Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -25,7 +25,7 @@ task_ids:
|
|
25 |
|
26 |
## Dataset Summary
|
27 |
|
28 |
-
**RepoBench-P (Pipeline)** is a subtask of [
|
29 |
|
30 |
## Settings
|
31 |
|
|
|
25 |
|
26 |
## Dataset Summary
|
27 |
|
28 |
+
**RepoBench-P (Pipeline)** is a subtask of *RepoBench*([GitHub](https://github.com/Leolty/repobench) [arXiv](https://arxiv.org/abs/2306.03091)), combinig the retrieval and code completion tasks. Specifically, the retrieval task is used to retrieve the most relevant code snippet first, and then do the code completion task with retrieved code snippet as cross-file context for next-line prediction, which mirrors complex real-world scenarios that a practical auto-completion system would face.
|
29 |
|
30 |
## Settings
|
31 |
|