Update README.md
Browse files
README.md
CHANGED
|
@@ -1,14 +1,77 @@
|
|
|
|
|
|
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
license: mit
|
| 3 |
datasets:
|
| 4 |
-
- bug-localization/BeetleBox
|
| 5 |
language:
|
| 6 |
-
- en
|
| 7 |
base_model:
|
| 8 |
-
- codesage/codesage-base
|
| 9 |
tags:
|
| 10 |
-
- bug
|
| 11 |
-
- localization
|
| 12 |
-
- embedding
|
| 13 |
-
- crosslanguage
|
| 14 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Here's the finalized `README.md` in Hugging Face-compatible format, including the metadata block and your full specifications:
|
| 2 |
+
|
| 3 |
---
|
| 4 |
+
|
| 5 |
+
```yaml
|
| 6 |
license: mit
|
| 7 |
datasets:
|
| 8 |
+
- bug-localization/BeetleBox
|
| 9 |
language:
|
| 10 |
+
- en
|
| 11 |
base_model:
|
| 12 |
+
- codesage/codesage-base
|
| 13 |
tags:
|
| 14 |
+
- bug
|
| 15 |
+
- localization
|
| 16 |
+
- embedding
|
| 17 |
+
- crosslanguage
|
| 18 |
+
```
|
| 19 |
+
|
| 20 |
+
# 🔥 BLAZE: Cross-Language and Cross-Project Bug Localization
|
| 21 |
+
|
| 22 |
+
**BLAZE** is a transformer-based bug localization model that works across languages and software projects. It enhances source-bug alignment using **dynamic chunking** and **hard example learning**, enabling precise bug localization in unseen codebases and programming languages.
|
| 23 |
+
|
| 24 |
+
[](https://doi.org/10.1109/TSE.2025.3579574)
|
| 25 |
+
[](https://zenodo.org/records/15122980)
|
| 26 |
+
|
| 27 |
+
---
|
| 28 |
+
|
| 29 |
+
## ✨ Highlights
|
| 30 |
+
|
| 31 |
+
* 📌 **Cross-project & cross-language** bug localization with no re-training
|
| 32 |
+
* 📏 **Dynamic Chunking** handles long files within LLM context windows
|
| 33 |
+
* 🧠 **Hard Example Learning** improves generalization and ranking accuracy
|
| 34 |
+
* 🌍 Supports Java, Python, C++, JavaScript, and Go
|
| 35 |
+
* 📊 Outperforms both cross-project and embedding-based baselines
|
| 36 |
+
|
| 37 |
+
---
|
| 38 |
+
|
| 39 |
+
## 📂 Dataset: BeetleBox
|
| 40 |
+
|
| 41 |
+
**BeetleBox** is the largest curated dataset for bug localization:
|
| 42 |
+
|
| 43 |
+
* 23,782 real-world bugs
|
| 44 |
+
* 29 repositories
|
| 45 |
+
* 5 programming languages
|
| 46 |
+
* Cleaned and de-duplicated to remove overlaps with training data
|
| 47 |
+
|
| 48 |
+
📥 [Available on Zenodo](https://zenodo.org/records/15122980)
|
| 49 |
+
📚 Also listed on Hugging Face Datasets: `bug-localization/BeetleBox`
|
| 50 |
+
|
| 51 |
+
---
|
| 52 |
+
|
| 53 |
+
## 🚀 Demo & Usage
|
| 54 |
+
|
| 55 |
+
All code, usage instructions, model files, and scripts are available via:
|
| 56 |
+
|
| 57 |
+
👉 **[BLAZE Repository & Demo (Zenodo)](https://zenodo.org/records/15122980)**
|
| 58 |
+
|
| 59 |
+
---
|
| 60 |
+
|
| 61 |
+
## 📝 Citation
|
| 62 |
+
|
| 63 |
+
Please cite the following paper if you use BLAZE or BeetleBox in your work:
|
| 64 |
+
|
| 65 |
+
```bibtex
|
| 66 |
+
@article{Chakraborty2025,
|
| 67 |
+
title = {BLAZE: Cross-Language and Cross-Project Bug Localization via Dynamic Chunking and Hard Example Learning},
|
| 68 |
+
ISSN = {2326-3881},
|
| 69 |
+
url = {http://dx.doi.org/10.1109/TSE.2025.3579574},
|
| 70 |
+
DOI = {10.1109/TSE.2025.3579574},
|
| 71 |
+
journal = {IEEE Transactions on Software Engineering},
|
| 72 |
+
publisher = {Institute of Electrical and Electronics Engineers (IEEE)},
|
| 73 |
+
author = {Chakraborty, Partha and Alfadel, Mahmoud and Nagappan, Meiyappan},
|
| 74 |
+
year = {2025},
|
| 75 |
+
pages = {1--14}
|
| 76 |
+
}
|
| 77 |
+
```
|