SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

WIP examples

Browse Source
This commit is contained in:
thomwolf
2019-07-10 15:33:34 +02:00
parent ed6c8d37f4
commit 50b7e52a7f
5 changed files with 361 additions and 209 deletions
Download Patch File Download Diff File Expand all files Collapse all files

289
examples/run_squad.py
View File

@@ -33,36 +33,156 @@ from tqdm import tqdm, trange
from tensorboardX import SummaryWriter
from pytorch_transformers import WEIGHTS_NAME, CONFIG_NAME
from pytorch_transformers.modeling_bert import BertForQuestionAnswering
from pytorch_transformers.optimization import BertAdam, WarmupLinearSchedule
from pytorch_transformers.tokenization_bert import BertTokenizer
from pytorch_transformers import (BertForQuestionAnswering, XLNetForQuestionAnswering,
XLMForQuestionAnswering, BERT_PRETRAINED_MODEL_ARCHIVE_MAP,
XLNET_PRETRAINED_MODEL_ARCHIVE_MAP, XLM_PRETRAINED_MODEL_ARCHIVE_MAP)
from pytorch_transformers import (BertTokenizer, XLNetTokenizer,
XLMTokenizer)
from utils_squad import read_squad_examples, convert_examples_to_features, RawResult, write_predictions
if sys.version_info[0] == 2:
import cPickle as pickle
else:
import pickle
logger = logging.getLogger(__name__)
ALL_MODELS = sum((tuple(m.keys()) for m in (BERT_PRETRAINED_MODEL_ARCHIVE_MAP,
XLNET_PRETRAINED_MODEL_ARCHIVE_MAP,
XLM_PRETRAINED_MODEL_ARCHIVE_MAP)), ())
MODEL_CLASSES = {
'bert': BertForQuestionAnswering,
'xlnet': XLNetForQuestionAnswering,
'xlm': XLMForQuestionAnswering,
}
TOKENIZER_CLASSES = {
'bert': BertTokenizer,
'xlnet': XLNetTokenizer,
'xlm': XLMTokenizer,
}
def train(args, train_dataset, model):
""" Train the model """
if args.local_rank in [-1, 0]:
tb_writer = SummaryWriter()
args.train_batch_size = args.train_batch_size // args.gradient_accumulation_steps
train_sampler = RandomSampler(train_dataset) if args.local_rank == -1 else DistributedSampler(train_dataset)
train_dataloader = DataLoader(train_dataset, sampler=train_sampler, batch_size=args.train_batch_size)
if args.max_steps > 0:
num_train_optimization_steps = args.max_steps
args.num_train_epochs = args.max_steps // (len(train_dataloader) // args.gradient_accumulation_steps) + 1
else:
num_train_optimization_steps = len(train_dataloader) // args.gradient_accumulation_steps * args.num_train_epochs
# Prepare optimizer
no_decay = ['bias', 'LayerNorm.weight']
optimizer_grouped_parameters = [
{'params': [p for n, p in model.named_parameters() if not any(nd in n for nd in no_decay)], 'weight_decay': 0.01},
{'params': [p for n, p in model.named_parameters() if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}
]
optimizer = BertAdam(optimizer_grouped_parameters, lr=args.learning_rate,
t_total=num_train_optimization_steps, warmup=args.warmup_proportion)
if args.fp16:
try:
from apex import amp
except ImportError:
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use fp16 training.")
model, optimizer = amp.initialize(model, optimizer, opt_level=args.fp16_opt_level)
