run_classifier.py 15.0 KB
Newer Older
T
tianxin04 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
#   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.
"""Finetuning on classification tasks."""

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

import os
import time
import multiprocessing

T
tianxin 已提交
24 25 26 27 28
# NOTE(paddle-dev): All of these flags should be
# set before `import paddle`. Otherwise, it would
# not take any effect.
os.environ['FLAGS_eager_delete_tensor_gb'] = '0'  # enable gc

T
tianxin04 已提交
29 30 31 32
import paddle.fluid as fluid

import reader.task_reader as task_reader
from model.ernie import ErnieConfig
T
tianxin 已提交
33
from finetune.classifier import create_model, evaluate, predict
T
tianxin04 已提交
34
from optimization import optimization
T
tianxin 已提交
35
from utils.args import print_arguments, check_cuda
T
tianxin04 已提交
36
from utils.init import init_pretraining_params, init_checkpoint
Z
zhengya01 已提交
37
from utils.cards import get_cards
T
format  
tianxin04 已提交
38
from finetune_args import parser
T
tianxin04 已提交
39 40 41

args = parser.parse_args()

T
format  
tianxin04 已提交
42

T
tianxin04 已提交
43 44 45 46 47 48 49 50 51 52 53 54
def main(args):
    ernie_config = ErnieConfig(args.ernie_config_path)
    ernie_config.print_config()

    if args.use_cuda:
        place = fluid.CUDAPlace(int(os.getenv('FLAGS_selected_gpus', '0')))
        dev_count = fluid.core.get_cuda_device_count()
    else:
        place = fluid.CPUPlace()
        dev_count = int(os.environ.get('CPU_NUM', multiprocessing.cpu_count()))
    exe = fluid.Executor(place)

T
format  
tianxin04 已提交
55 56 57 58 59 60
    reader = task_reader.ClassifyReader(
        vocab_path=args.vocab_path,
        label_map_config=args.label_map_config,
        max_seq_len=args.max_seq_len,
        do_lower_case=args.do_lower_case,
        in_tokens=args.in_tokens,
T
tianxin 已提交
61 62 63 64 65 66
        random_seed=args.random_seed,
        tokenizer=args.tokenizer,
        is_classify=args.is_classify,
        is_regression=args.is_regression,
        for_cn=args.for_cn,
        task_id=args.task_id)
T
tianxin04 已提交
67 68 69 70 71

    if not (args.do_train or args.do_val or args.do_test):
        raise ValueError("For args `do_train`, `do_val` and `do_test`, at "
                         "least one of them must be True.")

T
tianxin 已提交
72 73
    if args.do_test:
        assert args.test_save is not None
T
tianxin04 已提交
74 75 76 77
    startup_prog = fluid.Program()
    if args.random_seed is not None:
        startup_prog.random_seed = args.random_seed

T
tianxin 已提交
78 79
    if args.predict_batch_size == None:
        args.predict_batch_size = args.batch_size
T
tianxin04 已提交
80 81 82 83 84
    if args.do_train:
        train_data_generator = reader.data_generator(
            input_file=args.train_set,
            batch_size=args.batch_size,
            epoch=args.epoch,
T
tianxin 已提交
85 86
            dev_count=dev_count,
            shuffle=True,
T
tianxin04 已提交
87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103
            phase="train")

        num_train_examples = reader.get_num_examples(args.train_set)

        if args.in_tokens:
            max_train_steps = args.epoch * num_train_examples // (
                args.batch_size // args.max_seq_len) // dev_count
        else:
            max_train_steps = args.epoch * num_train_examples // args.batch_size // dev_count

        warmup_steps = int(max_train_steps * args.warmup_proportion)
        print("Device count: %d" % dev_count)
        print("Num train examples: %d" % num_train_examples)
        print("Max train steps: %d" % max_train_steps)
        print("Num warmup steps: %d" % warmup_steps)

        train_program = fluid.Program()
Z
zhengya01 已提交
104 105
        if args.random_seed is not None and args.enable_ce:
            train_program.random_seed = args.random_seed
T
tianxin04 已提交
106 107 108 109 110 111

