license: mit
task_categories:
- image-to-text
- text-to-image
language:
- en
pretty_name: simons ARC (abstraction & reasoning corpus) solve template version 9
size_categories:
- 1K<n<10K
configs:
- config_name: default
data_files:
- split: train
path: data.jsonl
Version 1
ARC-AGI Tasks where the job is to insert objects into templates.
example count: 2-3.
test count: 1-2.
image size: 8-10.
template size: 2-4.
number of rects: 2-4.
Version 2
Smaller images.
example count: 2-3.
test count: 1-2.
image size: 6-8.
template size: 2-2.
number of rects: 2-3.
Version 3
Added transformation: without_insertion_image
image size: 6-8.
template size: 2-3.
number of rects: 2-3.
Version 4
Added transformations: swap_one_to_many and swap_many_to_one.
image size: 6-10.
template size: 2-3.
number of rects: 2-3.
Version 5
image size: 6-12.
template size: 2-4.
number of rects: 2-3.
Version 6
Focus only on: without_insertion_image
image size: 6-10.
template size: 2-4.
number of rects: 2-3.
Version 7
Focus only on: without_insertion_image
image size: 6-12.
template size: 2-4.
number of rects: 2-3.
Version 8
Focus only on swap_one_to_many and swap_many_to_one.
image size: 6-10.
template size: 2-3.
number of rects: 2-3.
Version 9
Focus only on swap_many_to_one.
image size: 6-10.
template size: 2-3.
number of rects: 2-3.
Version 10
Focus only on swap_many_to_one.
image size: 6-12.
template size: 2-4.
number of rects: 2-3.