/ARC

Abstraction and Reasoning Corpus

Primary LanguageMathematicaMIT LicenseMIT

ARC

Abstraction and Reasoning Corpus

My attempt at:

https://github.com/fchollet/ARC

How it Works

How it Works

Percentage of Tasks Passing

  • Training tasks: 181 / 400 (45.2%)
    Percentage of Training Tasks Passing
    • Implemented: 93
    • Passing via generalization: 88
  • Evaluation tasks: 64 / 400 (16%)
    Percentage of Evaluation Tasks Passing

Progress over Time

Graph of Progress

Tasks Implemented

Core ARC Training Tasks (93)

Personally Created Training Tasks (16)

Tasks Passing via Generalization

Training Tasks (88)

The following ARC training tasks started passing after some different task was implemented.

Evaluation Tasks (64)

The following ARC evaluation tasks are passing. My intention is to not analyzed or implemented evaluation examples specifically, but I have implemented a few accidentally.