nlp：bert
run_boolq_bert.py
CUDA_VISIBLE_DEVICES=1 python run_boolq_bert.py --model_type bert --model_name_or_path bert-base-cased --do_train --do_eval --evaluate_during_training --do_lower_case --train_file train.jsonl --predict_file val.jsonl --test_file test.jsonl --tbname boolq_bert --max_seq_length 256 --learning_rate 1e-5 --num_train_epochs 1000.0 --logging_steps 500 --per_gpu_eval_batch_size=8 --per_gpu_train_batch_size=8 --output_dir boolq_bert_outputimport json
import argparse
import csv
import glob
import logging
import os
import random
import numpy as np
import torch
from torch.utils.data import DataLoader, RandomSampler, SequentialSampler, TensorDataset
from torch.utils.data.distributed import DistributedSampler
from tqdm import tqdm, trange
from transformers import (
WEIGHTS_NAME,
AdamW,
BertConfig,
BertForSequenceClassification,
BertTokenizer,
get_linear_schedule_with_warmup,
)
try:
from torch.utils.tensorboard import SummaryWriter
except ImportError:
from tensorboardX import SummaryWriter
logger = logging.getLogger(__name__)
MODEL_CLASSES = {
"bert": (BertConfig, BertForSequenceClassification, BertTokenizer),
}
class SwagExample(object):
"""A single training/test example for the SWAG dataset."""
def __init__(self, swag_id, context_sentence, start_ending, ending_0, ending_1, ending_2, ending_3, label=None):
self.swag_id = swag_id
self.context_sentence = context_sentence
self.start_ending = start_ending
self.endings = [
ending_0,
ending_1,
ending_2,
ending_3,
]
self.label = label
def __str__(self):
return self.__repr__()
def __repr__(self):
attributes = [
"swag_id: {}".format(self.swag_id),
"context_sentence: {}".format(self.context_sentence),
"start_ending: {}".format(self.start_ending),
"ending_0: {}".format(self.endings[0]),
"ending_1: {}".format(self.endings[1]),
"ending_2: {}".format(self.endings[2]),
"ending_3: {}".format(self.endings[3]),
]
if self.label is not None:
attributes.append("label: {}".format(self.label))
return ", ".join(attributes)
class BoolqExample(object):
"""A single training/test example for the Boolq dataset."""
def __init__(self, swag_id, context_sentence, start_ending, ending_0, label=None):
self.swag_id = swag_id
self.context_sentence = context_sentence
self.start_ending = start_ending
self.endings = [
ending_0
# ending_1,
]
self.label = label
def __str__(self):
return self.__repr__()
def __repr__(self):
attributes = [
"swag_id: {}".format(self.swag_id),
"context_sentence: {}".format(self.context_sentence),
"start_ending: {}".format(self.start_ending),
"ending_0: {}".format(self.endings[0]),
#"ending_1: {}".format(self.endings[1]),
]
if self.label is not None:
attributes.append("label: {}".format(self.label))
return ", ".join(attributes)
class MultiRCExample(object):
"""A single training/test example for the Boolq dataset."""
def __init__(self, swag_id, context_sentence, start_ending, ending_0, label=None):
self.swag_id = swag_id
self.context_sentence = context_sentence
self.start_ending = start_ending
self.endings = [
ending_0
# ending_1,
]
self.label = label
def __str__(self):
return self.__repr__()
def __repr__(self):
attributes = [
"swag_id: {}".format(self.swag_id),
"context_sentence: {}".format(self.context_sentence),
"start_ending: {}".format(self.start_ending),
"ending_0: {}".format(self.endings[0]),
#"ending_1: {}".format(self.endings[1]),
]
if self.label is not None:
attributes.append("label: {}".format(self.label))
return ", ".join(attributes)
'''class InputFeatures(object):
def __init__(self, example_id, choices_features, label):
self.example_id = example_id
self.choices_features = [
{"input_ids": input_ids, "input_mask": input_mask, "segment_ids": segment_ids}
for _, input_ids, input_mask, segment_ids in choices_features
]
self.label = label'''
class InputFeatures(object):
"""
A single set of features of data.
Args:
input_ids: Indices of input sequence tokens in the vocabulary.
attention_mask: Mask to avoid performing attention on padding token indices.
Mask values selected in ``[0, 1]``:
Usually ``1`` for tokens that are NOT MASKED, ``0`` for MASKED (padded) tokens.
token_type_ids: Segment token indices to indicate first and second portions of the inputs.
label: Label corresponding to the input
"""
def __init__(self, input_ids, attention_mask=None, token_type_ids=None, label=None):
self.input_ids = input_ids
self.attention_mask = attention_mask
self.token_type_ids = token_type_ids
self.label = label
def read_swag_examples(input_file, is_training=True):
with open(input_file, "r", encoding="utf-8") as f:
lines = list(csv.reader(f))
if is_training and lines[0][-1] != "label":
raise ValueError("For training, the input file must contain a label column.")
examples = [
SwagExample(
swag_id=line[2],
context_sentence=line[4],
start_ending=line[5], # in the swag dataset, the
# common beginning of each
# choice is stored in "sent2".
ending_0=line[7],
ending_1=line[8],
ending_2=line[9],
ending_3=line[10],
label=int(line[11]) if is_training else None,
)
for line in lines[1:] # we skip the line with the column names
]
return examples
def read_boolq_examples(input_file, is_training=True):
with open(input_file, "r", encoding="utf-8") as f:
lines = f.readlines()
examples = []
for line in lines:
data_raw = json.loads(line.strip("\n"))
if data_raw["label"]:
label_input = 1
else:
label_input = 0
examples.append(
BoolqExample(
swag_id=data_raw["idx"],
context_sentence=data_raw["passage"],
start_ending=data_raw["question"],
ending_0="",
label=label_input if is_training else None,
)
)
return examples
def convert_examples_to_features(examples, tokenizer, max_seq_length, is_training):
"""Loads a data file into a list of `InputBatch`s."""
