def __init__(self, task_def_path):
self._task_def_dic = yaml.safe_load(open(task_def_path))
global_map = {}
n_class_map = {}
data_type_map = {}
task_type_map = {}
metric_meta_map = {}
enable_san_map = {}
dropout_p_map = {}
encoderType_map = {}
loss_map = {}
kd_loss_map = {}
uniq_encoderType = set()
for task, task_def in self._task_def_dic.items():
assert "_" not in task, "task name should not contain '_', current task name: %s" % task
n_class_map[task] = task_def["n_class"]
data_format = DataFormat[task_def["data_format"]]
data_type_map[task] = data_format
task_type_map[task] = TaskType[task_def["task_type"]]
metric_meta_map[task] = tuple(
Metric[metric_name] for metric_name in task_def["metric_meta"])
enable_san_map[task] = task_def["enable_san"]
uniq_encoderType.add(EncoderModelType[task_def["encoder_type"]])
if "labels" in task_def:
labels = task_def["labels"]
label_mapper = Vocabulary(True)
for label in labels:
label_mapper.add(label)
global_map[task] = label_mapper
if "dropout_p" in task_def:
dropout_p_map[task] = task_def["dropout_p"]
# loss map
if "loss" in task_def:
t_loss = task_def["loss"]
loss_crt = LossCriterion[t_loss]
loss_map[task] = loss_crt
else:
loss_map[task] = None
if "kd_loss" in task_def:
t_loss = task_def["kd_loss"]
loss_crt = LossCriterion[t_loss]
kd_loss_map[task] = loss_crt
else:
kd_loss_map[task] = None
assert len(
uniq_encoderType) == 1, 'The shared encoder has to be the same.'
self.global_map = global_map
self.n_class_map = n_class_map
self.data_type_map = data_type_map
self.task_type_map = task_type_map
self.metric_meta_map = metric_meta_map
self.enable_san_map = enable_san_map
self.dropout_p_map = dropout_p_map
self.encoderType = uniq_encoderType.pop()
self.loss_map = loss_map
self.kd_loss_map = kd_loss_map
def __init__(self, task_def_path):
self._task_def_dic = yaml.safe_load(open(task_def_path))
global_map = {}
n_class_map = {}
data_type_map = {}
task_type_map = {}
metric_meta_map = {}
split_names_map = {}
enable_san_map = {}
dropout_p_map = {}
loss_map = {}
kd_loss_map = {}
for task, task_def in self._task_def_dic.items():
assert "_" not in task, "task name should not contain '_', current task name: %s" % task
n_class_map[task] = task_def["n_class"]
data_format = DataFormat[task_def["data_format"]]
data_type_map[task] = data_format
task_type_map[task] = TaskType[task_def["task_type"]]
metric_meta_map[task] = tuple(
Metric[metric_name] for metric_name in task_def["metric_meta"])
split_names_map[task] = task_def.get("split_names",
["train", "dev", "test"])
enable_san_map[task] = task_def["enable_san"]
if "labels" in task_def:
labels = task_def["labels"]
label_mapper = Vocabulary(True)
for label in labels:
label_mapper.add(label)
global_map[task] = label_mapper
if "dropout_p" in task_def:
dropout_p_map[task] = task_def["dropout_p"]
# loss map
if "loss" in task_def:
t_loss = task_def["loss"]
loss_crt = LossCriterion[t_loss]
loss_map[task] = loss_crt
else:
loss_map[task] = None
if "kd_loss" in task_def:
t_loss = task_def["kd_loss"]
loss_crt = LossCriterion[t_loss]
kd_loss_map[task] = loss_crt
else:
kd_loss_map[task] = None
self._global_map = global_map
self._n_class_map = n_class_map
self._data_type_map = data_type_map
self._task_type_map = task_type_map
self._metric_meta_map = metric_meta_map
self._split_names_map = split_names_map
self._enable_san_map = enable_san_map
self._dropout_p_map = dropout_p_map
self._loss_map = loss_map
self._kd_loss_map = kd_loss_map
self._task_def_dic = {}
def __init__(self, task_def_path):
self._task_def_dic = yaml.safe_load(open(task_def_path))
global_map = {}
n_class_map = {}
data_format_map = {}
data_type_map = {}
task_type_map = {}
metric_meta_map = {}
enable_san_map = {}
dropout_p_map = {}
tasks = []
split_names_map = {}
for task, task_def in self._task_def_dic.items():
tasks.append(task)
assert "_" not in task, "task name should not contain '_', current task name: %s" % task
n_class_map[task] = task_def["n_class"]
data_format = DataFormat[task_def["data_format"]]
data_format_map[task] = data_format
if data_format == DataFormat.PremiseOnly:
data_type_map[task] = 1
elif data_format in (DataFormat.PremiseAndMultiHypothesis,
DataFormat.PremiseAndOneHypothesis):
data_type_map[task] = 0
else:
raise ValueError(data_format)
task_type_map[task] = TaskType[task_def["task_type"]]
metric_meta_map[task] = tuple(
Metric[metric_name] for metric_name in task_def["metric_meta"])
enable_san_map[task] = task_def["enable_san"]
if "labels" in task_def:
labels = task_def["labels"]
label_mapper = Vocabulary(True)
for label in labels:
label_mapper.add(label)