# Train!
logger.info("***** Running training *****")
logger.info(" Num examples = %d", len(train_dataset))
logger.info(" Num Epochs = %d", args.num_train_epochs)
logger.info(" Batch size = %d", args.train_batch_size)
logger.info(" Total batch size (distributed) = %d", args.train_batch_size * (torch.distributed.get_world_size() if args.local_rank != -1 else 1))
logger.info(" Gradient Accumulation steps = %d", args.gradient_accumulation_steps)
logger.info(" Total optimization steps = %d", num_train_optimization_steps)
global_step = 0
tr_loss, logging_loss = 0.0, 0.0
model.train()
optimizer.zero_grad()
for _ in trange(int(args.num_train_epochs), desc="Epoch", disable=args.local_rank not in [-1, 0]):
for step, batch in enumerate(tqdm(train_dataloader, desc="Iteration", disable=args.local_rank not in [-1, 0])):
batch = tuple(t.to(args.device) for t in batch)
inputs = {'input_ids': batch[0],
'attention_mask': batch[1],
'token_type_ids': batch[2] if args.model_type in ['bert', 'xlnet'] else None, # XLM don't use segment_ids
'labels': batch[3]}
ouputs = model(**inputs)
loss = ouputs[0]
def evalutate(args, dataset, model):
""" Evaluate the model """
def load_and_cache_examples(args, tokenizer, training=True):
""" Load data features from cache or dataset file. """
cached_features_file = os.path.join(args.data_dir, 'cached_{}_{}_{}_{}'.format(
'dev' if evaluate else 'train',
list(filter(None, args.model_name.split('/'))).pop(),
str(args.max_seq_length),
str(task)))
if os.path.exists(cached_features_file):
logger.info("Loading features from cached file %s", cached_features_file)
features = torch.load(cached_features_file)
else:
logger.info("Creating features from dataset file at %s", args.data_dir)
label_list = processor.get_labels()
examples = read_squad_examples(input_file=args.train_file if training else args.predict_file,
is_training=training,
version_2_with_negative=args.version_2_with_negative)
features = convert_examples_to_features(
examples=examples,
tokenizer=tokenizer,
max_seq_length=args.max_seq_length,
doc_stride=args.doc_stride,
max_query_length=args.max_query_length,
is_training=training)
if args.local_rank in [-1, 0]:
logger.info("Num orig examples = %d", len(examples))
logger.info("Num split examples = %d", len(features))
logger.info("Saving features into cached file %s", cached_features_file)
torch.save(features, cached_features_file)
# Convert to Tensors and build dataset
all_input_ids = torch.tensor([f.input_ids for f in eval_features], dtype=torch.long)
all_input_mask = torch.tensor([f.input_mask for f in eval_features], dtype=torch.long)
all_segment_ids = torch.tensor([f.segment_ids for f in eval_features], dtype=torch.long)
if training:
all_start_positions = torch.tensor([f.start_position for f in train_features], dtype=torch.long)
all_end_positions = torch.tensor([f.end_position for f in train_features], dtype=torch.long)
dataset = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_start_positions, all_end_positions)
else:
all_example_index = torch.arange(all_input_ids.size(0), dtype=torch.long)
dataset = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_example_index)
return dataset
def main():
parser = argparse.ArgumentParser()
## Required parameters
parser.add_argument("--bert_model", default=None, type=str, required=True,
help="Bert pre-trained model selected in the list: bert-base-uncased, "
"bert-large-uncased, bert-base-cased, bert-large-cased, bert-base-multilingual-uncased, "
"bert-base-multilingual-cased, bert-base-chinese.")
parser.add_argument("--train_file", default=None, type=str, required=True,
help="SQuAD json for training. E.g., train-v1.1.json")
parser.add_argument("--predict_file", default=None, type=str, required=True,
help="SQuAD json for predictions. E.g., dev-v1.1.json or test-v1.1.json")
parser.add_argument("--model_name", default=None, type=str, required=True,
help="Bert/XLNet/XLM pre-trained model selected in the list: " + ", ".join(ALL_MODELS))
parser.add_argument("--output_dir", default=None, type=str, required=True,
help="The output directory where the model checkpoints and predictions will be written.")
## Other parameters
parser.add_argument("--train_file", default=None, type=str, help="SQuAD json for training. E.g., train-v1.1.json")
parser.add_argument("--predict_file", default=None, type=str,
help="SQuAD json for predictions. E.g., dev-v1.1.json or test-v1.1.json")
parser.add_argument('--version_2_with_negative', action='store_true',
help='If true, the SQuAD examples contain some that do not have an answer.')
parser.add_argument('--null_score_diff_threshold', type=float, default=0.0,
help="If null_score - best_non_null is greater than the threshold predict null.")
parser.add_argument('--overwrite_output_dir', action='store_true',
help="Overwrite the content of the output directory")
parser.add_argument("--max_seq_length", default=384, type=int,
help="The maximum total input sequence length after WordPiece tokenization. Sequences "
"longer than this will be truncated, and sequences shorter than this will be padded.")