# Swag is a multiple choice task. To perform this task using Bert,
# we will use the formatting proposed in "Improving Language
# Understanding by Generative Pre-Training" and suggested by
# @jacobdevlin-google in this issue
# https://github.com/google-research/bert/issues/38.
#
# Each choice will correspond to a sample on which we run the
# inference. For a given Swag example, we will create the 4
# following inputs:
# - [CLS] context [SEP] choice_1 [SEP]
# - [CLS] context [SEP] choice_2 [SEP]
# - [CLS] context [SEP] choice_3 [SEP]
# - [CLS] context [SEP] choice_4 [SEP]
# The model will output a single value for each input. To get the
# final decision of the model, we will run a softmax over these 4
# outputs.
features = []
for example_index, example in tqdm(enumerate(examples)):
context_tokens = tokenizer.tokenize(example.context_sentence)
start_ending_tokens = tokenizer.tokenize(example.start_ending)
context_tokens_choice = context_tokens[:]
ending_tokens = start_ending_tokens
_truncate_seq_pair(context_tokens_choice, ending_tokens, max_seq_length - 3)
tokens = ["[CLS]"] + context_tokens_choice + ["[SEP]"] + ending_tokens + ["[SEP]"]
segment_ids = [0] * (len(context_tokens_choice) + 2) + [1] * (len(ending_tokens) + 1)
input_ids = tokenizer.convert_tokens_to_ids(tokens)
input_mask = [1] * len(input_ids)
# Zero-pad up to the sequence length.
padding = [0] * (max_seq_length - len(input_ids))
input_ids += padding
input_mask += padding
segment_ids += padding
assert len(input_ids) == max_seq_length
assert len(input_mask) == max_seq_length
assert len(segment_ids) == max_seq_length
label = example.label
if example_index < 5:
logger.info("*** Example ***")
logger.info("swag_id: {}".format(example.swag_id))
logger.info("tokens: {}".format(" ".join(tokens)))
logger.info("input_ids: {}".format(" ".join(map(str, input_ids))))
logger.info("attention_mask: {}".format(" ".join(map(str, input_mask))))
logger.info("token_type_ids: {}".format(" ".join(map(str, segment_ids))))
if is_training:
logger.info("label: {}".format(label))
features.append(InputFeatures(input_ids = input_ids, attention_mask = input_mask, token_type_ids = segment_ids, label = label ))
return features
def _truncate_seq_pair(tokens_a, tokens_b, max_length):
"""Truncates a sequence pair in place to the maximum length."""
# This is a simple heuristic which will always truncate the longer sequence
# one token at a time. This makes more sense than truncating an equal percent
# of tokens from each, since if one sequence is very short then each token
# that's truncated likely contains more information than a longer sequence.
while True:
total_length = len(tokens_a) + len(tokens_b)
if total_length <= max_length:
break
if len(tokens_a) > len(tokens_b):
tokens_a.pop()
else:
tokens_b.pop()
'''def accuracy(out, labels):
outputs = np.argmax(out, axis=1)
return np.sum(outputs == labels)'''
'''def accuracy(out, labels):
outputs = (out >= 0.5).astype(np.int)
return np.sum(outputs == labels)'''
def accuracy(out, labels):
outputs = (out.squeeze(1) >= 0.5).astype(np.int)
return np.sum(outputs == labels)
def select_field(features, field):
return [[choice[field] for choice in feature.choices_features] for feature in features]
def set_seed(args):
random.seed(args.seed)
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.n_gpu > 0:
torch.cuda.manual_seed_all(args.seed)
def load_and_cache_examples(args, tokenizer, train = False, evaluate=False, test = False, output_examples=False):
if args.local_rank not in [-1, 0]:
torch.distributed.barrier() # Make sure only the first process in distributed training process the dataset, and the others will use the cache
# Load data features from cache or dataset file
# input_file = args.predict_file if evaluate else args.train_file
if train:
input_file = args.train_file
if evaluate:
input_file = args.predict_file
if test:
input_file = args.test_file
cached_features_file = os.path.join(
os.path.dirname(input_file),
"cached_{}_{}_{}".format(
"dev" if evaluate else "train",
list(filter(None, args.model_name_or_path.split("/"))).pop(),
str(args.max_seq_length),
),
)
if os.path.exists(cached_features_file) and not args.overwrite_cache and not output_examples:
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", input_file)
examples = read_boolq_examples(input_file)
features = convert_examples_to_features(examples, tokenizer, args.max_seq_length, not evaluate)
if args.local_rank in [-1, 0]:
logger.info("Saving features into cached file %s", cached_features_file)
torch.save(features, cached_features_file)
if args.local_rank == 0:
torch.distributed.barrier() # Make sure only the first process in distributed training process the dataset, and the others will use the cache
# Convert to Tensors and build dataset
# all_input_ids = torch.tensor(select_field(features, "input_ids"), dtype=torch.long)
# all_input_mask = torch.tensor(select_field(features, "input_mask"), dtype=torch.long)
# all_segment_ids = torch.tensor(select_field(features, "segment_ids"), dtype=torch.long)
# all_label = torch.tensor([f.label for f in features], dtype=torch.long)
all_input_ids = torch.tensor([f.input_ids for f in features], dtype=torch.long)
all_input_mask = torch.tensor([f.attention_mask for f in features], dtype=torch.long)
all_segment_ids = torch.tensor([f.token_type_ids for f in features], dtype=torch.long)
all_label = torch.tensor([f.label for f in features], dtype=torch.long)
if evaluate:
dataset = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label)
else:
dataset = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label)
if output_examples:
return dataset, examples, features
return dataset
def train(args, train_dataset, model, tokenizer):
""" Train the model """
'''if args.local_rank in [-1, 0]:
tb_writer = SummaryWriter()'''
if args.local_rank in [-1, 0]:
tb_writer = SummaryWriter(args.tbname)
args.train_batch_size = args.per_gpu_train_batch_size * max(1, args.n_gpu)
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)
print("len(train_dataset): ",len(train_dataset))
if args.max_steps > 0:
t_total = args.max_steps
args.num_train_epochs = args.max_steps // (len(train_dataloader) // args.gradient_accumulation_steps) + 1
else:
t_total = len(train_dataloader) // args.gradient_accumulation_steps * args.num_train_epochs
# Prepare optimizer and schedule (linear warmup and decay)
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": args.weight_decay,
},
{"params": [p for n, p in model.named_parameters() if any(nd in n for nd in no_decay)], "weight_decay": 0.0},
]
optimizer = AdamW(optimizer_grouped_parameters, lr=args.learning_rate, eps=args.adam_epsilon)
scheduler = get_linear_schedule_with_warmup(
optimizer, num_warmup_steps=args.warmup_steps, num_training_steps=t_total
)
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)
# multi-gpu training (should be after apex fp16 initialization)
if args.n_gpu > 1:
model = torch.nn.DataParallel(model)
# Distributed training (should be after apex fp16 initialization)
if args.local_rank != -1:
model = torch.nn.parallel.DistributedDataParallel(
model, device_ids=[args.local_rank], output_device=args.local_rank, find_unused_parameters=True
)
