Skip to content

Latest commit

 

History

History

datasets

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 

Datasets for ARC (Abstraction and Reasoning Corpus)

For simplicity, most of the datasets are contained in a single repo.

These datasets all follow the same json format as the original ARC dataset.

ARC-AGI - The original ARC dataset

Mini-ARC

  • Repo: MINI-ARC
  • Authors: Subin Kim, Prin Phunyaphibarn, Donghyun Ahn, Sundong Kim
  • License: Apache-2.0

ConceptARC

  • Repo: ConceptARC
  • Authors: Arseny Moskvichev, Victor Vikram Odouard, Melanie Mitchell
  • License: Unspecified

1D-ARC

  • Repo: 1D-ARC
  • Authors: Yudong (Will) Xu, Wenhao Li, Pashootan Vaezipoor, Scott Sanner, Elias B. Khalil
  • License: MIT

Sequence_ARC

  • Repo: Sequence_ARC
  • Authors: Seedling123
  • License: Apache-2.0

ARC_synthetic_extend

PQA: Perceptual Question Answering

synth_riddles

ARC dataset tama

ARC dataset diva

ARC Community

  • Repo: arc-community/arc
  • Authors: Yannic Kilcher, Peter Nemeth, Jack Cole.
  • License: Unspecified

Optional ARC

What's special about this dataset? Counter-examples. This dataset enhances the original ARC dataset, so every test pair have gotten 8 examples of what the output should NOT be.

Task generators

RE-ARC: Reverse-Engineering the Abstraction and Reasoning Corpus

  • Repo: RE-ARC
  • Authors: Michael Hodel
  • License: MIT

ARC gym

  • Repo: ARC gym
  • Authors: Simon Ouellette
  • License: Unspecified

BabyARC

  • Repo: BabyARC
  • Authors: Zhengxuan Wu.
  • License: MIT

ARC Generative DSL

ARC-Generator

  • Repo: ARC-Generator
  • Authors: Krzysztof Czarnecki
  • License: Unspecified

Corrections to the original ARC dataset

Communicating Natural Programs to Humans and Machines