Dataset Viewer
Auto-converted to Parquet Duplicate
id
string
task
string
category
string
image
string
image_sha256
string
image_source_filename
string
question
string
answer
unknown
AGC-0
attribute_group_counting
distributed_scanning
images/AGC-0.png
d82f8b770b02791aa925d5f6e4c085fb2241574f0d6831e2bf4cd2271cd0b253
attribute_group_counting_00000.gptimage_v1_stones.png
How many distinct shapes are in the image? The image contains many small pieces scattered across the canvas. Two pieces belong to the same type if and only if they have exactly the same silhouette. Pieces may differ in surface texture or colour tone — only the silhouette outline matters. Count the total number of disti...
3
AGC-1
attribute_group_counting
distributed_scanning
images/AGC-1.png
50ff27f0e307d0352e1eeb5601acbf2ba24bbf9cd169d7caab86a47cd9cc663b
attribute_group_counting_00001.gptimage_v2_cookies.png
How many distinct shapes are in the image? The image contains many small pieces scattered across the canvas. Two pieces belong to the same type if and only if they have exactly the same silhouette. Pieces may differ in surface texture or colour tone — only the silhouette outline matters. Count the total number of disti...
5
AGC-2
attribute_group_counting
distributed_scanning
images/AGC-2.png
7d05fa89c6a44d47b43b9f7c24fe3af4cf22c2b126dc1596a9ee0960981fb101
attribute_group_counting_00002.gptimage_v3_leaves.png
How many distinct shapes are in the image? The image contains many small pieces scattered across the canvas. Two pieces belong to the same type if and only if they have exactly the same silhouette. Pieces may differ in surface texture or colour tone — only the silhouette outline matters. Count the total number of disti...
6
AGC-3
attribute_group_counting
distributed_scanning
images/AGC-3.png
bd07263069e58898db048120e93e1079057e7e5e4de039f70e1c199df394c6b5
attribute_group_counting_00003.gptimage_v4_teacups.png
How many distinct shapes are in the image? The image contains many small pieces scattered across the canvas. Two pieces belong to the same type if and only if they have exactly the same silhouette. Pieces may differ in surface texture or colour tone — only the silhouette outline matters. Count the total number of disti...
8
AGC-4
attribute_group_counting
distributed_scanning
images/AGC-4.png
d0611a29bce166345a4409bd34c1158840e4ffc98b9214ad154b9357f5bd4ea1
attribute_group_counting_00004.gptimage_v5_claytiles.png
How many distinct shapes are in the image? The image contains many small pieces scattered across the canvas. Two pieces belong to the same type if and only if they have exactly the same silhouette. Pieces may differ in surface texture or colour tone — only the silhouette outline matters. Count the total number of disti...
10
BFC-0
bounded_faces_counting
distributed_scanning
images/BFC-0.png
3a249fbde7c1911df0d5f2f1f6fa294188c94a8517dbd20966d1d8bcc2d1efaa
bounded_faces_counting_00000.gptimage_v1_pool1_driftwood.png
How many distinct enclosed regions (bounded faces) are visible in this image? An enclosed region is a maximal area that is fully surrounded by drawn lines or strands on every side, with no opening to the outside background. The unbounded outside area does not count. Each enclosed region should be counted exactly once. ...
5
BFC-1
bounded_faces_counting
distributed_scanning
images/BFC-1.png
1a271517410294e0cef4642f2a4244fb06b8725953e0d2a988ed670c1fe3f880
bounded_faces_counting_00001.gptimage_v2_pool2_rebar.png
How many distinct enclosed regions (bounded faces) are visible in this image? An enclosed region is a maximal area that is fully surrounded by drawn lines or strands on every side, with no opening to the outside background. The unbounded outside area does not count. Each enclosed region should be counted exactly once. ...
8
BFC-2
bounded_faces_counting
distributed_scanning
images/BFC-2.png
252f724e27d6d8b6f95764c25efc26b05c3e4d979b870c22f4fe768a9a683b92
bounded_faces_counting_00002.gptimage_v3_pool3_mossstone.png
How many distinct enclosed regions (bounded faces) are visible in this image? An enclosed region is a maximal area that is fully surrounded by drawn lines or strands on every side, with no opening to the outside background. The unbounded outside area does not count. Each enclosed region should be counted exactly once. ...