# Train!
logger.info("***** Running training *****")
logger.info(" Num examples = %d", len(train_dataset))
logger.info(" Num Epochs = %d", args.num_train_epochs)
logger.info(" Instantaneous batch size per GPU = %d", args.per_gpu_train_batch_size)
logger.info(
" Total train batch size (w. parallel, distributed & accumulation) = %d",
args.train_batch_size
* args.gradient_accumulation_steps
* (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", t_total)
global_step = 0
tr_loss, logging_loss = 0.0, 0.0
model.zero_grad()
train_iterator = trange(int(args.num_train_epochs), desc="Epoch", disable=args.local_rank not in [-1, 0])
set_seed(args) # Added here for reproductibility
for _ in train_iterator:
epoch_iterator = tqdm(train_dataloader, desc="Iteration", disable=args.local_rank not in [-1, 0])
print("len(train_dataloader): ",len(train_dataloader))
for step, batch in enumerate(epoch_iterator):
model.train()
batch = tuple(t.to(args.device) for t in batch)
inputs = {
"input_ids": batch[0],
"attention_mask": batch[1],
# 'token_type_ids': None if args.model_type == 'xlm' else batch[2],
"token_type_ids": batch[2],
"labels": batch[3].float(),
}
# if args.model_type in ['xlnet', 'xlm']:
# inputs.update({'cls_index': batch[5],
# 'p_mask': batch[6]})
outputs = model(**inputs)
'''logit = outputs[1]
print(logit)
print(logit.shape)
print(batch[3])
print(batch[3].shape)'''
'''logit = outputs[1]
print("logit", logit)
print("logit.shape", logit.shape)
print("batch[3]", batch[3])
print("batch[3].shape", batch[3].shape)
print("batch[3].type", type(batch[3]))'''
loss = outputs[0] # model outputs are always tuple in transformers (see doc)
if args.n_gpu > 1:
loss = loss.mean() # mean() to average on multi-gpu parallel (not distributed) training
if args.gradient_accumulation_steps > 1:
loss = loss / args.gradient_accumulation_steps
if args.fp16:
with amp.scale_loss(loss, optimizer) as scaled_loss:
scaled_loss.backward()
torch.nn.utils.clip_grad_norm_(amp.master_params(optimizer), args.max_grad_norm)
else:
loss.backward()
torch.nn.utils.clip_grad_norm_(model.parameters(), args.max_grad_norm)
tr_loss += loss.item()
if (step + 1) % args.gradient_accumulation_steps == 0:
optimizer.step()
scheduler.step() # Update learning rate schedule
model.zero_grad()
global_step += 1
if args.local_rank in [-1, 0] and args.logging_steps > 0 and global_step % args.logging_steps == 0:
# Log metrics
#if (args.local_rank == -1 and args.evaluate_during_training):
if args.evaluate_during_training:
eval = True
test = False
results = evaluate(eval, test, args, model, tokenizer)
for key, value in results.items():
tb_writer.add_scalar("eval_{}".format(key), value, global_step)
# Save model checkpoint
output_dir = os.path.join(args.output_dir, "checkpoint-{}".format(global_step))
if not os.path.exists(output_dir):
os.makedirs(output_dir)
model_to_save = (
model.module if hasattr(model, "module") else model
) # Take care of distributed/parallel training
model_to_save.save_pretrained(output_dir)
tokenizer.save_vocabulary(output_dir)
torch.save(args, os.path.join(output_dir, "training_args.bin"))
logger.info("Saving model checkpoint to %s", output_dir)
tb_writer.add_scalar("lr", scheduler.get_lr()[0], global_step)
tb_writer.add_scalar("loss", (tr_loss - logging_loss) / args.logging_steps, global_step)
logging_loss = tr_loss
'''if args.local_rank in [-1, 0] and args.save_steps > 0 and global_step % args.save_steps == 0:
# Save model checkpoint
output_dir = os.path.join(args.output_dir, "checkpoint-{}".format(global_step))
if not os.path.exists(output_dir):
os.makedirs(output_dir)
model_to_save = (
model.module if hasattr(model, "module") else model
) # Take care of distributed/parallel training
model_to_save.save_pretrained(output_dir)
tokenizer.save_vocabulary(output_dir)
torch.save(args, os.path.join(output_dir, "training_args.bin"))
logger.info("Saving model checkpoint to %s", output_dir)'''
if args.max_steps > 0 and global_step > args.max_steps:
epoch_iterator.close()
break
if args.max_steps > 0 and global_step > args.max_steps:
train_iterator.close()
break
if args.local_rank in [-1, 0]:
tb_writer.close()
return global_step, tr_loss / global_step
# train = False, evaluate=False, test = False,
def evaluate(eval, test, args, model, tokenizer, prefix=""):
if eval :
dataset, examples, features = load_and_cache_examples(args, tokenizer, train=False, evaluate=True, test=False,
output_examples=True)
if test :
dataset, examples, features = load_and_cache_examples(args, tokenizer, train=False, evaluate=False, test=True,
output_examples=True)
# train = False, evaluate=False, test = False,
if not os.path.exists(args.output_dir) and args.local_rank in [-1, 0]:
os.makedirs(args.output_dir)
args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu)
# Note that DistributedSampler samples randomly
eval_sampler = SequentialSampler(dataset) if args.local_rank == -1 else DistributedSampler(dataset)
eval_dataloader = DataLoader(dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