        with fluid.program_guard(train_program, startup_prog):
            with fluid.unique_name.guard():
                train_pyreader, graph_vars = create_model(
                    args,
                    pyreader_name='train_reader',
T
tianxin 已提交
112 113 114 115
                    ernie_config=ernie_config,
                    is_classify=args.is_classify,
                    is_regression=args.is_regression)
                scheduled_lr, loss_scaling = optimization(
T
tianxin04 已提交
116 117 118 119 120 121 122 123
                    loss=graph_vars["loss"],
                    warmup_steps=warmup_steps,
                    num_train_steps=max_train_steps,
                    learning_rate=args.learning_rate,
                    train_program=train_program,
                    startup_prog=startup_prog,
                    weight_decay=args.weight_decay,
                    scheduler=args.lr_scheduler,
T
tianxin 已提交
124
                    use_fp16=args.use_fp16)
T
tianxin04 已提交
125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143

        if args.verbose:
            if args.in_tokens:
                lower_mem, upper_mem, unit = fluid.contrib.memory_usage(
                    program=train_program,
                    batch_size=args.batch_size // args.max_seq_len)
            else:
                lower_mem, upper_mem, unit = fluid.contrib.memory_usage(
                    program=train_program, batch_size=args.batch_size)
            print("Theoretical memory usage in training: %.3f - %.3f %s" %
                  (lower_mem, upper_mem, unit))

    if args.do_val or args.do_test:
        test_prog = fluid.Program()
        with fluid.program_guard(test_prog, startup_prog):
            with fluid.unique_name.guard():
                test_pyreader, graph_vars = create_model(
                    args,
                    pyreader_name='test_reader',
T
tianxin 已提交
144 145 146
                    ernie_config=ernie_config,
                    is_classify=args.is_classify,
                    is_regression=args.is_regression)
T
tianxin04 已提交
147 148

        test_prog = test_prog.clone(for_test=True)
T
tianxin 已提交
149 150
    nccl2_num_trainers = 1
    nccl2_trainer_id = 0
T
tianxin04 已提交
151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190
    exe.run(startup_prog)

    if args.do_train:
        if args.init_checkpoint and args.init_pretraining_params:
            print(
                "WARNING: args 'init_checkpoint' and 'init_pretraining_params' "
                "both are set! Only arg 'init_checkpoint' is made valid.")
        if args.init_checkpoint:
            init_checkpoint(
                exe,
                args.init_checkpoint,
                main_program=startup_prog,
                use_fp16=args.use_fp16)
        elif args.init_pretraining_params:
            init_pretraining_params(
                exe,
                args.init_pretraining_params,
                main_program=startup_prog,
                use_fp16=args.use_fp16)
    elif args.do_val or args.do_test:
        if not args.init_checkpoint:
            raise ValueError("args 'init_checkpoint' should be set if"
                             "only doing validation or testing!")
        init_checkpoint(
            exe,
            args.init_checkpoint,
            main_program=startup_prog,
            use_fp16=args.use_fp16)

    if args.do_train:
        exec_strategy = fluid.ExecutionStrategy()
        if args.use_fast_executor:
            exec_strategy.use_experimental_executor = True
        exec_strategy.num_threads = dev_count
        exec_strategy.num_iteration_per_drop_scope = args.num_iteration_per_drop_scope

        train_exe = fluid.ParallelExecutor(
            use_cuda=args.use_cuda,
            loss_name=graph_vars["loss"].name,
            exec_strategy=exec_strategy,
T
tianxin 已提交
191 192 193
            main_program=train_program,
            num_trainers=nccl2_num_trainers,
            trainer_id=nccl2_trainer_id)
T
tianxin04 已提交
194 195 196 197 198

        train_pyreader.decorate_tensor_provider(train_data_generator)
    else:
        train_exe = None

T
tianxin 已提交
199 200 201 202 203 204 205 206
    test_exe = exe
    if args.do_val or args.do_test:
        if args.use_multi_gpu_test:
            test_exe = fluid.ParallelExecutor(
                use_cuda=args.use_cuda,
                main_program=test_prog,
                share_vars_from=train_exe)