10
BFC-3
bounded_faces_counting
distributed_scanning
images/BFC-3.png
9a1ce169b39704bb71e74d7236505504bf7bc786a6e28f8d2b9430d623fd73eb
bounded_faces_counting_00003.gptimage_v4_pool4_tentstakes.png
How many distinct enclosed regions (bounded faces) are visible in this image? An enclosed region is a maximal area that is fully surrounded by drawn lines or strands on every side, with no opening to the outside background. The unbounded outside area does not count. Each enclosed region should be counted exactly once. ...
12
BFC-4
bounded_faces_counting
distributed_scanning
images/BFC-4.png
a0e9da7e438cb5ca0f21a5d734e95f3d3b199f2dad4df7066eb65d7bfc3ea729
bounded_faces_counting_00004.gptimage_v5_pool5_telegraph.png
How many distinct enclosed regions (bounded faces) are visible in this image? An enclosed region is a maximal area that is fully surrounded by drawn lines or strands on every side, with no opening to the outside background. The unbounded outside area does not count. Each enclosed region should be counted exactly once. ...
15
CCC-0
counting_connected_components
distributed_scanning
images/CCC-0.png
456cf0bd3b3a832dd6b71e1610e1c324e5ffbc4bd5cdf242e89b3b74989c8cf4
counting_connected_components_00000.gptimage_v1_driftwood.png
How many connected components are there in the image? A connected component is a maximal group of anchor points such that any two are linked by a path of one or more strands (directly or through other anchors in the same group). Anchors not linked by any chain belong to different components, even if they appear visuall...
10
CCC-1
counting_connected_components
distributed_scanning
images/CCC-1.png
bd91dfb264fe66cb942e6da65048f8f745f542a7f648b38728603434669d720c
counting_connected_components_00001.gptimage_v2_rebar.png
How many connected components are there in the image? A connected component is a maximal group of anchor points such that any two are linked by a path of one or more strands (directly or through other anchors in the same group). Anchors not linked by any chain belong to different components, even if they appear visuall...
12
CCC-2
counting_connected_components
distributed_scanning
images/CCC-2.png
b42a2ac6131b69280700b9f615abe5108b0ee4601d6e819c8ab05defd3c0ff30
counting_connected_components_00002.gptimage_v3_mossstone.png
How many connected components are there in the image? A connected component is a maximal group of anchor points such that any two are linked by a path of one or more strands (directly or through other anchors in the same group). Anchors not linked by any chain belong to different components, even if they appear visuall...
15
CCC-3
counting_connected_components
distributed_scanning
images/CCC-3.png
c236ae836fd47ef1d5795a6d9d110d8fb20900004eb0f29482ad03e4042a9efc
counting_connected_components_00003.gptimage_v4_tentstakes.png
How many connected components are there in the image? A connected component is a maximal group of anchor points such that any two are linked by a path of one or more strands (directly or through other anchors in the same group). Anchors not linked by any chain belong to different components, even if they appear visuall...
18
CCC-4
counting_connected_components
distributed_scanning
images/CCC-4.png
a326a583f1a1d5c3a40f71e0f7ae157cb404a16bc42284fcc97f6c2ea02b360e
counting_connected_components_00004.gptimage_v5_telegraph.png
How many connected components are there in the image? A connected component is a maximal group of anchor points such that any two are linked by a path of one or more strands (directly or through other anchors in the same group). Anchors not linked by any chain belong to different components, even if they appear visuall...
20
CR-0
counting_regions
distributed_scanning
images/CR-0.png
0ec1b5a70118a429ba78f324a46d7546a233ff36348cba4ac616ef52ce998b8f
counting_regions_00000.gptimage_v8_animaltrails.png
How many separated regions are inside the image? A region is a maximal area filled with one continuous fill. Boundaries between regions may appear as continuous lines OR as a chain of small individual markers — discrete dots, footprints, dashes, or short segments aligned along a curve — that you must visually link toge...
10
CR-1
counting_regions
distributed_scanning
images/CR-1.png
9c33ca1e80167e758322379f390c3941c443d58ba49c28f9f40e7aee97ca0e58
counting_regions_00001.gptimage_v9_footprintsnow.png
How many separated regions are inside the image? A region is a maximal area filled with one continuous fill. Boundaries between regions may appear as continuous lines OR as a chain of small individual markers — discrete dots, footprints, dashes, or short segments aligned along a curve — that you must visually link toge...