# Eval!
logger.info("***** Running evaluation {} *****".format(prefix))
logger.info(" Num examples = %d", len(dataset))
logger.info(" Batch size = %d", args.eval_batch_size)
eval_loss, eval_accuracy = 0, 0
nb_eval_steps, nb_eval_examples = 0, 0
for batch in tqdm(eval_dataloader, desc="Evaluating"):
model.eval()
batch = tuple(t.to(args.device) for t in batch)
with torch.no_grad():
inputs = {
"input_ids": batch[0],
"attention_mask": batch[1],
# 'token_type_ids': None if args.model_type == 'xlm' else batch[2] # XLM don't use segment_ids
"token_type_ids": batch[2],
"labels": batch[3].float(),
}
# if args.model_type in ['xlnet', 'xlm']:
# inputs.update({'cls_index': batch[4],
# 'p_mask': batch[5]})
outputs = model(**inputs)
tmp_eval_loss, logits = outputs[:2]
eval_loss += tmp_eval_loss.mean().item()
logits = logits.detach().cpu().numpy()
label_ids = inputs["labels"].to("cpu").numpy()
tmp_eval_accuracy = accuracy(logits, label_ids)
eval_accuracy += tmp_eval_accuracy
nb_eval_steps += 1
nb_eval_examples += inputs["input_ids"].size(0)
'''print("inputs[\"input_ids\"]: ", inputs["input_ids"])
print("inputs[\"input_ids\"].shape: ", inputs["input_ids"].shape)'''
eval_loss = eval_loss / nb_eval_steps
eval_accuracy = eval_accuracy / nb_eval_examples
result = {"eval_loss": eval_loss, "eval_accuracy": eval_accuracy}
output_eval_file = os.path.join(args.output_dir, "eval_results.txt")
with open(output_eval_file, "w") as writer:
logger.info("***** Eval results *****")
for key in sorted(result.keys()):
logger.info("%s = %s", key, str(result[key]))
writer.write("%s = %s\n" % (key, str(result[key])))
return result
def main():
parser = argparse.ArgumentParser()
# Required parameters
parser.add_argument(
"--train_file",
default=None,
type=str,
required=True,
help="SWAG csv for training. E.g., train.csv"
)
parser.add_argument(
"--predict_file",
default=None,
type=str,
required=True,
help="SWAG csv for predictions. E.g., val.csv",
)
parser.add_argument(
"--test_file",
default=None,
type=str,
required=True,
help="SWAG csv for test. E.g., test.csv",
)
parser.add_argument(
"--tbname",
default=None,
type=str,
required=True,
help="tbname"
)
parser.add_argument(
"--model_type",
default=None,
type=str,
required=True,
help="Model type selected in the list: " + ", ".join(MODEL_CLASSES.keys()),
)
parser.add_argument(
"--model_name_or_path",
default=None,
type=str,
required=True,
help="Path to pre-trained model or shortcut name selected in the list: " + ", ",
)
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(
"--config_name", default="", type=str, help="Pretrained config name or path if not the same as model_name"
)
parser.add_argument(
"--tokenizer_name",
default="",
type=str,
help="Pretrained tokenizer name or path if not the same as model_name",
)
parser.add_argument(
"--max_seq_length",
default=384,
type=int,
help="The maximum total input sequence length after tokenization. Sequences "
"longer than this will be truncated, and sequences shorter than this will be padded.",
)
parser.add_argument("--do_train", action="store_true", help="Whether to run training.")
parser.add_argument("--do_eval", action="store_true", help="Whether to run eval on the dev set.")
parser.add_argument(
"--evaluate_during_training", action="store_true", help="Rul evaluation during training at each logging step."
)
parser.add_argument(
"--do_lower_case", action="store_true", help="Set this flag if you are using an uncased model."
)
parser.add_argument("--per_gpu_train_batch_size", default=8, type=int, help="Batch size per GPU/CPU for training.")
parser.add_argument(
"--per_gpu_eval_batch_size", default=8, type=int, help="Batch size per GPU/CPU for evaluation."
)
parser.add_argument("--learning_rate", default=1e-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("--weight_decay", default=0.0, type=float, help="Weight deay if we apply some.")
parser.add_argument("--adam_epsilon", default=1e-6, type=float, help="Epsilon for Adam optimizer.")
parser.add_argument("--max_grad_norm", default=5.0, type=float, help="Max gradient norm.")
parser.add_argument(
"--num_train_epochs", default=3.0, type=float, help="Total number of training epochs to perform."
