum"] del loss_dict_unsup["fg_sum"] distill_weights = train_cfg['loss_weight'] loss_dict_unsup = { k: v * distill_weights[k] for k, v in loss_dict_unsup.items() } losses_unsup = sum([ metrics_value for metrics_value in loss_dict_unsup.values() ]) * unsup_weight losses_unsup.backward() loss_dict.update(loss_dict_unsup) loss_dict.update({'loss_unsup_sum': losses_unsup}) losses += losses_unsup.detach() loss_dict.update({"fg_sum": fg_num}) loss_dict['loss'] = losses self.optimizer.step() curr_lr = self.optimizer.get_lr() self.lr.step() self.optimizer.clear_grad() self.status['learning_rate'] = curr_lr if self._nranks < 2 or self._local_rank == 0: self.status['training_staus'].update(loss_dict) self.status['batch_time'].update(time.time() - iter_tic) self._compose_callback.on_step_end(self.status) # Note: ema_start_iters if self.use_ema and curr_iter == self.ema_start_iters: logger.info("***" * 30) logger.info('EMA starting ...') logger.info("***" * 30) self.ema.update(self.model, decay=0) elif self.use_ema and curr_iter > self.ema_start_iters: self.ema.update(self.model) iter_tic = time.time() is_snapshot = (self._nranks < 2 or self._local_rank == 0) \ and ((epoch_id + 1) % self.cfg.snapshot_epoch == 0 or epoch_id == self.end_epoch - 1) if is_snapshot and self.use_ema: # apply ema weight on model weight = copy.deepcopy(self.ema.model.state_dict()) for k, v in weight.items(): if paddle.is_floating_point(v): weight[k].stop_gradient = True self.status['weight'] = weight self._compose_callback.on_epoch_end(self.status) if validate and is_snapshot: if not hasattr(self, '_eval_loader'): # build evaluation dataset and loader self._eval_dataset = self.cfg.EvalDataset self._eval_batch_sampler = \ paddle.io.BatchSampler( self._eval_dataset, batch_size=self.cfg.EvalReader['batch_size']) # If metric is VOC, need to be set collate_batch=False. if self.cfg.metric == 'VOC': self.cfg['EvalReader']['collate_batch'] = False self._eval_loader = create('EvalReader')( self._eval_dataset, self.cfg.worker_num, batch_sampler=self._eval_batch_sampler) # if validation in training is enabled, metrics should be re-init # Init_mark makes sure this code will only execute once if validate and Init_mark == False: Init_mark = True self._init_metrics(validate=validate) self._reset_metrics() with paddle.no_grad(): self.status['save_best_model'] = True self._eval_with_loader(self._eval_loader) if is_snapshot and self.use_ema: self.status.pop('weight') self._compose_callback.on_train_end(self.status) def evaluate(self): # get distributed model if self.cfg.get('fleet', False): self.model = fleet.distributed_model(self.model) self.optimizer = fleet.distributed_optimizer(self.optimizer) elif self._nranks > 1: find_unused_parameters = self.cfg[ 'find_unused_parameters'] if 'find_unused_parameters' in self.cfg else False self.model = paddle.DataParallel( self.model, find_unused_parameters=find_unused_parameters) with paddle.no_grad(): self._eval_with_loader(self.loader) def _eval_with_loader(self, loader): sample_num = 0 tic = time.time() self._compose_callback.on_epoch_begin(self.status) self.status['mode'] = 'eval' test_cfg = self.cfg.DenseTeacher['test_cfg'] if test_cfg['inference_on'] == 'teacher': logger.info("***** teacher model evaluating *****") eval_model = self.ema.model else: logger.info("***** student model evaluating *****") eval_model = self.model eval_model.eval() if self.cfg.get('print_flops', False): flops_loader = create('{}Reader'.format(self.mode.capitalize()))( self.dataset, self.cfg.worker_num, self._eval_batch_sampler) self._flops(flops_loader) for step_id, data in enumerate(loader): self.status['step_id'] = step_id self._compose_callback.on_step_begin(self.status) # forward if self.use_amp: with paddle.amp.auto_cast( enable=self.cfg.use_gpu or self.cfg.use_mlu, custom_white_list=self.custom_white_list, custom_black_list=self.custom_black_list, level=self.amp_level): outs = eval_model(data) else: outs = eval_model(data) # update metrics for metric in self._metrics: metric.update(data, outs) # multi-scale inputs: all inputs have same im_id if isinstance(data, typing.Sequence): sample_num += data[0]['im_id'].numpy().shape[0] else: sample_num += data['im_id'].numpy().shape[0] self._compose_callback.on_step_end(self.status) self.status['sample_num'] = sample_num self.status['cost_time'] = time.time() - tic # accumulate metric to log out for metric in self._metrics: metric.accumulate() metric.log() self._compose_callback.on_epoch_end(self.status) # reset metric states for metric may performed multiple times self._reset_metrics()