12
CR-2
counting_regions
distributed_scanning
images/CR-2.png
1dd4b90d468b151f6cb322321b59de35cb50f7d7c0bb2a91433b488df3a1461d
counting_regions_00002.gptimage_v13_antmarches.png
How many separated regions are inside the image? A region is a maximal area filled with one continuous fill. Boundaries between regions may appear as continuous lines OR as a chain of small individual markers — discrete dots, footprints, dashes, or short segments aligned along a curve — that you must visually link toge...
15
CR-3
counting_regions
distributed_scanning
images/CR-3.png
cc2bce4441e9fdd6a9583915dd337926d73a7ca0dc1f14d9ddf16314ef4dae58
counting_regions_00003.gptimage_v14_parchmentink.png
How many separated regions are inside the image? A region is a maximal area filled with one continuous fill. Boundaries between regions may appear as continuous lines OR as a chain of small individual markers — discrete dots, footprints, dashes, or short segments aligned along a curve — that you must visually link toge...
18
CR-4
counting_regions
distributed_scanning
images/CR-4.png
40460bb780fc5b2a74edf666b725b8895179e97dfdaf5663a6563be906c4e572
counting_regions_00004.gptimage_v15_beachrope.png
How many separated regions are inside the image? A region is a maximal area filled with one continuous fill. Boundaries between regions may appear as continuous lines OR as a chain of small individual markers — discrete dots, footprints, dashes, or short segments aligned along a curve — that you must visually link toge...
20
TL-0
tangled_loops
distributed_scanning
images/TL-0.png
39d650e832cc6edd4af73344d31e9c5fbffdd9fabf903b290c7cfa10707ce1c0
tangled_loops_00000.gptimage_v1_pool1_oldrope.png
How many distinct closed loops are tangled together in this image? Each loop is a continuous closed curve (rope, cord, cable, etc.) with no loose endpoints anywhere. Loops may cross other loops or themselves freely. Count the total number of distinct closed loops and report the count as a positive integer. Provide your...
10
TL-1
tangled_loops
distributed_scanning
images/TL-1.png
8a4873a9620ebf807e52df9d86aa31de3810c283f4541e97553a74058d766511
tangled_loops_00001.gptimage_v2_pool2_cords.png
How many distinct closed loops are tangled together in this image? Each loop is a continuous closed curve (rope, cord, cable, etc.) with no loose endpoints anywhere. Loops may cross other loops or themselves freely. Count the total number of distinct closed loops and report the count as a positive integer. Provide your...
12
TL-2
tangled_loops
distributed_scanning
images/TL-2.png
7614b2cc36b63e66cf47b0af3780f98c829a9a7e58c54309bfe30a7fa66b7cef
tangled_loops_00002.gptimage_v3_pool3_chaos.png
How many distinct closed loops are tangled together in this image? Each loop is a continuous closed curve (rope, cord, cable, etc.) with no loose endpoints anywhere. Loops may cross other loops or themselves freely. Count the total number of distinct closed loops and report the count as a positive integer. Provide your...
15
TL-3
tangled_loops
distributed_scanning
images/TL-3.png
1d3d0b0d031949467381b78eba6b3e7ae38f49d3a98281b484018c2c47a5bb0c
tangled_loops_00003.gptimage_v4_pool4_climbing.png
How many distinct closed loops are tangled together in this image? Each loop is a continuous closed curve (rope, cord, cable, etc.) with no loose endpoints anywhere. Loops may cross other loops or themselves freely. Count the total number of distinct closed loops and report the count as a positive integer. Provide your...
18
TL-4
tangled_loops
distributed_scanning
images/TL-4.png
1f425f36465b1affbabf42d6b4368f2d5d229bf3ea8f0c231ae2384e0710bd7f
tangled_loops_00004.gptimage_v5_pool5_stagecables.png
How many distinct closed loops are tangled together in this image? Each loop is a continuous closed curve (rope, cord, cable, etc.) with no loose endpoints anywhere. Loops may cross other loops or themselves freely. Count the total number of distinct closed loops and report the count as a positive integer. Provide your...
20
AC-0
arrow_chain
sequential_traversal
images/AC-0.png
7417166a81381ac73660a2764bda82600a82ab566b3395df2219ad25d2db4e36
arrow_chain_00000.gptimage_v1_foot.png
The image shows many small pieces scattered across the canvas, each clearly oriented in a specific direction, plus several larger labeled terminus markers (each marker bears a distinct letter). The piece highlighted by a green ring is the starting point. From that piece, cast an infinitely thin ray (a mathematical half...