)
parser.add_argument(
"--max_steps",
default=-1,
type=int,
help="If > 0: set total number of training steps to perform. Override num_train_epochs.",
)
parser.add_argument("--warmup_steps", default=0, type=int, help="Linear warmup over warmup_steps.")
parser.add_argument("--logging_steps", type=int, default=50, help="Log every X updates steps.")
parser.add_argument("--save_steps", type=int, default=50, help="Save checkpoint every X updates steps.")
parser.add_argument(
"--eval_all_checkpoints",
action="store_true",
help="Evaluate all checkpoints starting with the same prefix as model_name ending and ending with step number",
)
parser.add_argument("--no_cuda", action="store_true", help="Whether not to use CUDA when available")
parser.add_argument(
"--overwrite_output_dir", action="store_true", help="Overwrite the content of the output directory"
)
parser.add_argument(
"--overwrite_cache", action="store_true", help="Overwrite the cached training and evaluation sets"
)
parser.add_argument("--seed", type=int, default=42, help="random seed for initialization")
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 (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()
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
)
)
# Setup distant debugging if needed
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
print("Waiting for debugger attach")
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")
args.n_gpu = 0 if args.no_cuda else 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)
torch.distributed.init_process_group(backend="nccl")
args.n_gpu = 1
args.device = device
# Setup logging
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.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,
)
# Set seed
set_seed(args)
'''MODEL_CLASSES = {
"bert": (BertConfig, BertForSequenceClassification, BertTokenizer),
}'''
# 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
args.model_type = args.model_type.lower()
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path)
tokenizer = tokenizer_class.from_pretrained(
args.tokenizer_name if args.tokenizer_name else args.model_name_or_path, do_lower_case=args.do_lower_case
)
config.num_labels = 1
model = model_class.from_pretrained(
args.model_name_or_path, from_tf=bool(".ckpt" in args.model_name_or_path), config=config
)
if args.local_rank == 0:
torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab
model.to(args.device)
logger.info("Training/evaluation parameters %s", args)
# Training
if args.do_train:
train_dataset = load_and_cache_examples(args, tokenizer, train = True, evaluate=False, test = False, output_examples=False)
# train = False, evaluate=False, test = False,
global_step, tr_loss = train(args, train_dataset, model, tokenizer)
logger.info(" global_step = %s, average loss = %s", global_step, tr_loss)
# Save the trained model and the tokenizer
if args.local_rank == -1 or torch.distributed.get_rank() == 0:
# Create output directory if needed
if not os.path.exists(args.output_dir) and args.local_rank in [-1, 0]:
os.makedirs(args.output_dir)
logger.info("Saving model checkpoint to %s", args.output_dir)
# Save a trained model, configuration and tokenizer using `save_pretrained()`.
# They can then be reloaded using `from_pretrained()`
model_to_save = (
model.module if hasattr(model, "module") else model
) # Take care of distributed/parallel training
model_to_save.save_pretrained(args.output_dir)
tokenizer.save_pretrained(args.output_dir)
# Good practice: save your training arguments together with the trained model
torch.save(args, os.path.join(args.output_dir, "training_args.bin"))
# Load a trained model and vocabulary that you have fine-tuned
model = model_class.from_pretrained(args.output_dir)
tokenizer = tokenizer_class.from_pretrained(args.output_dir)
model.to(args.device)
# Evaluation - we can ask to evaluate all the checkpoints (sub-directories) in a directory
results = {}
if args.do_eval and args.local_rank in [-1, 0]:
if args.do_train:
checkpoints = [args.output_dir]
else:
# if do_train is False and do_eval is true, load model directly from pretrained.
checkpoints = [args.model_name_or_path]
if args.eval_all_checkpoints:
checkpoints = list(
os.path.dirname(c) for c in sorted(glob.glob(args.output_dir + "/**/" + WEIGHTS_NAME, recursive=True))
)
logging.getLogger("transformers.modeling_utils").setLevel(logging.WARN) # Reduce model loading logs
logger.info("Evaluate the following checkpoints: %s", checkpoints)
for checkpoint in checkpoints:
# Reload the model
global_step = checkpoint.split("-")[-1] if len(checkpoints) > 1 else ""
model = model_class.from_pretrained(checkpoint)
tokenizer = tokenizer_class.from_pretrained(checkpoint)
model.to(args.device)
# Evaluate
eval = False
test = True
result = evaluate(eval, test, args, model, tokenizer, prefix=global_step,)
# train = False, evaluate=False, test = False,
result = dict((k + ("_{}".format(global_step) if global_step else ""), v) for k, v in result.items())
results.update(result)
logger.info("Results: {}".format(results))
return results
if __name__ == "__main__":
main()
use_boolq_py.py
CUDA_VISIBLE_DEVICES=0,2 python run_boolq_roberta.py --model_type roberta --model_name_or_path roberta-large --do_train --do_eval --evaluate_during_training --do_lower_case --train_file train.jsonl --predict_file val.jsonl --test_file test.jsonl --tbname boolq_roberta --max_seq_length 256 --learning_rate 1e-5 --num_train_epochs 1000.0 --logging_steps 500 --per_gpu_eval_batch_size=4 --per_gpu_train_batch_size=4 --output_dir boolq_roberta_outputimport json
import argparse
import csv
import glob
import logging
import os
import random
import numpy as np
import torch
from torch.utils.data import DataLoader, RandomSampler, SequentialSampler, TensorDataset
from torch.utils.data.distributed import DistributedSampler
from tqdm import tqdm, trange
from transformers import WEIGHTS_NAME, AdamW, AutoConfig, AutoTokenizer, get_linear_schedule_with_warmup
from transformers.modeling_auto import AutoModelForMultipleChoice
from transformers import (
WEIGHTS_NAME,
AdamW,
BertConfig,
BertForSequenceClassification,
BertTokenizer,
get_linear_schedule_with_warmup,
)
try:
from torch.utils.tensorboard import SummaryWriter
except ImportError:
from tensorboardX import SummaryWriter
MODEL_CLASSES = {
"bert": (BertConfig, BertForSequenceClassification, BertTokenizer),
}
logger = logging.getLogger(__name__)
class BoolqExample(object):
"""A single training/test example for the Dream dataset."""
