Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,29 @@
|
|
| 1 |
---
|
| 2 |
license: cc-by-sa-4.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: cc-by-sa-4.0
|
| 3 |
+
task_categories:
|
| 4 |
+
- text-generation
|
| 5 |
+
- text2text-generation
|
| 6 |
+
language:
|
| 7 |
+
- en
|
| 8 |
+
- es
|
| 9 |
+
- ja
|
| 10 |
+
- ru
|
| 11 |
+
size_categories:
|
| 12 |
+
- n<1K
|
| 13 |
---
|
| 14 |
+
|
| 15 |
+
__ODEX__ is an Open-Domain EXecution-based NL-to-Code generation data benchmark.
|
| 16 |
+
It contains 945 samples with a total of 1,707 human-written test cases, covering intents in four different natural languages -- 439 in English, 90 in Spanish, 164 in Japanese, and 252 in Russian.
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
If you find our dataset useful, please cite the paper
|
| 21 |
+
|
| 22 |
+
```
|
| 23 |
+
@article{wang2022execution,
|
| 24 |
+
title={Execution-Based Evaluation for Open-Domain Code Generation},
|
| 25 |
+
author={Zhiruo Wang, Shuyan Zhou, Daniel Fried, Graham Neubig},
|
| 26 |
+
journal={arXiv preprint arXiv:2212.10481},
|
| 27 |
+
year={2022}
|
| 28 |
+
}
|
| 29 |
+
```
|