Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,89 @@
+---
+license: cc-by-4.0
+language:
+- en
+tags:
+- unity
+- unity6
+- csharp
+- code-review
+- bug-detection
+- llm-evaluation
+- code
+pretty_name: UnityBench (by 14Dimension Enterprise)
+task_categories:
+- text-generation
+size_categories:
+- n<1K
+configs:
+- config_name: review_bugs
+  data_files: data/review/bugs.jsonl
+- config_name: review_clean
+  data_files: data/review/clean.jsonl
+- config_name: knowledge
+  data_files: data/knowledge/*.jsonl
+---
+
+# UnityBench — by **14Dimension Enterprise**
+
+**A two-track evaluation for Unity 6 / C# coding models: domain *knowledge* + real-world *bug review*.**
+
+Generalist code models are strong — but how well do they actually handle **Unity 6**, and can they
+find **real bugs in real game code** without crying wolf? There was no good public answer. This is one.
+
+Built by **14Dimension Enterprise** from a shipping Unity TRPG — bugs and fixes reconstructed from real
+version-control history. **Code:** https://github.com/leegod/unitybench
+
+---
+
+## Tracks
+
+**1. Knowledge** (`data/knowledge/`, ~146 items) — Unity 6 lifecycle/API Q&A + short
+"write this MonoBehaviour" compile tasks. Tests whether a model actually knows *current* Unity 6
+(post-cutoff API, common pitfalls).
+
+**2. Review** (`data/review/`, 24 buggy + 22 clean) — real Unity **gameplay** methods, each in two
+states: the **pre-fix (buggy)** version and the **post-fix (clean)** version, reconstructed from
+version-control history. Measures two things at once:
+- **Recall** — does it flag the *real* bug? (`review_bugs`)
+- **Specificity** — does it stay quiet on already-fixed/clean code instead of crying wolf? (`review_clean`)
+
+## 🏆 Leaderboard — Review track
+
+Bug detection on real Unity gameplay code (24 bugs / 22 clean). Recall = real bugs flagged.
+Quiet = clean methods left alone. Balance = harmonic mean.
+
+| Model | Recall | Quiet (specificity) | Balance |
+|---|---|---|---|
+| **Qwen3-Coder + project-context (RAG)** | 67% | **82%** | **73** |
+| Qwen3-Coder (open, no context) | 92% | 23% | 36 |
+| Gemini 2.5 Pro | 67% | 18% | 29 |
+| Claude Sonnet 4.5 | 46% | 18% | 26 |
+| GPT-5.5 | 100% | 9% | 17 |
+
+**What jumps out:** every frontier model over-flags on real game code (high recall, low
+specificity); injecting the project's own conventions (RAG) is the single biggest lever — it nearly
+**4×'d** the balance of the same base model. Binary recall/specificity is necessary but not
+sufficient — see the [repo README](https://github.com/leegod/unitybench) for the judge-based
+precision analysis.
+
+## Fields
+
+- `review_bugs`: `id`, `code` (buggy method), `bug` (description), `severity`, `context_needed`
+- `review_clean`: `id`, `code` (fixed method)
+- `knowledge`: Unity 6 Q&A / compile tasks
+
+## License & citation
+
+**Data:** CC-BY-4.0 — free to use, including commercially, **with attribution to 14Dimension Enterprise**.
+
+```bibtex
+@misc{unitybench2026,
+  title  = {UnityBench: Unity 6 Knowledge \& Real-World Bug-Review Evaluation},
+  author = {14Dimension Enterprise},
+  year   = {2026},
+  url    = {https://github.com/leegod/unitybench}
+}
+```
+
+— Built and measured hands-on by **14Dimension Enterprise** while shipping a Unity game.