T
tianxin04 已提交
207 208 209 210 211 212
    if args.do_train:
        train_pyreader.start()
        steps = 0
        if warmup_steps > 0:
            graph_vars["learning_rate"] = scheduled_lr

Z
zhengya01 已提交
213
        ce_info = []
T
tianxin04 已提交
214
        time_begin = time.time()
T
tianxin 已提交
215 216
        last_epoch = 0
        current_epoch = 0
T
tianxin04 已提交
217 218 219 220 221 222
        while True:
            try:
                steps += 1
                if steps % args.skip_steps != 0:
                    train_exe.run(fetch_list=[])
                else:
T
tianxin 已提交
223 224 225 226 227 228 229 230 231
                    outputs = evaluate(
                        train_exe,
                        train_program,
                        train_pyreader,
                        graph_vars,
                        "train",
                        metric=args.metric,
                        is_classify=args.is_classify,
                        is_regression=args.is_regression)
T
tianxin04 已提交
232 233 234 235 236 237 238 239 240 241 242 243

                    if args.verbose:
                        verbose = "train pyreader queue size: %d, " % train_pyreader.queue.size(
                        )
                        verbose += "learning rate: %f" % (
                            outputs["learning_rate"]
                            if warmup_steps > 0 else args.learning_rate)
                        print(verbose)

                    current_example, current_epoch = reader.get_train_progress()
                    time_end = time.time()
                    used_time = time_end - time_begin
T
tianxin 已提交
244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260

                    if args.is_classify:
                        print(
                            "epoch: %d, progress: %d/%d, step: %d, ave loss: %f, "
                            "ave acc: %f, speed: %f steps/s" %
                            (current_epoch, current_example, num_train_examples,
                             steps, outputs["loss"], outputs["accuracy"],
                             args.skip_steps / used_time))
                        ce_info.append(
                            [outputs["loss"], outputs["accuracy"], used_time])
                    if args.is_regression:
                        print(
                            "epoch: %d, progress: %d/%d, step: %d, ave loss: %f, "
                            " speed: %f steps/s" %
                            (current_epoch, current_example, num_train_examples,
                             steps, outputs["loss"],
                             args.skip_steps / used_time))
T
tianxin04 已提交
261 262 263 264 265 266 267
                    time_begin = time.time()

                if steps % args.save_steps == 0:
                    save_path = os.path.join(args.checkpoints,
                                             "step_" + str(steps))
                    fluid.io.save_persistables(exe, save_path, train_program)

T
tianxin 已提交
268
                if steps % args.validation_steps == 0 or last_epoch != current_epoch:
T
tianxin04 已提交
269 270
                    # evaluate dev set
                    if args.do_val:
T
tianxin 已提交
271 272 273 274
                        evaluate_wrapper(args, reader, exe, test_prog,
                                         test_pyreader, graph_vars,
                                         current_epoch, steps)

T
tianxin04 已提交
275
                    if args.do_test:
T
tianxin 已提交
276 277 278 279 280 281 282
                        predict_wrapper(args, reader, exe, test_prog,
                                        test_pyreader, graph_vars,
                                        current_epoch, steps)

                if last_epoch != current_epoch:
                    last_epoch = current_epoch

T
tianxin04 已提交
283 284 285 286 287
            except fluid.core.EOFException:
                save_path = os.path.join(args.checkpoints, "step_" + str(steps))
                fluid.io.save_persistables(exe, save_path, train_program)
                train_pyreader.reset()
                break
Z
zhengya01 已提交
288 289 290 291 292 293 294 295 296 297 298
        if args.enable_ce:
            card_num = get_cards()
            ce_loss = 0
            ce_acc = 0
            ce_time = 0
            try:
                ce_loss = ce_info[-2][0]
                ce_acc = ce_info[-2][1]
                ce_time = ce_info[-2][2]
            except:
                print("ce info error")
T
tianxin 已提交
299 300 301
            print("kpis\ttrain_duration_card%s\t%s" % (card_num, ce_time))
            print("kpis\ttrain_loss_card%s\t%f" % (card_num, ce_loss))
            print("kpis\ttrain_acc_card%s\t%f" % (card_num, ce_acc))
T
tianxin04 已提交
302 303 304