"H"
AC-1
arrow_chain
sequential_traversal
images/AC-1.png
3e58261808a2e435d27e9c008860796489e37647458cc30bd25bcbc610dd7f27
arrow_chain_00001.gptimage_v2_fish.png
The image shows many small pieces scattered across the canvas, each clearly oriented in a specific direction, plus several larger labeled terminus markers (each marker bears a distinct letter). The piece highlighted by a green ring is the starting point. From that piece, cast an infinitely thin ray (a mathematical half...
"B"
AC-2
arrow_chain
sequential_traversal
images/AC-2.png
1b9f82255b54c3c326ef98f0f9b5e274da5854b9dc1d87a1b2ba4d438a2d89ac
arrow_chain_00002.gptimage_v3_bird.png
The image shows many small pieces scattered across the canvas, each clearly oriented in a specific direction, plus several larger labeled terminus markers (each marker bears a distinct letter). The piece highlighted by a green ring is the starting point. From that piece, cast an infinitely thin ray (a mathematical half...
"B"
AC-3
arrow_chain
sequential_traversal
images/AC-3.png
dcdf8a1800e5dfca85ea527e178298d17030edc4003bfcf2092e8ee576fa8047
arrow_chain_00003.gptimage_v4_key.png
The image shows many small pieces scattered across the canvas, each clearly oriented in a specific direction, plus several larger labeled terminus markers (each marker bears a distinct letter). The piece highlighted by a green ring is the starting point. From that piece, cast an infinitely thin ray (a mathematical half...
"J"
AC-4
arrow_chain
sequential_traversal
images/AC-4.png
951a6ebde1307c2eca593506d25aac4c044226508422d80375ed96492c74dd9c
arrow_chain_00004.gptimage_v5_leaf.png
The image shows many small pieces scattered across the canvas, each clearly oriented in a specific direction, plus several larger labeled terminus markers (each marker bears a distinct letter). The piece highlighted by a green ring is the starting point. From that piece, cast an infinitely thin ray (a mathematical half...
"C"
CZ-0
color_zone_sequence
sequential_traversal
images/CZ-0.png
88b50223c38c896244842a8f41d0282c683e4abe5d4b8280c8167d9ef242de45
color_zone_00000.gptimage_demo.png
The image shows several distinct regions, each carrying a single-letter label and distinguished by its own fill (colour, texture, or material). A single smooth curve starts at the position marked 'Start' and travels through several regions, potentially re-entering earlier ones. Follow the curve from start to end and li...
"H, D, F, B, I, A, C, J, A, J, H"
CZ-1
color_zone_sequence
sequential_traversal
images/CZ-1.png
b4cde60a8e87fd1a7328d35e0dc0d9c302fe6a38f74f086f2a2ee4866dd2a4b6
color_zone_00001.gptimage_v2_nautical.png
The image shows several distinct regions, each carrying a single-letter label and distinguished by its own fill (colour, texture, or material). A single smooth curve starts at the position marked 'Start' and travels through several regions, potentially re-entering earlier ones. Follow the curve from start to end and li...
"B, D, C, G, C, E, I, A, J, F, C"
CZ-2
color_zone_sequence
sequential_traversal
images/CZ-2.png
7251a813552432d3bc01544cc3b2fe25014b2b47d9c79a95a00f0dd6ae139032
color_zone_00002.gptimage_v3_storybook.png
The image shows several distinct regions, each carrying a single-letter label and distinguished by its own fill (colour, texture, or material). A single smooth curve starts at the position marked 'Start' and travels through several regions, potentially re-entering earlier ones. Follow the curve from start to end and li...
"F, E, F, I, C, D, A, J, H, G, E"
CZ-3
color_zone_sequence
sequential_traversal
images/CZ-3.png
40f669624624fa3481909c9ed943d1d49aaa062fa85dd3f4cd047d1e792243b6
color_zone_00003.gptimage_v4_transit.png
The image shows several distinct regions, each carrying a single-letter label and distinguished by its own fill (colour, texture, or material). A single smooth curve starts at the position marked 'Start' and travels through several regions, potentially re-entering earlier ones. Follow the curve from start to end and li...
