Upload README.md with huggingface_hub

README.md

@@ -2,7 +2,6 @@
 license: mit
 task_categories:
   - text-generation
-  - text2text-generation
 language:
   - en
   - code
@@ -30,8 +29,8 @@ Each row captures a moment where a human code reviewer left an inline comment on
 
 ### Key Features
 
-- **500K+ triplets** from 10,000+ top GitHub repositories
-- **10+ programming languages** (Python, JavaScript, TypeScript, Java, Go, Rust, C++, C#, Ruby, PHP, and more)
+- **118K+ triplets** from 562 top GitHub repositories
+- **20+ programming languages** (Python, TypeScript, Go, C++, Rust, JavaScript, C#, Java, Kotlin, Swift, and more)
 - **Quality-filtered**: bot comments, noise ("LGTM", "+1"), and auto-generated content removed
 - **Chunk-focused**: ~50 lines of context around the reviewed code, not entire files
 - **Permissive licenses only**: all source repos use MIT, Apache-2.0, BSD, or similar licenses
@@ -65,7 +64,7 @@ Each row captures a moment where a human code reviewer left an inline comment on
 ```python
 from datasets import load_dataset
 
-ds = load_dataset("ronantakizawa/code-review-diffs")
+ds = load_dataset("ronantakizawa/github-codereview")
 
 # Get a training example
 example = ds["train"][0]
@@ -134,6 +133,6 @@ If you use this dataset, please cite:
   author={Takizawa, Ronan},
   year={2026},
   publisher={Hugging Face},
-  url={https://huggingface.co/datasets/ronantakizawa/code-review-diffs}
+  url={https://huggingface.co/datasets/ronantakizawa/github-codereview}
 }
 ```