README.md

---
license: mit
language:
- en
pretty_name: ARC-Bench
size_categories:
- n<1K
task_categories:
- other
tags:
- autonomous-research
- ai-agents
- llm-agents
- scientific-discovery
- benchmark
- machine-learning
- high-energy-physics
- quantum-computing
- systems-biology
- statistics
configs:
- config_name: default
  data_files:
  - split: test
    path: data/arc_bench.jsonl
- config_name: ml
  data_files:
  - split: test
    path: data/ml.jsonl
- config_name: physics
  data_files:
  - split: test
    path: data/physics.jsonl
- config_name: quantum
  data_files:
  - split: test
    path: data/quantum.jsonl
- config_name: biology
  data_files:
  - split: test
    path: data/biology.jsonl
- config_name: statistics
  data_files:
  - split: test
    path: data/statistics.jsonl
---

<p align="center">
  <img src="assets/logo.png" width="700" alt="AutoResearchClaw Logo">
</p>

<h2 align="center"><b>ARC-Bench: An Open-Ended Autonomous-Research Benchmark Across Five Scientific Domains</b></h2>

<p align="center">
  The benchmark released with <a href="https://arxiv.org/abs/2605.20025"><i>AutoResearchClaw: Self-Reinforcing Autonomous Research with Human-AI Collaboration</i></a>.
</p>

<p align="center">
  <a href="https://arxiv.org/abs/2605.20025"><img src="https://img.shields.io/badge/arXiv-2605.20025-b31b1b?logo=arxiv&logoColor=white" alt="arXiv"></a>
  <a href="https://huggingface.co/papers/2605.20025"><img src="https://img.shields.io/badge/%F0%9F%A4%97%20Paper-2605.20025-yellow" alt="HF Paper"></a>
  <a href="https://github.com/aiming-lab/AutoResearchClaw"><img src="https://img.shields.io/badge/GitHub-AutoResearchClaw-181717?logo=github" alt="GitHub"></a>
  <a href="LICENSE"><img src="https://img.shields.io/badge/License-MIT-yellow.svg" alt="MIT License"></a>
</p>

---

**ARC-Bench** is a **55-topic, open-ended autonomous-research benchmark** spanning
five scientific domains. Each topic is not a fixed-input/fixed-output task — it is a
*research question* plus a structured briefing. A research agent (or a human) must
take a topic from question → experiment design → code → measurements → claims →
write-up, and the deliverable is graded against a weighted, multi-criteria rubric.

| Domain | Count | IDs | Typical execution |
|---|---|---|---|
| Machine learning | 25 | `ML01`–`ML25` | CPU, `numpy`/`scipy`/`sklearn`/`statsmodels` |
| High-energy physics | 10 | `P01`–`P10` | Lagrangian → MadGraph MC → analysis → figure (paper reproduction) |
| Quantum | 10 | `Q01`–`Q10` | CPU, Qiskit 2.x statevector / VQE / QML |
| Systems biology | 7 | `B01`–`B07` | Constraint-based modelling (COBRApy / BiGG) |
| Statistics | 3 | `S01`–`S03` | Simulation studies (`numpy`/`scipy`/`statsmodels`) |

The ML, quantum, and statistics topics are **open research questions** (the agent
designs the experiment); the physics topics are **published-paper reproductions**
(each scoped to a specific reference figure); the biology topics are
**constraint-based metabolic-modelling** studies.

## Load it

```python
from datasets import load_dataset

# all 55 topics
ds = load_dataset("AIMING-Lab-UNC/ARC-Bench", split="test")

# a single domain subset
ml = load_dataset("AIMING-Lab-UNC/ARC-Bench", "ml", split="test")

row = ds[0]
print(row["id"], row["title"])
print(row["metric_key"], row["metric_direction"])
```

Available config names: `default` (all 55), `ml`, `physics`, `quantum`,
`biology`, `statistics`.

## Schema

Each row describes one benchmark topic. Deeply-nested / variable-shape fields are
stored as **JSON-encoded strings** so the table schema is stable across all
domains; parse them with `json.loads`.

