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

Fix CI with change of name of nlp (#7054)

Browse Source 
* nlp -> datasets

* More nlp -> datasets

* Woopsie

* More nlp -> datasets

* One last
This commit is contained in:
Sylvain Gugger
2020-09-10 14:51:08 -04:00
committed by GitHub
parent e9a2f772bc
commit 514486739c
13 changed files with 42 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
examples/seq2seq/download_wmt.py
View File

@@ -5,25 +5,25 @@ from tqdm import tqdm
def download_wmt_dataset(src_lang="ro", tgt_lang="en", dataset="wmt16", save_dir=None) -> None:
"""Download a dataset using the nlp package and save it to the format expected by finetune.py
"""Download a dataset using the datasets package and save it to the format expected by finetune.py
Format of save_dir: train.source, train.target, val.source, val.target, test.source, test.target.
Args:
src_lang: <str> source language
tgt_lang: <str> target language
dataset: <str> wmt16, wmt17, etc. wmt16 is a good start as it's small. To get the full list run `import nlp; print([d.id for d in nlp.list_datasets() if "wmt" in d.id])`
dataset: <str> wmt16, wmt17, etc. wmt16 is a good start as it's small. To get the full list run `import datasets; print([d.id for d in datasets.list_datasets() if "wmt" in d.id])`
save_dir: <str>, where to save the datasets, defaults to f'{dataset}-{src_lang}-{tgt_lang}'
Usage:
>>> download_wmt_dataset('ro', 'en', dataset='wmt16') # saves to wmt16-ro-en
"""
try:
import nlp
import datasets
except (ModuleNotFoundError, ImportError):
raise ImportError("run pip install nlp")
raise ImportError("run pip install datasets")
pair = f"{src_lang}-{tgt_lang}"
print(f"Converting {dataset}-{pair}")
ds = nlp.load_dataset(dataset, pair)
ds = datasets.load_dataset(dataset, pair)
if save_dir is None:
save_dir = f"{dataset}-{pair}"
save_dir = Path(save_dir)