def __init__(self, swag_id, context_sentence, start_ending, ending_0, label=None):
self.swag_id = swag_id
self.context_sentence = context_sentence
self.start_ending = start_ending
self.endings = [
ending_0
# ending_1,
]
self.label = label
def __str__(self):
return self.__repr__()
def __repr__(self):
attributes = [
"swag_id: {}".format(self.swag_id),
"context_sentence: {}".format(self.context_sentence),
"start_ending: {}".format(self.start_ending),
"ending_0: {}".format(self.endings[0]),
#"ending_1: {}".format(self.endings[1]),
]
if self.label is not None:
attributes.append("label: {}".format(self.label))
return ", ".join(attributes)
'''class InputFeatures(object):
def __init__(self, example_id, choices_features, label):
self.example_id = example_id
self.choices_features = [
{"input_ids": input_ids, "input_mask": input_mask, "segment_ids": segment_ids}
for _, input_ids, input_mask, segment_ids in choices_features
]
self.label = label'''
class InputFeatures(object):
"""
A single set of features of data.
Args:
input_ids: Indices of input sequence tokens in the vocabulary.
attention_mask: Mask to avoid performing attention on padding token indices.
Mask values selected in ``[0, 1]``:
Usually ``1`` for tokens that are NOT MASKED, ``0`` for MASKED (padded) tokens.
token_type_ids: Segment token indices to indicate first and second portions of the inputs.
label: Label corresponding to the input
"""
def __init__(self, input_ids, attention_mask=None, token_type_ids=None, label=None):
self.input_ids = input_ids
self.attention_mask = attention_mask
self.token_type_ids = token_type_ids
self.label = label
def read_boolq_examples(input_file, is_training=True):
with open(input_file, "r", encoding="utf-8") as f:
lines = f.readlines()
examples = []
for line in lines:
data_raw = json.loads(line.strip("\n"))
if data_raw["label"]:
label_input = 1
else:
label_input = 0
examples.append(
BoolqExample(
swag_id=data_raw["idx"],
context_sentence=data_raw["passage"],
start_ending=data_raw["question"],
ending_0="",
# ending_1="",
label=label_input if is_training else None,
)
)
return examples
def convert_examples_to_features(examples, tokenizer, max_seq_length, is_training):
"""Loads a data file into a list of `InputBatch`s."""
# Swag is a multiple choice task. To perform this task using Bert,
# we will use the formatting proposed in "Improving Language
# Understanding by Generative Pre-Training" and suggested by
# @jacobdevlin-google in this issue
# https://github.com/google-research/bert/issues/38.
#
# Each choice will correspond to a sample on which we run the
# inference. For a given Swag example, we will create the 4
# following inputs:
# - [CLS] context [SEP] choice_1 [SEP]
# - [CLS] context [SEP] choice_2 [SEP]
# - [CLS] context [SEP] choice_3 [SEP]
# - [CLS] context [SEP] choice_4 [SEP]
# The model will output a single value for each input. To get the
# final decision of the model, we will run a softmax over these 4
# outputs.
features = []
for example_index, example in tqdm(enumerate(examples)):
context_tokens = tokenizer.tokenize(example.context_sentence)
start_ending_tokens = tokenizer.tokenize(example.start_ending)
context_tokens_choice = context_tokens[:]
ending_tokens = start_ending_tokens
_truncate_seq_pair(context_tokens_choice, ending_tokens, max_seq_length - 3)
tokens = ["[CLS]"] + context_tokens_choice + ["[SEP]"] + ending_tokens + ["[SEP]"]
segment_ids = [0] * (len(context_tokens_choice) + 2) + [1] * (len(ending_tokens) + 1)
input_ids = tokenizer.convert_tokens_to_ids(tokens)
input_mask = [1] * len(input_ids)
# Zero-pad up to the sequence length.
padding = [0] * (max_seq_length - len(input_ids))
input_ids += padding
input_mask += padding
segment_ids += padding
assert len(input_ids) == max_seq_length
assert len(input_mask) == max_seq_length
assert len(segment_ids) == max_seq_length
label = example.label
if example_index < 5:
logger.info("*** Example ***")
logger.info("swag_id: {}".format(example.swag_id))
logger.info("tokens: {}".format(" ".join(tokens)))
logger.info("input_ids: {}".format(" ".join(map(str, input_ids))))
logger.info("attention_mask: {}".format(" ".join(map(str, input_mask))))
logger.info("token_type_ids: {}".format(" ".join(map(str, segment_ids))))
if is_training:
logger.info("label: {}".format(label))
features.append(InputFeatures(input_ids = input_ids, attention_mask = input_mask, token_type_ids = segment_ids, label = label ))
return features
def _truncate_seq_pair(tokens_a, tokens_b, max_length):
"""Truncates a sequence pair in place to the maximum length."""