| Column | Type | Description |
|---|---|---|
| `id` | string | Topic id (`ML01`, `P03`, `Q07`, `B01`, `S02`, …) |
| `domain` | string | One of `ml` / `physics` / `quantum` / `biology` / `statistics` |
| `title` | string | Human-readable topic title |
| `topic` | string | One-line topic statement (from the domain registry) |
| `domains` | list[string] | Subfield tags (e.g. `["machine-learning","calibration"]`) |
| `arxiv_id` | string \| null | Source paper (physics reproductions; null for open questions) |
| `venue` | string | Benchmark venue label |
| `metric_key` | string | Headline metric name |
| `metric_direction` | string | `maximize` / `minimize` / `match_reference` |
| `gpu_required` | bool | Whether a GPU is needed (all topics are CPU-friendly → `false`) |
| `est_wall_clock_sec` | int | Rough single-run wall-clock budget |
| `synthesis` | string | The research briefing: background + what a credible study includes |
| `num_hypotheses` | int | Number of pre-registered hypotheses |
| `hypotheses` | string (JSON) | List of `{id, statement, measurable}` |
| `experiment_design` | string (JSON) | `research_question`, `conditions`, `baselines`, `metrics`, `datasets`, `compute_requirements` |
| `requirements` | string (JSON) | Agent-mode pass/fail gating items (physics + biology; `""` otherwise) |
| `rubric` | string (JSON) | Hierarchical weighted scoring rubric (code / execution / results buckets) |
| `rubric_num_leaves` | int | Number of leaf criteria in the rubric |
| `manifest_file` | string | Path to the raw manifest inside this repo (`tasks/…`) |
| `rubric_file` | string | Path to the raw rubric inside this repo (`tasks/…`) |

### Raw inputs

The flattened `data/*.jsonl` is convenient for `load_dataset`. The authoritative,
human-readable benchmark inputs are also shipped verbatim under `tasks/`:

```
tasks/
├── meta_paper_quality.json         # shared paper-quality meta-rubric (manual grading)
└── <domain>/
    ├── topics.yaml                 # the domain topic registry
    ├── manifests/<ID>.yaml         # full per-topic briefing
    └── rubrics/<ID>.json           # weighted scoring rubric
```

## How a topic is scored

Each topic carries a hierarchical rubric. For ML / quantum / statistics it has
three buckets — **Code Development**, **Code Execution**, **Result Analysis**
(weighted roughly 25 : 25 : 50). Physics and biology add a fourth
**Reproducibility** bucket. Leaf criteria are graded on scientific substance and
directional correctness of the evidence, not on rigid threshold matching (see each
rubric's `judging_note`).

A second, optional layer — `tasks/meta_paper_quality.json` — grades the **paper
output** (writing, code orchestration, figure quality, factual accuracy) and is
intended for manual / vision-equipped grading rather than fast automated scoring.

## Intended use

- Evaluating autonomous-research / AI-scientist agents end-to-end.
- Studying agent behavior across heterogeneous scientific domains with a *single*
  task format.
- As a stimulus set for human-in-the-loop or framework-comparison studies.

The runner harness, baseline adapters, and judges are **not** part of this dataset;
they live in the source repository (link below).

## Attribution

The benchmark glue (manifests, rubrics, registries) is the authors' own work. Some
domain pipelines are driven by **external Claude-Code agents**, which should be
credited when reporting domain results:

| Topic family | External agent | Upstream |
|---|---|---|
| `P01`–`P10` (HEP) | ColliderAgent | <https://github.com/HET-AGI/ColliderAgent> |
| `B01`–`B07` (metabolic) | Biology-Agent | constraint-based modelling pipeline |

## Links

- 💻 **Code / harness:** <https://github.com/aiming-lab/AutoResearchClaw>
- 📄 **Paper (arXiv):** <https://arxiv.org/abs/2605.20025>
- 🤗 **Paper page:** <https://huggingface.co/papers/2605.20025>

## Citation

If you find ARC-Bench or AutoResearchClaw useful, please cite:

```bibtex
@misc{liu2026autoresearchclawselfreinforcingautonomousresearch,
      title={AutoResearchClaw: Self-Reinforcing Autonomous Research with Human-AI Collaboration},
      author={Jiaqi Liu and Shi Qiu and Mairui Li and Bingzhou Li and Haonian Ji and Siwei Han and Xinyu Ye and Peng Xia and Zihan Dong and Congyu Zhang and Letian Zhang and Guiming Chen and Haoqin Tu and Xinyu Yang and Lu Feng and Xujiang Zhao and Haifeng Chen and Jiawei Zhou and Xiao Wang and Weitong Zhang and Hongtu Zhu and Yun Li and Jieru Mei and Hongliang Fei and Jiaheng Zhang and Linjie Li and Linjun Zhang and Yuyin Zhou and Sheng Wang and Caiming Xiong and James Zou and Zeyu Zheng and Cihang Xie and Mingyu Ding and Huaxiu Yao},
      year={2026},
      eprint={2605.20025},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2605.20025},
}
```

## License

Released under the [MIT License](LICENSE).

<p align="center">
  <sub>Built with 🦞 by the AutoResearchClaw team</sub>
</p>