def download_xnli_data_and_write_config(task_data_base_path: str, task_config_base_path: str):
xnli_temp_path = py_io.create_dir(task_data_base_path, "xnli_temp")
download_utils.download_and_unzip(
"https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip", xnli_temp_path,
)
full_val_data = py_io.read_jsonl(os.path.join(xnli_temp_path, "XNLI-1.0", "xnli.dev.jsonl"))
val_data = datastructures.group_by(full_val_data, key_func=lambda elem: elem["language"])
full_test_data = py_io.read_jsonl(os.path.join(xnli_temp_path, "XNLI-1.0", "xnli.test.jsonl"))
test_data = datastructures.group_by(full_test_data, lambda elem: elem["language"])
languages = sorted(list(val_data))
for lang in languages:
task_name = f"xnli_{lang}"
task_data_path = py_io.create_dir(task_data_base_path, task_name)
val_path = os.path.join(task_data_path, "val.jsonl")
test_path = os.path.join(task_data_path, "test.jsonl")
py_io.write_jsonl(data=val_data[lang], path=val_path)
py_io.write_jsonl(data=test_data[lang], path=test_path)
py_io.write_json(
data={
"task": "xnli",
"paths": {"val": val_path, "test": test_path},
"name": task_name,
"kwargs": {"language": lang},
},
path=os.path.join(task_config_base_path, f"{task_name}_config.json"),
)
shutil.rmtree(xnli_temp_path)
def download_fever_nli_data_and_write_config(
task_name: str, task_data_path: str, task_config_path: str
):
os.makedirs(task_data_path, exist_ok=True)
download_utils.download_and_unzip(
("https://www.dropbox.com/s/hylbuaovqwo2zav/nli_fever.zip?dl=1"), task_data_path,
)
# Since the FEVER NLI dataset doesn't have labels for the dev set, we also download the original
# FEVER dev set and match example CIDs to obtain labels.
orig_dev_path = os.path.join(task_data_path, "fever-dev-temp.jsonl")
download_utils.download_file(
"https://s3-eu-west-1.amazonaws.com/fever.public/shared_task_dev.jsonl", orig_dev_path,
)
id_to_label = {}
for line in py_io.read_jsonl(orig_dev_path):
if "id" not in line:
logging.warning("FEVER dev dataset is missing ID.")
continue
if "label" not in line:
logging.warning("FEVER dev dataset is missing label.")
continue
id_to_label[line["id"]] = line["label"]
os.remove(orig_dev_path)
dev_path = os.path.join(task_data_path, "nli_fever", "dev_fitems.jsonl")
dev_examples = []
for line in py_io.read_jsonl(dev_path):
if "cid" not in line:
logging.warning("Data in {} is missing CID.".format(dev_path))
continue
if int(line["cid"]) not in id_to_label:
logging.warning("Could not match CID {} to dev data.".format(line["cid"]))
continue
dev_example = line
dev_example["label"] = id_to_label[int(line["cid"])]
dev_examples.append(dev_example)
py_io.write_jsonl(dev_examples, os.path.join(task_data_path, "val.jsonl"))
os.remove(dev_path)
for phase in ["train", "test"]:
os.rename(
os.path.join(task_data_path, "nli_fever", f"{phase}_fitems.jsonl"),
os.path.join(task_data_path, f"{phase}.jsonl"),
)
shutil.rmtree(os.path.join(task_data_path, "nli_fever"))
py_io.write_json(
data={
"task": task_name,
"paths": {
"train": os.path.join(task_data_path, "train.jsonl"),
"val": os.path.join(task_data_path, "val.jsonl"),
"test": os.path.join(task_data_path, "test.jsonl"),
},
"name": task_name,
},
path=task_config_path,
)
def load_log(fol_path):
all_paths = filesystem.find_files_with_ext(fol_path, "zlog")
log_data = {}
for path in all_paths:
key = os.path.abspath(path).replace(os.path.abspath(fol_path),
"")[1:].replace(".zlog", "")
log_data[key] = py_io.read_jsonl(path)
return log_data
def read_examples(cls, path: str, set_type: str):
if path.endswith(".gz"):
with gzip.open(path, "r") as f:
data = [json.loads(row) for row in f.readlines()]
elif path.endswith(".jsonl"):
data = read_jsonl(path)
else:
raise KeyError(f"Unknown format: {path}")
# First row is a header row
assert "header" in data[0]
examples = []
for i, line in enumerate(data[1:]):
for elem in line["qas"]:
if set_type == PHASE.TRAIN:
# Just use first answer for training (e.g. "Car", "Vehicle")
answer = elem["detected_answers"][0]