@@ -71,65 +191,53 @@ def main():
parser.add_argument("--max_query_length", default=64, type=int,
help="The maximum number of tokens for the question. Questions longer than this will "
"be truncated to this length.")
parser.add_argument("--do_train", action='store_true', help="Whether to run training.")
parser.add_argument("--do_predict", action='store_true', help="Whether to run eval on the dev set.")
parser.add_argument("--train_batch_size", default=32, type=int, help="Total batch size for training.")
parser.add_argument("--predict_batch_size", default=8, type=int, help="Total batch size for predictions.")
parser.add_argument("--learning_rate", default=5e-5, type=float, help="The initial learning rate for Adam.")
parser.add_argument("--do_train", action='store_true',
help="Whether to run training.")
parser.add_argument("--do_predict", action='store_true',
help="Whether to run eval on the dev set.")
parser.add_argument("--do_lower_case", action='store_true',
help="Whether to lower case the input text. True for uncased models, False for cased models.")
parser.add_argument("--train_batch_size", default=32, type=int,
help="Total batch size for training.")
parser.add_argument("--predict_batch_size", default=8, type=int,
help="Total batch size for predictions.")
parser.add_argument("--learning_rate", default=5e-5, type=float,
help="The initial learning rate for Adam.")
parser.add_argument('--gradient_accumulation_steps', type=int, default=1,
help="Number of updates steps to accumulate before performing a backward/update pass.")
parser.add_argument("--num_train_epochs", default=3.0, type=float,
help="Total number of training epochs to perform.")
parser.add_argument("--warmup_proportion", default=0.1, type=float,
help="Proportion of training to perform linear learning rate warmup for. E.g., 0.1 = 10%% "
"of training.")
help="Proportion of training with linear learning rate warmup (0.1 = 10%% of training).")
parser.add_argument("--n_best_size", default=20, type=int,
help="The total number of n-best predictions to generate in the nbest_predictions.json "
"output file.")
help="The total number of n-best predictions to generate in the nbest_predictions.json output file.")
parser.add_argument("--max_answer_length", default=30, type=int,
help="The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another.")
parser.add_argument("--verbose_logging", action='store_true',
help="If true, all of the warnings related to data processing will be printed. "
"A number of warnings are expected for a normal SQuAD evaluation.")
parser.add_argument("--no_cuda",
action='store_true',
parser.add_argument("--no_cuda", action='store_true',
help="Whether not to use CUDA when available")
parser.add_argument('--seed',
type=int,
default=42,
parser.add_argument('--seed', type=int, default=42,
help="random seed for initialization")
parser.add_argument('--gradient_accumulation_steps',
type=int,
default=1,
help="Number of updates steps to accumulate before performing a backward/update pass.")
parser.add_argument("--do_lower_case",
action='store_true',
help="Whether to lower case the input text. True for uncased models, False for cased models.")
parser.add_argument("--local_rank",
type=int,
default=-1,
parser.add_argument("--local_rank", type=int, default=-1,
help="local_rank for distributed training on gpus")
parser.add_argument('--fp16',
action='store_true',
help="Whether to use 16-bit float precision instead of 32-bit")
parser.add_argument('--overwrite_output_dir',
action='store_true',
help="Overwrite the content of the output directory")
parser.add_argument('--loss_scale',
type=float, default=0,
help="Loss scaling to improve fp16 numeric stability. Only used when fp16 set to True.\n"
"0 (default value): dynamic loss scaling.\n"
"Positive power of 2: static loss scaling value.\n")
parser.add_argument('--version_2_with_negative',
action='store_true',
help='If true, the SQuAD examples contain some that do not have an answer.')
parser.add_argument('--null_score_diff_threshold',
type=float, default=0.0,
help="If null_score - best_non_null is greater than the threshold predict null.")
parser.add_argument('--fp16', action='store_true',
help="Whether to use 16-bit (mixed) precision (through NVIDIA apex) instead of 32-bit")
parser.add_argument('--fp16_opt_level', type=str, default='O1',
help="For fp16: Apex AMP optimization level selected in ['O0', 'O1', 'O2', and 'O3']."