# This is a simple heuristic which will always truncate the longer sequence
# one token at a time. This makes more sense than truncating an equal percent
# of tokens from each, since if one sequence is very short then each token
# that's truncated likely contains more information than a longer sequence.
while True:
total_length = len(tokens_a) + len(tokens_b)
if total_length <= max_length:
break
if len(tokens_a) > len(tokens_b):
tokens_a.pop()
else:
tokens_b.pop()
'''def accuracy(out, labels):
outputs = np.argmax(out, axis=1)
return np.sum(outputs == labels)'''
'''def accuracy(out, labels):
outputs = (out >= 0.5).astype(np.int)
return np.sum(outputs == labels)'''
def accuracy(out, labels,n,data):
outputs = (out.squeeze(1) >= 0.5).astype(np.int)
for i in outputs:
n[0] = n[0] + 1
print(n[0], ": ", i)
#data = open("out2.txt", 'a', encoding='utf-8')
print("{\"idx\":", n[0], ", \"label\":", i, "}", file=data)
#data.close()
return np.sum(outputs == labels)
def select_field(features, field):
return [[choice[field] for choice in feature.choices_features] for feature in features]
def set_seed(args):
random.seed(args.seed)
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.n_gpu > 0:
torch.cuda.manual_seed_all(args.seed)
def load_and_cache_examples(args, tokenizer, train = False, evaluate=False, test = False, output_examples=False):
if args.local_rank not in [-1, 0]:
torch.distributed.barrier() # Make sure only the first process in distributed training process the dataset, and the others will use the cache
# Load data features from cache or dataset file
# input_file = args.predict_file if evaluate else args.train_file
if train:
input_file = args.train_file
if evaluate:
input_file = args.predict_file
if test:
input_file = args.test_file
cached_features_file = os.path.join(
os.path.dirname(input_file),
"cached_{}_{}_{}".format(
"dev" if evaluate else "train",
list(filter(None, args.model_name_or_path.split("/"))).pop(),
str(args.max_seq_length),
),
)
if os.path.exists(cached_features_file) and not args.overwrite_cache and not output_examples:
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", input_file)
examples = read_boolq_examples(input_file)
features = convert_examples_to_features(examples, tokenizer, args.max_seq_length, not evaluate)
if args.local_rank in [-1, 0]:
logger.info("Saving features into cached file %s", cached_features_file)
torch.save(features, cached_features_file)
if args.local_rank == 0:
torch.distributed.barrier() # Make sure only the first process in distributed training process the dataset, and the others will use the cache
# Convert to Tensors and build dataset
# all_input_ids = torch.tensor(select_field(features, "input_ids"), dtype=torch.long)
# all_input_mask = torch.tensor(select_field(features, "input_mask"), dtype=torch.long)
# all_segment_ids = torch.tensor(select_field(features, "segment_ids"), dtype=torch.long)
# all_label = torch.tensor([f.label for f in features], dtype=torch.long)
all_input_ids = torch.tensor([f.input_ids for f in features], dtype=torch.long)
all_input_mask = torch.tensor([f.attention_mask for f in features], dtype=torch.long)
all_segment_ids = torch.tensor([f.token_type_ids for f in features], dtype=torch.long)
all_label = torch.tensor([f.label for f in features], dtype=torch.long)
if evaluate:
dataset = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label)
else:
dataset = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label)
if output_examples:
return dataset, examples, features
return dataset
# train = False, evaluate=False, test = False,
def evaluate(data, eval, test, args, model, tokenizer, prefix=""):
if eval :
dataset, examples, features = load_and_cache_examples(args, tokenizer, train=False, evaluate=True, test=False,
output_examples=True)
if test :
dataset, examples, features = load_and_cache_examples(args, tokenizer, train=False, evaluate=False, test=True,
output_examples=True)
# train = False, evaluate=False, test = False,
if not os.path.exists(args.output_dir) and args.local_rank in [-1, 0]:
os.makedirs(args.output_dir)
args.eval_batch_size = args.per_gpu_eval_batch_size * max(1, args.n_gpu)
# Note that DistributedSampler samples randomly
eval_sampler = SequentialSampler(dataset) if args.local_rank == -1 else DistributedSampler(dataset)
eval_dataloader = DataLoader(dataset, sampler=eval_sampler, batch_size=args.eval_batch_size)
# Eval!
logger.info("***** Running evaluation {} *****".format(prefix))
logger.info(" Num examples = %d", len(dataset))
logger.info(" Batch size = %d", args.eval_batch_size)
eval_loss, eval_accuracy = 0, 0
nb_eval_steps, nb_eval_examples = 0, 0
n = [0]
for batch in tqdm(eval_dataloader, desc="Evaluating"):
model.eval()
batch = tuple(t.to(args.device) for t in batch)
with torch.no_grad():
inputs = {
"input_ids": batch[0],
"attention_mask": batch[1],
# 'token_type_ids': None if args.model_type == 'xlm' else batch[2] # XLM don't use segment_ids
"token_type_ids": batch[2],
"labels": batch[3],
}
# if args.model_type in ['xlnet', 'xlm']:
# inputs.update({'cls_index': batch[4],
# 'p_mask': batch[5]})
outputs = model(**inputs)
tmp_eval_loss, logits = outputs[:2]
eval_loss += tmp_eval_loss.mean().item()
logits = logits.detach().cpu().numpy()
label_ids = inputs["labels"].to("cpu").numpy()
#print(logits,label_ids)
tmp_eval_accuracy = accuracy(logits, label_ids, n, data)
eval_accuracy += tmp_eval_accuracy
nb_eval_steps += 1
nb_eval_examples += inputs["input_ids"].size(0)
eval_loss = eval_loss / nb_eval_steps
eval_accuracy = eval_accuracy / nb_eval_examples
result = {"eval_loss": eval_loss, "eval_accuracy": eval_accuracy}
output_eval_file = os.path.join(args.output_dir, "eval_results.txt")
with open(output_eval_file, "w") as writer:
logger.info("***** Eval results *****")
for key in sorted(result.keys()):
logger.info("%s = %s", key, str(result[key]))
writer.write("%s = %s\n" % (key, str(result[key])))
return result
def main():
parser = argparse.ArgumentParser()
data = open("out2.txt", 'w', encoding='utf-8')
# Required parameters
parser.add_argument(
"--train_file",
default=None,
type=str,
required=True,
help="SWAG csv for training. E.g., train.csv"
)
parser.add_argument(
"--predict_file",
default=None,
type=str,
required=True,
help="SWAG csv for predictions. E.g., val.csv",
)
parser.add_argument(
"--test_file",
default=None,
type=str,
required=True,
help="SWAG csv for test. E.g., test.csv",
)
parser.add_argument(
"--model_type",
default=None,
type=str,
required=True,
help="Model type selected in the list: ",
)
parser.add_argument(
"--model_name_or_path",
default=None,
type=str,
required=True,
help="Path to pre-trained model or shortcut name selected in the list: " ,
)
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(
"--config_name", default="", type=str, help="Pretrained config name or path if not the same as model_name"
)
parser.add_argument(
"--tokenizer_name",
default="",
type=str,
help="Pretrained tokenizer name or path if not the same as model_name",
)
parser.add_argument(
"--max_seq_length",
default=384,
type=int,
help="The maximum total input sequence length after tokenization. Sequences "
"longer than this will be truncated, and sequences shorter than this will be padded.",
)
parser.add_argument("--do_train", action="store_true", help="Whether to run training.")
