# Copyright 2020 Huawei Technologies Co., Ltd # # 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. # ============================================================================ """task distill script""" import os import re import argparse import mindspore.common.dtype as mstype from mindspore import context from mindspore.train.model import Model from mindspore.train.callback import TimeMonitor from mindspore.train.serialization import load_checkpoint, load_param_into_net from mindspore.nn.wrap.loss_scale import DynamicLossScaleUpdateCell from mindspore.nn.optim import AdamWeightDecay from mindspore import log as logger from src.dataset import create_tinybert_dataset, DataType from src.utils import LossCallBack, ModelSaveCkpt, EvalCallBack, BertLearningRate from src.assessment_method import Accuracy, F1 from src.td_config import phase1_cfg, phase2_cfg, eval_cfg, td_teacher_net_cfg, td_student_net_cfg from src.tinybert_for_gd_td import BertEvaluationWithLossScaleCell, BertNetworkWithLoss_td, BertEvaluationCell from src.tinybert_model import BertModelCLS, BertModelNER _cur_dir = os.getcwd() td_phase1_save_ckpt_dir = os.path.join(_cur_dir, 'tinybert_td_phase1_save_ckpt') td_phase2_save_ckpt_dir = os.path.join(_cur_dir, 'tinybert_td_phase2_save_ckpt') if not os.path.exists(td_phase1_save_ckpt_dir): os.makedirs(td_phase1_save_ckpt_dir) if not os.path.exists(td_phase2_save_ckpt_dir): os.makedirs(td_phase2_save_ckpt_dir) def parse_args(): """ parse args """ parser = argparse.ArgumentParser(description='tinybert task distill') parser.add_argument("--device_target", type=str, default="Ascend", choices=['Ascend', 'GPU', 'CPU'], help='device where the code will be implemented. (Default: Ascend)') parser.add_argument("--do_train", type=str, default="true", choices=["true", "false"], help="Do train task, default is true.") parser.add_argument("--do_eval", type=str, default="true", choices=["true", "false"], help="Do eval task, default is true.") parser.add_argument("--td_phase1_epoch_size", type=int, default=10, help="Epoch size for td phase 1, default is 10.") parser.add_argument("--td_phase2_epoch_size", type=int, default=3, help="Epoch size for td phase 2, default is 3.") parser.add_argument("--device_id", type=int, default=0, help="Device id, default is 0.") parser.add_argument("--do_shuffle", type=str, default="true", choices=["true", "false"], help="Enable shuffle for dataset, default is true.") parser.add_argument("--enable_data_sink", type=str, default="true", choices=["true", "false"], help="Enable data sink, default is true.") parser.add_argument("--save_ckpt_step", type=int, default=100, help="Enable data sink, default is true.") parser.add_argument("--max_ckpt_num", type=int, default=1, help="Enable data sink, default is true.") parser.add_argument("--data_sink_steps", type=int, default=1, help="Sink steps for each epoch, default is 1.") parser.add_argument("--load_teacher_ckpt_path", type=str, default="", help="Load checkpoint file path") parser.add_argument("--load_gd_ckpt_path", type=str, default="", help="Load checkpoint file path") parser.add_argument("--load_td1_ckpt_path", type=str, default="", help="Load checkpoint file path") parser.add_argument("--train_data_dir", type=str, default="", help="Data path, it is better to use absolute path") parser.add_argument("--eval_data_dir", type=str, default="", help="Data path, it is better to use absolute path") parser.add_argument("--schema_dir", type=str, default="", help="Schema path, it is better to use absolute path") parser.add_argument("--task_type", type=str, default="classification", choices=["classification", "ner"], help="The type of the task to train.") parser.add_argument("--task_name", type=str, default="", choices=["SST-2", "QNLI", "MNLI", "TNEWS", "CLUENER"], help="The name of the task to train.") parser.add_argument("--assessment_method", type=str, default="accuracy", choices=["accuracy", "bf1", "mf1"], help="assessment_method include: [accuracy, bf1, mf1], default is accuracy") parser.add_argument("--dataset_type", type=str, default="tfrecord", help="dataset