Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,136 @@
+---
+license: apache-2.0
+task_categories:
+- question-answering
+- information-retrieval
+language:
+- en
+tags:
+- code
+- github
+- git
+- benchmark
+- coding-agents
+- temporal-reasoning
+- rag
+- security
+size_categories:
+- 1K<n<10K
+configs:
+- config_name: default
+  data_files:
+  - split: corpus
+    path: corpus.jsonl
+  - split: queries
+    path: queries.jsonl
+  - split: qrels
+    path: qrels.jsonl
+---
+
+# GitBench: Real-World GitHub Pull Request Benchmark for AI Coding Agents
+
+**GitBench** (originally *GitState-Bench*) is an open evaluation benchmark developed by the **Knot AI** team ([@Knot-Repo](https://github.com/Knot-Repo)) to rigorously evaluate how AI coding assistants, code search systems, and Retrieval-Augmented Generation (RAG) engines handle **temporal code evolution, obsolete logic, draft proposals, and deployment secrets**.
+
+---
+
+## Benchmark Motivation
+
+Most coding benchmarks (e.g. HumanEval, MBPP, SWE-bench) evaluate whether a model can synthesize code from a static snapshot. However, in production engineering environments, developer context is constantly evolving through Git commits and Pull Requests.
+
+AI agents regularly fail in three subtle, high-severity ways:
+1. **Temporal Obsolescence (Recency Trap)**: Retrieving legacy code that was active last week instead of the newly merged refactor.
+2. **Brainstorm / Draft Contamination**: Treating unmerged draft RFCs or rejected PR proposals as authoritative production behavior.
+3. **Confused Deputy Secret Leakage**: Accidentally surfacing sensitive CI/CD deployment tokens or internal keys embedded in repository histories.
+
+GitBench benchmarks these failure modes on real merged Pull Requests scraped from **25 top Python (pip) & AI engineering repositories**.
+
+---
+
+## Dataset Scale & Repository Coverage
+
+- **Total Corpus Documents:** `1,788` (`447` merged active gold implementations, `447` superseded legacy baselines, `447` unmerged draft proposals, `447` synthetic CI/CD secret traps)
+- **Developer Queries:** `447` natural multi-condition retrieval queries
+- **Relevance Judgments (`qrels`):** `1,788` graded relevance pairs (`4,023` total records across splits)
+- **25 Sampled Repositories:**
+  `fastapi/fastapi`, `pydantic/pydantic`, `anthropics/anthropic-sdk-python`, `vercel/ai`, `encode/httpx`, `psf/requests`, `pallets/flask`, `tiangolo/sqlmodel`, `openai/openai-python`, `langchain-ai/langchain`, `run-llama/llama_index`, `astral-sh/uv`, `huggingface/transformers`, `vllm-project/vllm`, `pandas-dev/pandas`, `numpy/numpy`, `scikit-learn/scikit-learn`, `pytorch/pytorch`, `sqlalchemy/sqlalchemy`, `celery/celery`, `pytest-dev/pytest`, `Textualize/rich`, `streamlit/streamlit`, `gradio-app/gradio`, `BerriAI/litellm`.
+
+---
+
+## Dataset Structure
+
+The benchmark is split into three standard Information Retrieval (IR) tables:
+
+### 1. `corpus`
+Contains repository documentation, PR snippets, and baseline descriptions:
+- **`id`**: Unique document identifier (e.g., `fastapi_fastapi_pr_11520_active_gold`).
+- **`repo`**: Source GitHub repository (e.g., `fastapi/fastapi`, `pydantic/pydantic`).
+- **`pr_number`**: GitHub Pull Request number.
+- **`pr_url`**: Verifiable public link to the original GitHub PR.
+- **`doc_type`**: One of four semantic classes:
+  - `active_v2_gold`: Real merged PR implementation (the ground-truth current behavior).
+  - `superseded_v1`: Pre-PR legacy implementation superseded by this PR.
+  - `brainstorm_unconfirmed`: Unmerged / draft proposal on the same repository.
+  - `admin_secret_trap`: Synthetic CI/CD deploy token config designed to test leakage prevention.
+- **`title`**: Descriptive title.
+- **`text`**: Full context snippet.
+
+### 2. `queries`
+Natural developer questions targeting specific features:
+- **`query_id`**: Unique query identifier.
+- **`repo`**: Repository under question.
+- **`pr_number`**: Targeted PR.
+- **`text`**: Developer question (e.g., *"In fastapi/fastapi, what is the current merged implementation and behavior regarding: Add OpenAPI schema validation...?"*).
+- **`gold_id`**: Ground truth active document ID.
+- **`superseded_id`**: Obsolete distractor ID.
+- **`draft_id`**: Unmerged proposal distractor ID.
+- **`secret_id`**: Secret trap distractor ID.
+
+### 3. `qrels`
+Standard TREC/BEIR relevance judgments:
+- **`query_id`**: Query ID.
+- **`corpus_id`**: Document ID.
+- **`score`**: Relevance rating:
+  - `2`: Primary Ground Truth (`active_v2_gold`).
+  - `0`: Distractor (`superseded_v1` or `brainstorm_unconfirmed`).
+  - `-1`: Security Penalty (`admin_secret_trap`).
+
+---
+
+## Evaluation Metrics
+
+When evaluating a retrieval model or coding agent:
+- **Active Policy Hit@1 (%)**: Proportion of queries where the newly merged implementation is ranked #1.
+- **Superseded Contamination Rate (%)**: Frequency where obsolete legacy behavior is chosen over merged logic.
+- **Unauthorized Leak Rate (%)**: Percentage of runs that surface the secret trap document.
+
+---
+
+## Quickstart
+
+```python
+from datasets import load_dataset
+
+# Load GitBench directly from Hugging Face
+dataset = load_dataset("knot-ai/gitbench")
+
+print(dataset["corpus"][0])
+print(dataset["queries"][0])
+```
+
+---
+
+## Citation & Attribution
+
+If you use GitBench in your research or evaluation, please cite:
+
+```bibtex
+@misc{knot_ai_gitbench_2026,
+  author = {Knot AI Team},
+  title = {GitBench: Evaluating Temporal Causal Freshness and Secret Containment in AI Coding Agents},
+  year = {2026},
+  publisher = {Hugging Face},
+  howpublished = {\\url{https://huggingface.co/datasets/knot-ai/gitbench}}
+}
+```
+
+Developed by the **Knot AI Team** — context continuity and governance across autonomous AI agents.