parser.add_argument("--do_eval", action="store_true", help="Whether to run eval on the dev set.")
parser.add_argument(
"--evaluate_during_training", action="store_true", help="Rul evaluation during training at each logging step."
)
parser.add_argument(
"--do_lower_case", action="store_true", help="Set this flag if you are using an uncased model."
)
parser.add_argument("--per_gpu_train_batch_size", default=8, type=int, help="Batch size per GPU/CPU for training.")
parser.add_argument(
"--per_gpu_eval_batch_size", default=8, type=int, help="Batch size per GPU/CPU for evaluation."
)
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("--weight_decay", default=0.0, type=float, help="Weight deay if we apply some.")
parser.add_argument("--adam_epsilon", default=1e-8, type=float, help="Epsilon for Adam optimizer.")
parser.add_argument("--max_grad_norm", default=1.0, type=float, help="Max gradient norm.")
parser.add_argument(
"--num_train_epochs", default=3.0, type=float, help="Total number of training epochs to perform."
)
parser.add_argument(
"--max_steps",
default=-1,
type=int,
help="If > 0: set total number of training steps to perform. Override num_train_epochs.",
)
parser.add_argument("--warmup_steps", default=0, type=int, help="Linear warmup over warmup_steps.")
parser.add_argument("--logging_steps", type=int, default=50, help="Log every X updates steps.")
parser.add_argument("--save_steps", type=int, default=50, help="Save checkpoint every X updates steps.")
parser.add_argument(
"--eval_all_checkpoints",
action="store_true",
help="Evaluate all checkpoints starting with the same prefix as model_name ending and ending with step number",
)
parser.add_argument("--no_cuda", action="store_true", help="Whether not to use CUDA when available")
parser.add_argument(
"--overwrite_output_dir", action="store_true", help="Overwrite the content of the output directory"
)
parser.add_argument(
"--overwrite_cache", action="store_true", help="Overwrite the cached training and evaluation sets"
)
parser.add_argument("--seed", type=int, default=42, help="random seed for initialization")
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 (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()
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
)
)
# Setup distant debugging if needed
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
print("Waiting for debugger attach")
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")
args.n_gpu = 0 if args.no_cuda else 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)
torch.distributed.init_process_group(backend="nccl")
args.n_gpu = 1
args.device = device
# Setup logging
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.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,
)
# Set seed
set_seed(args)
'''MODEL_CLASSES = {
"bert": (BertConfig, BertForSequenceClassification, BertTokenizer),
}'''
# 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
args.model_type = args.model_type.lower()
config_class, model_class, tokenizer_class = MODEL_CLASSES[args.model_type]
config = config_class.from_pretrained(args.config_name if args.config_name else args.model_name_or_path)
tokenizer = tokenizer_class.from_pretrained(
args.tokenizer_name if args.tokenizer_name else args.model_name_or_path, do_lower_case=args.do_lower_case
)
config.num_labels = 1
model = model_class.from_pretrained(
args.model_name_or_path, from_tf=bool(".ckpt" in args.model_name_or_path), config=config
)
if args.local_rank == 0:
torch.distributed.barrier() # Make sure only the first process in distributed training will download model & vocab
model.to(args.device)
logger.info("Training/evaluation parameters %s", args)
# Evaluation - we can ask to evaluate all the checkpoints (sub-directories) in a directory
results = {}
if args.do_eval and args.local_rank in [-1, 0]:
checkpoints = [args.output_dir]
logger.info("Evaluate the following checkpoints: %s", checkpoints)
for checkpoint in checkpoints:
# Reload the model
global_step = checkpoint.split("-")[-1] if len(checkpoints) > 1 else ""
model = model_class.from_pretrained(checkpoint)
tokenizer = tokenizer_class.from_pretrained(checkpoint)
model.to(args.device)
# Evaluate
eval = False
test = True
result = evaluate(data, eval, test, args, model, tokenizer, global_step,)
# train = False, evaluate=False, test = False,
result = dict((k + ("_{}".format(global_step) if global_step else ""), v) for k, v in result.items())
results.update(result)
logger.info("Results: {}".format(results))
data.close()
return results
'''MODEL_CLASSES = {
"bert": (BertConfig, BertForSequenceClassification, BertTokenizer),
}'''
if __name__ == "__main__":
main()