"See details at https://nvidia.github.io/apex/amp.html")
parser.add_argument('--server_ip', type=str, default='', help="Can be used for distant debugging.")
parser.add_argument('--server_port', type=str, default='', help="Can be used for distant debugging.")
args = parser.parse_args()
print(args)
if os.path.exists(args.output_dir) and os.listdir(args.output_dir) and args.do_train and not args.overwrite_output_dir:
raise ValueError("Output directory ({}) already exists and is not empty. Use --overwrite_output_dir to overcome.".format(args.output_dir))
if args.server_ip and args.server_port:
# Distant debugging - see https://code.visualstudio.com/docs/python/debugging#_attach-to-a-local-script
import ptvsd
@@ -137,71 +245,52 @@ def main():
ptvsd.enable_attach(address=(args.server_ip, args.server_port), redirect_output=True)
ptvsd.wait_for_attach()
# Setup CUDA, GPU & distributed training
if args.local_rank == -1 or args.no_cuda:
device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
n_gpu = torch.cuda.device_count()
else:
args.n_gpu = torch.cuda.device_count()
else: # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
torch.cuda.set_device(args.local_rank)
device = torch.device("cuda", args.local_rank)
n_gpu = 1
# Initializes the distributed backend which will take care of sychronizing nodes/GPUs
torch.distributed.init_process_group(backend='nccl')
args.n_gpu = 1
args.device = device
logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s - %(message)s',
datefmt = '%m/%d/%Y %H:%M:%S',
level = logging.INFO if args.local_rank in [-1, 0] else logging.WARN)
logger.info("device: {} n_gpu: {}, distributed training: {}, 16-bits training: {}".format(
device, n_gpu, bool(args.local_rank != -1), args.fp16))
if args.gradient_accumulation_steps < 1:
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, should be >= 1".format(
args.gradient_accumulation_steps))
args.train_batch_size = args.train_batch_size // args.gradient_accumulation_steps
# Setup logging
logging.basicConfig(level = logging.INFO if args.local_rank in [-1, 0] else logging.WARN)
logger.warning("Process rank: %s, device: %s, n_gpu: %s, distributed training: %s, 16-bits training: %s",
args.local_rank, device, args.n_gpu, bool(args.local_rank != -1), args.fp16)
# Setup seeds
random.seed(args.seed)
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if n_gpu > 0:
if args.n_gpu > 0:
torch.cuda.manual_seed_all(args.seed)
if not args.do_train and not args.do_predict:
raise ValueError("At least one of `do_train` or `do_predict` must be True.")
if args.do_train:
if not args.train_file:
raise ValueError(
"If `do_train` is True, then `train_file` must be specified.")
if args.do_predict:
if not args.predict_file:
raise ValueError(
"If `do_predict` is True, then `predict_file` must be specified.")
if os.path.exists(args.output_dir) and os.listdir(args.output_dir) and args.do_train and not args.overwrite_output_dir:
raise ValueError("Output directory {} already exists and is not empty. Use --overwrite_output_dir to overcome.".format(args.output_dir))
if not os.path.exists(args.output_dir):
os.makedirs(args.output_dir)
# Load pretrained model and tokenizer
if args.local_rank not in [-1, 0]:
torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab
torch.distributed.barrier() # Make sure only 1st process in distributed training download model & vocab
args.model_type = args.model_name.lower().split('-')[0]
tokenizer_class = TOKENIZER_CLASSES[args.model_type]
model_class = MODEL_CLASSES[args.model_type]
tokenizer = tokenizer_class.from_pretrained(args.model_name, do_lower_case=args.do_lower_case)
model = model_class.from_pretrained(args.model_name, num_labels=num_labels)
tokenizer = BertTokenizer.from_pretrained(args.bert_model, do_lower_case=args.do_lower_case)
model = BertForQuestionAnswering.from_pretrained(args.bert_model)
if args.local_rank == 0:
torch.distributed.barrier()
if args.fp16:
model.half()
model.to(device)
# Distributed and parrallel training
model.to(args.device)
if args.local_rank != -1:
model = torch.nn.parallel.DistributedDataParallel(model,
device_ids=[args.local_rank],
model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.local_rank],
output_device=args.local_rank,
find_unused_parameters=True)
elif n_gpu > 1:
elif args.n_gpu > 1:
model = torch.nn.DataParallel(model)
# Training
if args.do_train:
if args.local_rank in [-1, 0]:
tb_writer = SummaryWriter()