# Just use first occurrence of answer for training ("Every car is a car.")
answer_span = answer["char_spans"][0]
answer_text = line["context"][
answer_span[0]:answer_span[1] + 1]
# assert len(elem["detected_answers"][0]["char_spans"]) == 1
examples.append(
Example(
qas_id=f"{set_type}-{i}",
question_text=elem["question"],
context_text=line["context"],
answer_text=answer_text,
start_position_character=answer_span[0],
title="",
is_impossible=False,
answers=[],
))
else:
answers = []
for answer in elem["detected_answers"]:
for answer_span in answer["char_spans"]:
answers.append({
"answer_start":
answer_span[0],
"text":
line["context"][answer_span[0]:answer_span[1] +
1],
})
examples.append(
Example(
qas_id=f"{set_type}-{i}",
question_text=elem["question"],
context_text=line["context"],
answer_text=None,
start_position_character=None,
title="",
is_impossible=False,
answers=answers,
))
return examples
def _create_examples(cls, path, set_type):
for i, row in enumerate(py_io.read_jsonl(path)):
yield Example(
guid="%s-%s" % (set_type, i),
tokenized_text=row["tokenized_text"],
masked_spans=[
ExclusiveSpan(start, end)
for start, end in row["masked_spans"]
],
)
def default_get_test_labels(task: Task, possible_labels):
"""test setのラベルを獲得する．
本関数が動作するかは，サブクラスの実装による．
多くのサブクラスは(上位クラスで明示されない)同一のinterfaceを持っているため，この実装でうまくいく．
"""
labels = [
example.label for example in task._create_examples(
read_jsonl(task.test_path), "valid")
]
if any([label not in possible_labels
for label in labels]): # masked labels
labels = [possible_labels[-1]] * len(labels)
return labels
def read_examples(cls, path: str, set_type: str):
examples = []
for entry in read_jsonl(path):
for qa in entry["qas"]:
answer_text = entry["text"][
qa["answer"]["s"]:qa["answer"]["e"]]
examples.append(
Example(
qas_id=f"{set_type}-{len(examples)}",
question_text=qa["question"],
context_text=entry["text"],
answer_text=answer_text,
start_position_character=qa["answer"]["s"],
title="",
is_impossible=False,
answers=[{
"answer_start": qa["answer"]["s"],
"text": answer_text
}],
))
return examples
def read_examples(cls, path: str, set_type: str):
examples = []
for i, line in enumerate(read_jsonl(path)):
if set_type == PHASE.TRAIN:
for j, (answer_start, answer_text) in enumerate(
zip(line["answers"]["answer_start"],
line["answers"]["text"])):
examples.append(
Example(
qas_id=f"{set_type}-{i}",
question_text=line["question"],
context_text=line["context"],
answer_text=answer_text,
start_position_character=answer_start,
title=line["title"],
is_impossible=False,
answers=[],
))
else:
answers = [{
"answer_start": answer_start,
"text": answer_text
} for answer_start, answer_text in zip(
line["answers"]["answer_start"], line["answers"]["text"])]
examples.append(
Example(
qas_id=f"{set_type}-{i}",
question_text=line["question"],
context_text=line["context"],
answer_text=None,
start_position_character=None,
title=line["title"],
is_impossible=False,
answers=answers,
))
return examples
def get_test_examples(self):
return self._create_examples(
lines=read_jsonl(self.test_path),
set_type="test",
language=self.language,
)
def get_val_examples(self):
return self._create_examples(
lines=read_jsonl(self.val_path),
set_type="val",
language=self.language,
)
def get_val_examples(self):
return self._create_examples(lines=read_jsonl(self.path_dict['validation']), set_type="val")
def get_train_examples(self):
return self._create_examples(lines=read_jsonl(self.train_path), set_type="train")