"G, B, D, G, A, I, C, I, A, G, D, F"
CZ-4
color_zone_sequence
sequential_traversal
images/CZ-4.png
8b1d2112f7125b7999981c58da367935bbe42b7a244aaf4b3463d5f3daca7059
color_zone_00004.gptimage_v5_floorplan.png
The image shows several distinct regions, each carrying a single-letter label and distinguished by its own fill (colour, texture, or material). A single smooth curve starts at the position marked 'Start' and travels through several regions, potentially re-entering earlier ones. Follow the curve from start to end and li...
"C, A, G, A, F, E, D, J, B, C, H, I, G"
LI-0
line_intersections
sequential_traversal
images/LI-0.png
7d7c33dd7d09f1b8abfc346a63b2dac969beb465d0f03f1301140cdfd492d133
line_intersections_00000.gptimage_v1_beach.png
The image shows several curves or strands tangled together. Each curve has a single labeled endpoint on the perimeter, marked with a distinct letter. Starting from the endpoint labeled A, follow that curve through the interior tangle. At every crossing you pass through, record the label of the OTHER curve involved — wh...
"C, C, D, C, D, D, C, C, C, B, C"
LI-1
line_intersections
sequential_traversal
images/LI-1.png
1504a6fd0ec76e2513a23ad71be4e274e197c470b3b3d9d5d282ed8d94439467
line_intersections_00001.gptimage_v2_workshop.png
The image shows several curves or strands tangled together. Each curve has a single labeled endpoint on the perimeter, marked with a distinct letter. Starting from the endpoint labeled A, follow that curve through the interior tangle. At every crossing you pass through, record the label of the OTHER curve involved — wh...
"D, B, B, C, C, B, B, B, C"
LI-2
line_intersections
sequential_traversal
images/LI-2.png
d95525d70708fd7213fdb5981ae9a5f522cb36fbe3f7340078243fdf9f7ac23a
line_intersections_00002.gptimage_v3_forest.png
The image shows several curves or strands tangled together. Each curve has a single labeled endpoint on the perimeter, marked with a distinct letter. Starting from the endpoint labeled A, follow that curve through the interior tangle. At every crossing you pass through, record the label of the OTHER curve involved — wh...
"D, C, C, B, B, D, B, D, C"
LI-3
line_intersections
sequential_traversal
images/LI-3.png
4328215d10303c8a3dfcf81b6b5fa5738cd46e51568766205b0d3d2e08b96751
line_intersections_00003.gptimage_v4_hoses.png
The image shows several curves or strands tangled together. Each curve has a single labeled endpoint on the perimeter, marked with a distinct letter. Starting from the endpoint labeled A, follow that curve through the interior tangle. At every crossing you pass through, record the label of the OTHER curve involved — wh...
"D, C, B, C, D, C, D, C, C"
LI-4
line_intersections
sequential_traversal
images/LI-4.png
ffcd8c76dad10674c3584a6fda965598cfb1b4c8dcf95311ccc3cd80be85f4df
line_intersections_00004.gptimage_v5_ivy.png
The image shows several curves or strands tangled together. Each curve has a single labeled endpoint on the perimeter, marked with a distinct letter. Starting from the endpoint labeled A, follow that curve through the interior tangle. At every crossing you pass through, record the label of the OTHER curve involved — wh...
"C, C, B, B, C, B, C, B, C, D, B"
MZ-0
maze
sequential_traversal
images/MZ-0.png
8c5d171a2367fe52fcb00d0b54f2da6b03bdb6895e5ffb17512d69e274e3d5fb
maze_00000.gptimage_v1_hedge.png
The image shows a maze with several labeled openings on its border. Exactly one pair of openings is connected by a path through the maze; all other pairs are blocked by walls. Identify the connected pair. A path exists only if you can travel from one opening to another through corridors without crossing any wall. Repor...
"E-F"
MZ-1
maze
sequential_traversal
images/MZ-1.png
30f0d0e0a9a0f5c63d119cafc794dff22632f61823194c47e1fd97de941e44e3
maze_00001.gptimage_v2_dungeon.png
The image shows a maze with several labeled openings on its border. Exactly one pair of openings is connected by a path through the maze; all other pairs are blocked by walls. Identify the connected pair. A path exists only if you can travel from one opening to another through corridors without crossing any wall. Repor...