global_map[task] = label_mapper
if "dropout_p" in task_def:
dropout_p_map[task] = task_def["dropout_p"]
split_names = task_def.get("split_names", ["train", "dev", "test"])
split_names_map[task] = split_names
self.tasks = tasks
self.global_map = global_map
self.n_class_map = n_class_map
self.data_format_map = data_format_map
self.data_type_map = data_type_map
self.task_type_map = task_type_map
self.metric_meta_map = metric_meta_map
self.enable_san_map = enable_san_map
self.dropout_p_map = dropout_p_map
self.split_names_map = split_names_map
def load_dict(path):
vocab = Vocabulary(neat=True)
vocab.add('<s>')
vocab.add('<pad>')
vocab.add('</s>')
vocab.add('<unk>')
with open(path, 'r', encoding='utf8') as reader:
for line in reader:
idx = line.rfind(' ')
if idx == -1:
raise ValueError(
"Incorrect dictionary format, expected '<token> <cnt>'")
word = line[:idx]
vocab.add(word)
return vocab
def __init__(self, task_def_path):
self._task_def_dic = yaml.safe_load(open(task_def_path))
global_map = {}
n_class_map = {}
data_type_map = {}
task_type_map = {}
metric_meta_map = {}
enable_san_map = {}
dropout_p_map = {}
encoderType_map = {}
uniq_encoderType = set()
for task, task_def in self._task_def_dic.items():
assert "_" not in task, "task name should not contain '_', current task name: %s" % task
n_class_map[task] = task_def["n_class"]
data_format = DataFormat[task_def["data_format"]]
if data_format == DataFormat.PremiseOnly:
data_type_map[task] = 1
elif data_format in (DataFormat.PremiseAndMultiHypothesis,
DataFormat.PremiseAndOneHypothesis):
data_type_map[task] = 0
else:
raise ValueError(data_format)
task_type_map[task] = TaskType[task_def["task_type"]]
metric_meta_map[task] = tuple(
Metric[metric_name] for metric_name in task_def["metric_meta"])
enable_san_map[task] = task_def["enable_san"]
uniq_encoderType.add(EncoderModelType[task_def["encoder_type"]])
if "labels" in task_def:
labels = task_def["labels"]
label_mapper = Vocabulary(True)
for label in labels:
label_mapper.add(label)
global_map[task] = label_mapper
if "dropout_p" in task_def:
dropout_p_map[task] = task_def["dropout_p"]
assert len(
uniq_encoderType) == 1, 'The shared encoder has to be the same.'
self.global_map = global_map
self.n_class_map = n_class_map
self.data_type_map = data_type_map
self.task_type_map = task_type_map
self.metric_meta_map = metric_meta_map
self.enable_san_map = enable_san_map
self.dropout_p_map = dropout_p_map
self.encoderType = uniq_encoderType.pop()
def main(args):
## hyper param
do_lower_case = args.do_lower_case
root = args.root_dir
assert os.path.exists(root)
is_uncased = False
if 'uncased' in args.bert_model:
is_uncased = True
bert_tokenizer = BertTokenizer.from_pretrained(args.bert_model, do_lower_case=do_lower_case)
mt_dnn_suffix = 'mt_dnn'
if is_uncased:
mt_dnn_suffix = '{}_uncased'.format(mt_dnn_suffix)
else:
mt_dnn_suffix = '{}_cased'.format(mt_dnn_suffix)
if do_lower_case:
mt_dnn_suffix = '{}_lower'.format(mt_dnn_suffix)
mt_dnn_root = os.path.join(root, mt_dnn_suffix)
if not os.path.isdir(mt_dnn_root):
os.mkdir(mt_dnn_root)
task_def_dic = yaml.safe_load(open(args.task_def))
for task, task_def in task_def_dic.items():
logger.info("Task %s" % task)
data_format = DataFormat[task_def["data_format"]]
task_type = TaskType[task_def["task_type"]]
label_mapper = None
if "labels" in task_def:
labels = task_def["labels"]
label_mapper = Vocabulary(True)
for label in labels:
label_mapper.add(label)
split_names = task_def.get("split_names", ["train", "dev", "test"])
for split_name in split_names:
rows = load_data(os.path.join(root, "%s_%s.tsv" % (task, split_name)), data_format, task_type, label_mapper)
dump_path = os.path.join(mt_dnn_root, "%s_%s.json" % (task, split_name))
logger.info(dump_path)
build_data(rows, dump_path, bert_tokenizer, data_format)
# Copyright (c) Microsoft. All rights reserved.
from data_utils.vocab import Vocabulary
from data_utils.metrics import compute_acc, compute_f1, compute_mcc, compute_pearson, compute_spearman
# scitail
ScitailLabelMapper = Vocabulary(True)
ScitailLabelMapper.add('neutral')
ScitailLabelMapper.add('entails')
# label map
SNLI_LabelMapper = Vocabulary(True)
SNLI_LabelMapper.add('contradiction')
SNLI_LabelMapper.add('neutral')
SNLI_LabelMapper.add('entailment')
# qnli
QNLILabelMapper = Vocabulary(True)
QNLILabelMapper.add('not_entailment')
QNLILabelMapper.add('entailment')
GLOBAL_MAP = {
'scitail': ScitailLabelMapper,
'mnli': SNLI_LabelMapper,
'snli': SNLI_LabelMapper,
'qnli': QNLILabelMapper,
'qnnli': QNLILabelMapper,
'rte': QNLILabelMapper,
'diag': SNLI_LabelMapper,
}