Datasets:
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -32,8 +32,8 @@ This provides a natural signal for training models to:
|
|
| 32 |
|
| 33 |
### Key Features
|
| 34 |
|
| 35 |
-
- **
|
| 36 |
-
- **40K+ negative examples** (~
|
| 37 |
- **36 programming languages** (Python, TypeScript, Go, Rust, C++, JavaScript, C#, Java, Kotlin, Swift, and more)
|
| 38 |
- **Quality-filtered**: bot comments, noise ("LGTM", "+1"), and auto-generated content removed
|
| 39 |
- **Chunk-focused**: ~50 lines of context around the reviewed code, not entire files
|
|
|
|
| 32 |
|
| 33 |
### Key Features
|
| 34 |
|
| 35 |
+
- **128K+ positive triplets** from 617 top GitHub repositories
|
| 36 |
+
- **40K+ negative examples** (~24% of dataset) of clean code labeled "No issues found."
|
| 37 |
- **36 programming languages** (Python, TypeScript, Go, Rust, C++, JavaScript, C#, Java, Kotlin, Swift, and more)
|
| 38 |
- **Quality-filtered**: bot comments, noise ("LGTM", "+1"), and auto-generated content removed
|
| 39 |
- **Chunk-focused**: ~50 lines of context around the reviewed code, not entire files
|