update intro
Browse files
README.md
CHANGED
|
@@ -4,11 +4,21 @@ license: mit
|
|
| 4 |
|
| 5 |
# SyncBench
|
| 6 |
|
|
|
|
|
|
|
| 7 |
## Dataset Description
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
|
| 9 |
-
### Overview
|
| 10 |
|
| 11 |
-
##
|
| 12 |
- Format: JSON and CSV
|
| 13 |
- Fields:
|
| 14 |
- **instance_id:** SyncBench instance ID
|
|
@@ -43,7 +53,7 @@ license: mit
|
|
| 43 |
- Caller: syncbench_caller_150.csv
|
| 44 |
- Size: 300
|
| 45 |
|
| 46 |
-
##
|
| 47 |
```python
|
| 48 |
from datasets import load_dataset
|
| 49 |
dataset = load_dataset("xuehang/SyncBench")
|
|
|
|
| 4 |
|
| 5 |
# SyncBench
|
| 6 |
|
| 7 |
+
In tackling the challenge of *agent out-of-sync*, **SyncBench** is released as a part of our work [SyncMind: Measuring Agent Out-of-Sync Recovery In Collaborative Software Engineering](https://xhguo7.github.io/SyncMind/)
|
| 8 |
+
|
| 9 |
## Dataset Description
|
| 10 |
+
**SyncBench** for *agent out-of-sync* recovery:
|
| 11 |
+
- Task: Agent *out-of-sync*
|
| 12 |
+
- SyncBench
|
| 13 |
+
- In our current version, we construct *SyncBench* based on 21 popular repositories, while our dataset construction approach allows adaptive upsampling and downsampling for custom needs
|
| 14 |
+
- Expand to incorporate other repositories as data sources
|
| 15 |
+
- Downsampled to smaller ealuation subsets
|
| 16 |
+
- Datasets: Caller and Callee
|
| 17 |
+
- **Caller:** Testing code out-of-sync
|
| 18 |
+
- **Callee:** Tested code out-of-sync
|
| 19 |
|
|
|
|
| 20 |
|
| 21 |
+
## Data Structure
|
| 22 |
- Format: JSON and CSV
|
| 23 |
- Fields:
|
| 24 |
- **instance_id:** SyncBench instance ID
|
|
|
|
| 53 |
- Caller: syncbench_caller_150.csv
|
| 54 |
- Size: 300
|
| 55 |
|
| 56 |
+
## Usage
|
| 57 |
```python
|
| 58 |
from datasets import load_dataset
|
| 59 |
dataset = load_dataset("xuehang/SyncBench")
|