"C-E"
MZ-2
maze
sequential_traversal
images/MZ-2.png
b94e9b199c054e6ff90c251be8ba9b0df17189dd9647bbdd61d5dc692942b12b
maze_00002.gptimage_v3_cornfield.png
The image shows a maze with several labeled openings on its border. Exactly one pair of openings is connected by a path through the maze; all other pairs are blocked by walls. Identify the connected pair. A path exists only if you can travel from one opening to another through corridors without crossing any wall. Repor...
"A-G"
MZ-3
maze
sequential_traversal
images/MZ-3.png
a165f4c4c84a16660f11aea536c50555f7637114973fc89d5283955a39ac4314
maze_00003.gptimage_v4_library.png
The image shows a maze with several labeled openings on its border. Exactly one pair of openings is connected by a path through the maze; all other pairs are blocked by walls. Identify the connected pair. A path exists only if you can travel from one opening to another through corridors without crossing any wall. Repor...
"C-G"
MZ-4
maze
sequential_traversal
images/MZ-4.png
08177b3d624709e9a53a733fec5efd8f1fcd43a1c3a74b2dc0ba8e4d6dcadb85
maze_00004.gptimage_v5_snowmaze.png
The image shows a maze with several labeled openings on its border. Exactly one pair of openings is connected by a path through the maze; all other pairs are blocked by walls. Identify the connected pair. A path exists only if you can travel from one opening to another through corridors without crossing any wall. Repor...
"B-C"
TO-0
traverse_ordering
sequential_traversal
images/TO-0.png
a55eaaab754a98dcc7c2a72c62aef5dcfd3a713a9d4d82d92e7642d855cc39c3
traverse_ordering_00000.gptimage_v1_beach.png
The image shows two curves or strands that cross each other. Several labeled points are marked on the two curves combined. A green 'S' marks the start endpoint of one of the curves. Starting from S, follow THAT curve (the one S lies on) all the way to its other endpoint, and list the labels of the marked points you vis...
"L, G, H, F, C, K"
TO-1
traverse_ordering
sequential_traversal
images/TO-1.png
802513182beca8954adb481bf7351958cb7c74909e80a94496bcc11cae8fd313
traverse_ordering_00001.gptimage_v2_workshop.png
The image shows two curves or strands that cross each other. Several labeled points are marked on the two curves combined. A green 'S' marks the start endpoint of one of the curves. Starting from S, follow THAT curve (the one S lies on) all the way to its other endpoint, and list the labels of the marked points you vis...
"J, I, A, L, B, E"
TO-2
traverse_ordering
sequential_traversal
images/TO-2.png
616488cce7610f749e86bc449415960cdba9a60d8c479a1a928bf87ba9a2cd36
traverse_ordering_00002.gptimage_v3_forest.png
The image shows two curves or strands that cross each other. Several labeled points are marked on the two curves combined. A green 'S' marks the start endpoint of one of the curves. Starting from S, follow THAT curve (the one S lies on) all the way to its other endpoint, and list the labels of the marked points you vis...
"I, A, E, F, C, H"
TO-3
traverse_ordering
sequential_traversal
images/TO-3.png
4f275ac1edda0bfc2581550d1d4bab670d15a050399e207f8b5776bf69e8b7c4
traverse_ordering_00003.gptimage_v4_hoses.png
The image shows two curves or strands that cross each other. Several labeled points are marked on the two curves combined. A green 'S' marks the start endpoint of one of the curves. Starting from S, follow THAT curve (the one S lies on) all the way to its other endpoint, and list the labels of the marked points you vis...
"B, H, D, A, F, G"
TO-4
traverse_ordering
sequential_traversal
images/TO-4.png
3d08aa963f8aa0560967a28a8f6bee9a27ff8a4994654ba215080bd8998f8625
traverse_ordering_00004.gptimage_v5_ivy.png
The image shows two curves or strands that cross each other. Several labeled points are marked on the two curves combined. A green 'S' marks the start endpoint of one of the curves. Starting from S, follow THAT curve (the one S lies on) all the way to its other endpoint, and list the labels of the marked points you vis...
"D, F, A, C, B, L"
CMC-0
constellation_match_count
visual_attribute_transfer
images/CMC-0.png
5f929a6fea5039fae7174bd6913c431052e366f284482c92496ef6def6b13877
constellation_match_count_00000.gptimage_v1_stars.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a small constellation of marker points. The right panel shows the Field: a larger scene with many marker points. Both panels are drawn at the SAME pixel scale. Count how many copies of the Template pattern appear in the F...
