lannliat
commited on
Commit
•
cb1a1bb
1
Parent(s):
3ffddfd
Fix imports, filename, README.
Browse files- .gitignore +4 -0
- README.md +21 -1
- load_script.py → opus.py +2 -4
.gitignore
CHANGED
@@ -1,3 +1,7 @@
|
|
|
|
|
|
|
|
|
|
1 |
# Byte-compiled / optimized / DLL files
|
2 |
__pycache__/
|
3 |
*.py[cod]
|
|
|
1 |
+
data
|
2 |
+
test_loader.ipynb
|
3 |
+
preprocess.ipynb
|
4 |
+
|
5 |
# Byte-compiled / optimized / DLL files
|
6 |
__pycache__/
|
7 |
*.py[cod]
|
README.md
CHANGED
@@ -48,7 +48,8 @@ task_ids: []
|
|
48 |
## Dataset Description
|
49 |
Loads [OPUS](https://opus.nlpl.eu/) as HuggingFace dataset. OPUS is an open parallel corpus covering 700+ languages and 1100+ datasets.
|
50 |
|
51 |
-
Given a `src` and `tgt` language, this repository
|
|
|
52 |
|
53 |
**Requirements**. Pip install my fork of `OpusTools`.
|
54 |
|
@@ -57,6 +58,25 @@ git clone https://github.com/larrylawl/OpusTools.git
|
|
57 |
pip install -e OpusTools/opustools_pkg
|
58 |
```
|
59 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
**Disclaimer**.
|
61 |
|
62 |
This repository is still in active development. Do make a PR if there're any issues!
|
|
|
48 |
## Dataset Description
|
49 |
Loads [OPUS](https://opus.nlpl.eu/) as HuggingFace dataset. OPUS is an open parallel corpus covering 700+ languages and 1100+ datasets.
|
50 |
|
51 |
+
Given a `src` and `tgt` language, this repository can load *all* available parallel corpus. To my knowledge, other OPUS datasets on Huggingface loads a specific corpus
|
52 |
+
|
53 |
|
54 |
**Requirements**. Pip install my fork of `OpusTools`.
|
55 |
|
|
|
58 |
pip install -e OpusTools/opustools_pkg
|
59 |
```
|
60 |
|
61 |
+
**Example Usage**.
|
62 |
+
|
63 |
+
```
|
64 |
+
# args follows `opustools`: https://pypi.org/project/opustools/
|
65 |
+
|
66 |
+
src="en"
|
67 |
+
tgt="id"
|
68 |
+
download_dir="data" # dir to save downloaded files
|
69 |
+
corpus="bible-uedin" # corpus name. Leave as `None` to download all available corpus for the src-tgt pair.
|
70 |
+
|
71 |
+
|
72 |
+
dataset = load_dataset("larrylawl/opus",
|
73 |
+
src=src,
|
74 |
+
tgt=tgt,
|
75 |
+
download_dir=download_dir,
|
76 |
+
corpus=corpus)
|
77 |
+
)
|
78 |
+
```
|
79 |
+
|
80 |
**Disclaimer**.
|
81 |
|
82 |
This repository is still in active development. Do make a PR if there're any issues!
|
load_script.py → opus.py
RENAMED
@@ -22,9 +22,7 @@ import os
|
|
22 |
from typing import List
|
23 |
import datasets
|
24 |
import pandas as pd
|
25 |
-
|
26 |
-
from OpusTools.opustools_pkg.opustools.opus_read import OpusRead
|
27 |
-
from OpusTools.opustools_pkg.opustools.opus_get import OpusRead
|
28 |
|
29 |
_DESCRIPTION = """\
|
30 |
Downloads OPUS data using `opustools`.
|
@@ -43,7 +41,7 @@ class OpusConfig(datasets.BuilderConfig):
|
|
43 |
Args:
|
44 |
src: str, source language
|
45 |
tgt: str, target language
|
46 |
-
corpus:
|
47 |
download_dir: str, dir to save downloaded files
|
48 |
|
49 |
"""
|
|
|
22 |
from typing import List
|
23 |
import datasets
|
24 |
import pandas as pd
|
25 |
+
import opustools
|
|
|
|
|
26 |
|
27 |
_DESCRIPTION = """\
|
28 |
Downloads OPUS data using `opustools`.
|
|
|
41 |
Args:
|
42 |
src: str, source language
|
43 |
tgt: str, target language
|
44 |
+
corpus: str, corpus name. Leave as `None` to download all available corpus for the src-tgt pair.
|
45 |
download_dir: str, dir to save downloaded files
|
46 |
|
47 |
"""
|