File size: 1,742 Bytes
d5bfab8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
# ARC-Interactive-History-Dataset

This is a project that records human interactions while solving tasks from ARC, [Abstraction and Reasoning Corpus](https://github.com/fchollet/ARC) and similar datasets. This repository is populated with data via the [BrainGridGame](https://braingridgame.com/) website, where enthusiasts like you can challenge themselves.

## About the Project

This project aims to build a comprehensive dataset of interaction histories to facilitate ARC research. Each time a user solves a task on BrainGridGame, their interaction is anonymously recorded and contributes to this growing dataset, providing valuable insights into human problem-solving processes.

[video, 2 minutes, showing various ways of puzzle solving](https://www.youtube.com/watch?v=vQt7UZsYooQ).

Interaction history examples:
[05a7bcf2](https://neoneye.github.io/arc/history.html?dataset=ARC&task=05a7bcf2historyIndex=0&historyJson=%5B%2225%2F50%22%2C%2227%2F90%22%2C%228%2F70%22%5D), 
[08573cc6](https://neoneye.github.io/arc/history.html?dataset=ARC&task=08573cc6historyIndex=0&historyJson=%5B%227%2F33%22%5D), 
[3f23242b](https://neoneye.github.io/arc/history.html?dataset=ARC&task=3f23242bhistoryIndex=0&historyJson=[%224%2F39%22]).

## Contribute by solving tasks

To contribute to this dataset by solving tasks, you'll need to:

- Visit [BrainGridGame](https://braingridgame.com/)
- Sign in using your Google account (no password needed).
- Solve some of the harder tasks. There are already many solutions for the easiest tasks.

Every solved task helps enrich this dataset.

## Contact

For any questions or suggestions, feel free to open an issue here on GitHub or contact the project maintainer, Simon Strandgaard, at <[email protected]>.