    # final eval on dev set
    if args.do_val:
T
tianxin 已提交
305 306 307 308 309 310 311 312 313 314
        evaluate_wrapper(args, reader, exe, test_prog, test_pyreader,
                         graph_vars, current_epoch, steps)

    # final eval on test set
    if args.do_test:
        predict_wrapper(args, reader, exe, test_prog, test_pyreader, graph_vars,
                        current_epoch, steps)

    # final eval on dianostic, hack for glue-ax
    if args.diagnostic:
T
tianxin04 已提交
315 316
        test_pyreader.decorate_tensor_provider(
            reader.data_generator(
T
tianxin 已提交
317
                args.diagnostic,
T
format  
tianxin04 已提交
318 319
                batch_size=args.batch_size,
                epoch=1,
T
tianxin 已提交
320
                dev_count=1,
T
tianxin04 已提交
321 322
                shuffle=False))

T
tianxin 已提交
323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344
        print("Final diagnostic")
        qids, preds, probs = predict(
            test_exe,
            test_prog,
            test_pyreader,
            graph_vars,
            is_classify=args.is_classify,
            is_regression=args.is_regression)
        assert len(qids) == len(preds), '{} v.s. {}'.format(
            len(qids), len(preds))
        with open(args.diagnostic_save, 'w') as f:
            for id, s, p in zip(qids, preds, probs):
                f.write('{}\t{}\t{}\n'.format(id, s, p))

        print("Done final diagnostic, saving to {}".format(
            args.diagnostic_save))


def evaluate_wrapper(args, reader, exe, test_prog, test_pyreader, graph_vars,
                     epoch, steps):
    # evaluate dev set
    for ds in args.dev_set.split(','):
T
tianxin04 已提交
345 346
        test_pyreader.decorate_tensor_provider(
            reader.data_generator(
T
tianxin 已提交
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
                ds,
                batch_size=args.predict_batch_size,
                epoch=1,
                dev_count=1,
                shuffle=False))
        print("validation result of dataset {}:".format(ds))
        evaluate_info = evaluate(
            exe,
            test_prog,
            test_pyreader,
            graph_vars,
            "dev",
            metric=args.metric,
            is_classify=args.is_classify,
            is_regression=args.is_regression)
        print(evaluate_info + ', file: {}, epoch: {}, steps: {}'.format(
            ds, epoch, steps))


def predict_wrapper(args, reader, exe, test_prog, test_pyreader, graph_vars,
                    epoch, steps):
    test_sets = args.test_set.split(',')
    save_dirs = args.test_save.split(',')
    assert len(test_sets) == len(save_dirs)

    for test_f, save_f in zip(test_sets, save_dirs):
        test_pyreader.decorate_tensor_provider(
            reader.data_generator(
                test_f,
                batch_size=args.predict_batch_size,
T
tianxin04 已提交
377
                epoch=1,
T
tianxin 已提交
378
                dev_count=1,
T
tianxin04 已提交
379
                shuffle=False))
T
tianxin 已提交
380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397

        save_path = save_f + '.' + str(epoch) + '.' + str(steps)
        print("testing {}, save to {}".format(test_f, save_path))
        qids, preds, probs = predict(
            exe,
            test_prog,
            test_pyreader,
            graph_vars,
            is_classify=args.is_classify,
            is_regression=args.is_regression)

        save_dir = os.path.dirname(save_path)
        if not os.path.exists(save_dir):
            os.makedirs(save_dir)

        with open(save_path, 'w') as f:
            for id, s, p in zip(qids, preds, probs):
                f.write('{}\t{}\t{}\n'.format(id, s, p))
T
tianxin04 已提交
398 399 400 401


if __name__ == '__main__':
    print_arguments(args)
T
tianxin 已提交
402
    check_cuda(args.use_cuda)
T
tianxin04 已提交
403
    main(args)