type tfrecord/mindrecord, default is tfrecord") args = parser.parse_args() if args.do_train.lower() != "true" and args.do_eval.lower() != "true": raise ValueError("do train or do eval must have one be true, please confirm your config") if args.task_name in ["SST-2", "QNLI", "MNLI", "TNEWS"] and args.task_type != "classification": raise ValueError(f"{args.task_name} is a classification dataset, please set --task_type=classification") if args.task_name in ["CLUENER"] and args.task_type != "ner": raise ValueError(f"{args.task_name} is a ner dataset, please set --task_type=ner") if args.task_name in ["SST-2", "QNLI", "MNLI"] and \ (td_teacher_net_cfg.vocab_size != 30522 or td_student_net_cfg.vocab_size != 30522): logger.warning(f"{args.task_name} is an English dataset. Usually, we use 21128 for CN vocabs and 30522 for "\ "EN vocabs according to the origin paper.") if args.task_name in ["TNEWS", "CLUENER"] and \ (td_teacher_net_cfg.vocab_size != 21128 or td_student_net_cfg.vocab_size != 21128): logger.warning(f"{args.task_name} is a Chinese dataset. Usually, we use 21128 for CN vocabs and 30522 for " \ "EN vocabs according to the origin paper.") return args args_opt = parse_args() if args_opt.dataset_type == "tfrecord": dataset_type = DataType.TFRECORD elif args_opt.dataset_type == "mindrecord": dataset_type = DataType.MINDRECORD else: raise Exception("dataset format is not supported yet") DEFAULT_NUM_LABELS = 2 DEFAULT_SEQ_LENGTH = 128 task_params = {"SST-2": {"num_labels": 2, "seq_length": 64}, "QNLI": {"num_labels": 2, "seq_length": 128}, "MNLI": {"num_labels": 3, "seq_length": 128}, "TNEWS": {"num_labels": 15, "seq_length": 128}, "CLUENER": {"num_labels": 43, "seq_length": 128}} class Task: """ Encapsulation class of get the task parameter. """ def __init__(self, task_name): self.task_name = task_name @property def num_labels(self): if self.task_name in task_params and "num_labels" in task_params[self.task_name]: return task_params[self.task_name]["num_labels"] return DEFAULT_NUM_LABELS @property def seq_length(self): if self.task_name in task_params and "seq_length" in task_params[self.task_name]: return task_params[self.task_name]["seq_length"] return DEFAULT_SEQ_LENGTH task = Task(args_opt.task_name) def run_predistill(): """ run predistill """ cfg = phase1_cfg load_teacher_checkpoint_path = args_opt.load_teacher_ckpt_path load_student_checkpoint_path = args_opt.load_gd_ckpt_path netwithloss = BertNetworkWithLoss_td(teacher_config=td_teacher_net_cfg, teacher_ckpt=load_teacher_checkpoint_path, student_config=td_student_net_cfg, student_ckpt=load_student_checkpoint_path, is_training=True, task_type=args_opt.task_type, num_labels=task.num_labels, is_predistill=True) rank = 0 device_num = 1 dataset = create_tinybert_dataset('td', cfg.batch_size, device_num, rank, args_opt.do_shuffle, args_opt.train_data_dir, args_opt.schema_dir, data_type=dataset_type) dataset_size = dataset.get_dataset_size() print('td1 dataset size: ', dataset_size) print('td1 dataset repeatcount: ', dataset.get_repeat_count()) if args_opt.enable_data_sink == 'true': repeat_count = args_opt.td_phase1_epoch_size * dataset_size // args_opt.data_sink_steps time_monitor_steps = args_opt.data_sink_steps else: repeat_count = args_opt.td_phase1_epoch_size time_monitor_steps = dataset_size optimizer_cfg = cfg.optimizer_cfg lr_schedule = BertLearningRate(learning_rate=optimizer_cfg.AdamWeightDecay.learning_rate, end_learning_rate=optimizer_cfg.AdamWeightDecay.end_learning_rate, warmup_steps=int(dataset_size / 10), decay_steps=int(dataset_size * args_opt.td_phase1_epoch_size), power=optimizer_cfg.AdamWeightDecay.power) params = netwithloss.trainable_params() decay_params = list(filter(optimizer_cfg.AdamWeightDecay.decay_filter, params)) other_params = list(filter(lambda x: not optimizer_cfg.AdamWeightDecay.decay_filter(x), params)) group_params = [{'params': decay_params, 'weight_decay': optimizer_cfg.AdamWeightDecay.weight_decay}, {'params': other_params, 'weight_decay': 0.0}, {'order_params': params}] optimizer = AdamWeightDecay(group_params, learning_rate=lr_schedule, eps=optimizer_cfg.AdamWeightDecay.eps) callback = [TimeMonitor(time_monitor_steps), LossCallBack(), ModelSaveCkpt(netwithloss.bert, args_opt.save_ckpt_step, args_opt.max_ckpt_num, td_phase1_save_ckpt_dir)] if enable_loss_scale: update_cell = DynamicLossScaleUpdateCell(loss_scale_value=cfg.loss_scale_value, scale_factor=cfg.scale_factor, scale_window=cfg.scale_window) netwithgrads = BertEvaluationWithLossScaleCell(netwithloss, optimizer=optimizer, scale_update_cell=update_cell) else: netwithgrads = BertEvaluationCell(netwithloss, optimizer=optimizer) model = Model(netwithgrads) model.train(repeat_count, dataset, callbacks=callback, dataset_sink_mode=(args_opt.enable_data_sink == 'true'), sink_size=args_opt.data_sink_steps) def run_task_distill(ckpt_file): """ run task distill """ if ckpt_file == '': raise ValueError("Student ckpt file should not be None") cfg = phase2_cfg load_teacher_checkpoint_path = args_opt.load_teacher_ckpt_path load_student_checkpoint_path = ckpt_file netwithloss = BertNetworkWithLoss_td(teacher_config=td_teacher_net_cfg, teacher_ckpt=load_teacher_checkpoint_path, student_config=td_student_net_cfg, student_ckpt=load_student_checkpoint_path, is_training=True, task_type=args_opt.task_type, num_labels=task.num_labels, is_predistill=False) rank = 0 device_num = 1 train_dataset = create_tinybert_dataset('td', cfg.batch_size, device_num, rank, args_opt.do_shuffle, args_opt.train_data_dir, args_opt.schema_dir, data_type=dataset_type) dataset_size = train_dataset.get_dataset_size() print('td2 train dataset size: ', dataset_size) print('td2 train dataset repeatcount: ', train_dataset.get_repeat_count()) if args_opt.enable_data_sink == 'true': repeat_count = args_opt.td_phase2_epoch_size * train_dataset.get_dataset_size() // args_opt.data_sink_steps time_monitor_steps = args_opt.data_sink_steps else: repeat_count = args_opt.td_phase2_epoch_size time_monitor_steps = dataset_size optimizer_cfg = cfg.optimizer_cfg lr_schedule = BertLearningRate(learning_rate=optimizer_cfg.AdamWeightDecay.learning_rate, end_learning_rate=optimizer_cfg.AdamWeightDecay.end_learning_rate, warmup_steps=int(dataset_size * args_opt.td_phase2_epoch_size / 10), decay_steps=int(dataset_size * args_opt.td_phase2_epoch_size), power=optimizer_cfg.AdamWeightDecay.power) params = netwithloss.trainable_params() decay_params = list(filter(optimizer_cfg.AdamWeightDecay.decay_filter, params)) other_params = list(filter(lambda x: not optimizer_cfg.AdamWeightDecay.decay_filter(x), params)) group_params = [{'params': decay_params, 'weight_decay': optimizer_cfg.AdamWeightDecay.weight_decay}, {'params': other_params, 'weight_decay': 0.0}, {'order_params': params}] optimizer = AdamWeightDecay(group_params, learning_rate=lr_schedule, eps=optimizer_cfg.AdamWeightDecay.eps) eval_dataset = create_tinybert_dataset('td', eval_cfg.batch_size, device_num, rank, args_opt.do_shuffle, args_opt.eval_data_dir, args_opt.schema_dir, data_type=dataset_type) print('td2 eval dataset size: ', eval_dataset.get_dataset_size()) if args_opt.do_eval.lower() == "true": callback = [TimeMonitor(time_monitor_steps), LossCallBack(), EvalCallBack(netwithloss.bert, eval_dataset)] else: callback = [TimeMonitor(time_monitor_steps), LossCallBack(), ModelSaveCkpt(netwithloss.bert, args_opt.save_ckpt_step, args_opt.max_ckpt_num, td_phase2_save_ckpt_dir)] if enable_loss_scale: update_cell = DynamicLossScaleUpdateCell(loss_scale_value=cfg.loss_scale_value, scale_factor=cfg.scale_factor, scale_window=cfg.scale_window) netwithgrads = BertEvaluationWithLossScaleCell(netwithloss, optimizer=optimizer, scale_update_cell=update_cell) else: netwithgrads = BertEvaluationCell(netwithloss, optimizer=optimizer) model = Model(netwithgrads) model.train(repeat_count, train_dataset, callbacks=callback, dataset_sink_mode=(args_opt.enable_data_sink == 'true'), sink_size=args_opt.data_sink_steps) def eval_result_print(assessment_method="accuracy", callback=None): """print eval result""" if assessment_method == "accuracy": print("============== acc is {}".format(callback.acc_num / callback.total_num)) elif assessment_method == "bf1": print("Precision {:.6f} ".format(callback.TP / (callback.TP + callback.FP))) print("Recall {:.6f} ".format(callback.TP / (callback.TP + callback.FN))) print("F1 {:.6f} ".format(2 * callback.TP / (2 * callback.TP + callback.FP + callback.FN))) elif assessment_method == "mf1": print("F1 {:.6f} ".format(callback.eval())) else: raise ValueError("Assessment method not supported, support: [accuracy, f1]") def do_eval_standalone(): """ do eval standalone """ ckpt_file = args_opt.load_td1_ckpt_path if ckpt_file == '': raise ValueError("Student ckpt file should not be None") if args_opt.task_type == "classification": eval_model = BertModelCLS(td_student_net_cfg, False, task.num_labels, 0.0, phase_type="student") elif args_opt.task_type == "ner": eval_model = BertModelNER(td_student_net_cfg, False, task.num_labels, 0.0, phase_type="student") else: raise ValueError(f"Not support the task type {args_opt.task_type}") param_dict = load_checkpoint(ckpt_file) new_param_dict = {} for key, value in param_dict.items(): new_key = re.sub('tinybert_', 'bert_', key) new_key = re.sub('^bert.', '', new_key) new_param_dict[new_key] = value load_param_into_net(eval_model, new_param_dict) eval_model.set_train(False) eval_dataset = create_tinybert_dataset('td', batch_size=eval_cfg.batch_size, device_num=1, rank=0, do_shuffle="false", data_dir=args_opt.eval_data_dir, schema_dir=args_opt.schema_dir, data_type=dataset_type) print('eval dataset size: ', eval_dataset.get_dataset_size()) print('eval dataset batch size: ', eval_dataset.get_batch_size()) if args_opt.assessment_method == "accuracy": callback = Accuracy() elif args_opt.assessment_method == "bf1": callback = F1(num_labels=task.num_labels) elif args_opt.assessment_method == "mf1": callback = F1(num_labels=task.num_labels, mode="MultiLabel") else: raise ValueError("Assessment method not supported, support: [accuracy, f1]") columns_list = ["input_ids", "input_mask", "segment_ids", "label_ids"] for data in eval_dataset.create_dict_iterator(num_epochs=1): input_data = [] for i in columns_list: input_data.append(data[i]) input_ids, input_mask, token_type_id, label_ids = input_data logits = eval_model(input_ids, token_type_id, input_mask) callback.update(logits, label_ids) print("==============================================================") eval_result_print(args_opt.assessment_method, callback) print("==============================================================") if __name__ == '__main__': context.set_context(mode=context.GRAPH_MODE, device_target=args_opt.device_target, reserve_class_name_in_scope=False) if args_opt.device_target == "Ascend": context.set_context(device_id=args_opt.device_id) enable_loss_scale = True if args_opt.device_target == "GPU": if td_student_net_cfg.compute_type != mstype.float32: logger.warning('Compute about the student only support float32 temporarily, run with float32.') td_student_net_cfg.compute_type = mstype.float32 # Backward of the network are calculated using fp32, # and the loss scale is not necessary enable_loss_scale = False if args_opt.device_target == "CPU": logger.warning('CPU only support float32 temporarily, run with float32.') td_teacher_net_cfg.dtype = mstype.float32 td_teacher_net_cfg.compute_type = mstype.float32 td_student_net_cfg.dtype = mstype.float32 td_student_net_cfg.compute_type = mstype.float32 enable_loss_scale = False td_teacher_net_cfg.seq_length = task.seq_length td_student_net_cfg.seq_length = task.seq_length if args_opt.do_train == "true": # run predistill run_predistill() lists = os.listdir(td_phase1_save_ckpt_dir) if lists: lists.sort(key=lambda fn: os.path.getmtime(td_phase1_save_ckpt_dir+'/'+fn)) name_ext = os.path.splitext(lists[-1]) if name_ext[-1] != ".ckpt": raise ValueError("Invalid file, checkpoint file should be .ckpt file") newest_ckpt_file = os.path.join(td_phase1_save_ckpt_dir, lists[-1]) # run task distill run_task_distill(newest_ckpt_file) else: raise ValueError("Checkpoint file not exists, please make sure ckpt file has been saved") else: do_eval_standalone()