Reasoning

ARC-AGI

ARC Prize's released model attempts on the public evaluation tasks of ARC-AGI-1 and ARC-AGI-2, which test the inference of transformations from example grids.

526items

80subjects

ARC task definitions: Apache-2.0. ARC Prize result datasets and captured benchmarking harness: MIT.license

reasoningdomain

gridmodality

item-level responses released

Saturation status: Yes

Full data on Hugging Face Back to the gallery

Paper and credit

2019

On the Measure of Intelligence (opens in a new tab)

Original source

Author

François Chollet

Citation

BibTeX citationShow

BibTeX

1@misc{chollet2019measureintelligence,
2      title={On the Measure of Intelligence},
3      author={François Chollet},
4      year={2019},
5      eprint={1911.01547},
6      archivePrefix={arXiv},
7      primaryClass={cs.AI},
8      url={https://arxiv.org/abs/1911.01547},
9}

Response matrix

plus leaderboard aggregate scores

Loading response matrix…

lowhighUnobserved

Scale: Score in [0, 1]