def __init__(self, taskpath, max_seq_len, load_data, seed=1111):
logging.debug('***** Transfer task : ANLI Entailment*****\n\n')
self.seed = seed
targ_map = {'neutral': 0, 'entailment': 1, 'contradiction': 2}
train = sort_split(self.loadFile(os.path.join(taskpath, 'multinli_1.0_train.txt'),
max_seq_len, targ_map, load_data))
valid = sort_split(self.loadFile(os.path.join(taskpath, 'multinli_1.0_dev_matched.txt'),
max_seq_len, targ_map, load_data))
test = sort_split(self.loadAux(os.path.join(taskpath, 'adversarial_nli.tsv'),
max_seq_len, targ_map, load_data))
self.samples = train[0] + train[1] + valid[0] + valid[1] + test[0] + test[1]
self.data = {'train': train, 'valid': valid, 'test': test}
def __init__(self, taskpath, max_seq_len, load_data, seed=1111):
logging.debug('***** Transfer task : CoLA *****\n\n')
self.seed = seed
train = sort_split(self.loadFile(os.path.join(taskpath, 'train.tsv'),
max_seq_len, load_data),
pair_input=0)
valid = sort_split(self.loadFile(os.path.join(taskpath, 'dev.tsv'),
max_seq_len, load_data),
pair_input=0)
test = sort_split(self.loadTest(os.path.join(taskpath, 'test.tsv'),
max_seq_len, load_data),
pair_input=0)
self.samples = train[0] + valid[0] + test[0]
self.data = {'train': train, 'valid': valid, 'test': test}
def __init__(self, taskpath, max_seq_len, load_data, seed=1111):
logging.debug(
'***** Transfer task : Recognizing Textual Entailment*****\n\n')
self.seed = seed
train = sort_split(
self.loadFile(os.path.join(taskpath, 'train.tsv'), max_seq_len,
load_data))
valid = sort_split(
self.loadFile(os.path.join(taskpath, 'dev.tsv'), max_seq_len,
load_data))
test = sort_split(
self.loadTest(os.path.join(taskpath, 'test.tsv'), max_seq_len,
load_data))
self.samples = train[0] + train[1] + valid[0] + valid[1] + test[
0] + test[1]
self.data = {'train': train, 'valid': valid, 'test': test}
def __init__(self, taskpath, max_seq_len, load_data, seed=1111):
logging.debug('***** Transfer task : QNLI Classification *****\n\n')
self.seed = seed
train = sort_split(
self.loadFile(os.path.join(taskpath, "train.tsv"), max_seq_len,
load_data))
valid = sort_split(
self.loadFile(os.path.join(taskpath, "dev.tsv"), max_seq_len,
load_data))
test = sort_split(
self.loadTest(os.path.join(taskpath, "test.tsv"), max_seq_len,
load_data))
# sort data (by s2 first) to reduce padding
self.samples = train[0] + train[1] + valid[0] + valid[1] + test[
0] + test[1]
self.data = {'train': train, 'valid': valid, 'test': test}
def __init__(self, taskpath, max_seq_len, load_data, seed=1111):
''' Class for QQP task '''
logging.debug(
'***** Transfer task : Quora Question Similarity*****\n\n')
self.seed = seed
train = sort_split(
self.loadFile(os.path.join(taskpath, 'train.tsv'), max_seq_len,
load_data))
valid = sort_split(
self.loadFile(os.path.join(taskpath, 'dev.tsv'), max_seq_len,
load_data))
test = sort_split(
self.loadTest(os.path.join(taskpath, 'test.tsv'), max_seq_len,
load_data))
self.samples = train[0] + train[1] + valid[0] + valid[1] + test[
0] + test[1]
self.data = {'train': train, 'valid': valid, 'test': test}
def __init__(self, taskpath, max_seq_len=50, load_data=1, seed=1111):
logging.debug('***** Transfer task : MNLI Entailment*****\n\n')
self.seed = seed
targ_map = {'neutral': 0, 'entailment': 1, 'contradiction': 2}
train = sort_split(self.loadFile(os.path.join(taskpath, 'multinli_1.0_train.txt'),
max_seq_len, targ_map, load_data))
valid = sort_split(self.loadFile(os.path.join(taskpath, 'multinli_1.0_dev_both.txt'),
max_seq_len, targ_map, load_data))
test_m = self.loadTest(os.path.join(taskpath, 'mnli_matched_test_ans.tsv'),
max_seq_len, targ_map, load_data)
test_m = sort_split(test_m)
test_mm = sort_split(self.loadTest(os.path.join(taskpath, 'mnli_mismatched_test_ans.tsv'),
max_seq_len, targ_map, load_data))
test_d = sort_split(self.loadTest(os.path.join(taskpath, 'diagnostic_test_ans.tsv'),
max_seq_len, targ_map, load_data))
self.samples = train[0] + train[1] + valid[0] + valid[1] + test_m[0] + test_m[1] + \
test_mm[0] + test_mm[1] + test_d[0] + test_d[1]
self.data = {'train': train, 'valid': valid, 'test': test_m,
'test_mismatched': test_mm, 'diagnostic': test_d}