5
CMC-1
constellation_match_count
visual_attribute_transfer
images/CMC-1.png
18f56e2d6618ebc9a0d310b79d2cce471567ee9b2c4c9ca529aa1b2802691751
constellation_match_count_00001.gptimage_v2_gemstones.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a small constellation of marker points. The right panel shows the Field: a larger scene with many marker points. Both panels are drawn at the SAME pixel scale. Count how many copies of the Template pattern appear in the F...
6
CMC-2
constellation_match_count
visual_attribute_transfer
images/CMC-2.png
f39221f5dd4c3a532ad331cfb87a1e032e91f31115e8fe871b464f7fd423a54a
constellation_match_count_00002.gptimage_v3_lights.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a small constellation of marker points. The right panel shows the Field: a larger scene with many marker points. Both panels are drawn at the SAME pixel scale. Count how many copies of the Template pattern appear in the F...
8
CMC-3
constellation_match_count
visual_attribute_transfer
images/CMC-3.png
09de618818bd66d57a79d75bf18d2c1f85f731cb3db5f6da915cd8bf5e544a10
constellation_match_count_00003.gptimage_v4_candy.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a small constellation of marker points. The right panel shows the Field: a larger scene with many marker points. Both panels are drawn at the SAME pixel scale. Count how many copies of the Template pattern appear in the F...
9
CMC-4
constellation_match_count
visual_attribute_transfer
images/CMC-4.png
7cb4f3e981a45bca948a79e54c7a68aa9141c116d28436f47bda1e594b564485
constellation_match_count_00004.gptimage_v5_shells.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a small constellation of marker points. The right panel shows the Field: a larger scene with many marker points. Both panels are drawn at the SAME pixel scale. Count how many copies of the Template pattern appear in the F...
10
CSC-0
contour_silhouette_count
visual_attribute_transfer
images/CSC-0.png
b650e26df2081f8e62d66dafc33486cf90549a0e10d1660cda1ed83cc900ed05
contour_silhouette_count_00000.gptimage_v1_pool1_kitchen.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a single closed silhouette. The right panel shows the Field: many closed silhouettes scattered across the canvas, all at the SAME pixel scale as the Template. Count the number of Field silhouettes that are exact copies of...
5
CSC-1
contour_silhouette_count
visual_attribute_transfer
images/CSC-1.png
034e4e90dfd985f8dfc2e105cf3de8137b6e2cf2642566e855688a3cef29d70b
contour_silhouette_count_00001.gptimage_v2_pool2_jewellery.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a single closed silhouette. The right panel shows the Field: many closed silhouettes scattered across the canvas, all at the SAME pixel scale as the Template. Count the number of Field silhouettes that are exact copies of...
8
CSC-2
contour_silhouette_count
visual_attribute_transfer
images/CSC-2.png
e516d6b0daddb0bf1197ec78fe3b082d94f00ad321f8dd3a5ed4750c8ed70f4e
contour_silhouette_count_00002.gptimage_v3_pool3_fruits.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a single closed silhouette. The right panel shows the Field: many closed silhouettes scattered across the canvas, all at the SAME pixel scale as the Template. Count the number of Field silhouettes that are exact copies of...
10
CSC-3
contour_silhouette_count
visual_attribute_transfer
images/CSC-3.png
60ea7ea5a0166726afa95812d7c848dcd0d1d7faeab3cc7a5aa265024dd353b4
contour_silhouette_count_00003.gptimage_v4_pool4_industrial.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a single closed silhouette. The right panel shows the Field: many closed silhouettes scattered across the canvas, all at the SAME pixel scale as the Template. Count the number of Field silhouettes that are exact copies of...
12
CSC-4
contour_silhouette_count
visual_attribute_transfer
images/CSC-4.png
f38c3ecaf3ecaec06633806d8cfa1b1b67cb79effc7b89c2743ead0236ace0a1
contour_silhouette_count_00004.gptimage_v5_pool5_textile.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a single closed silhouette. The right panel shows the Field: many closed silhouettes scattered across the canvas, all at the SAME pixel scale as the Template. Count the number of Field silhouettes that are exact copies of...
15
SCD-0
spot_the_contour_diff
visual_attribute_transfer
images/SCD-0.png
3bf0b8dabd91d3cf952ecafa2818fd7758f2feeb6ffd54299bb471c83eb4005d
contour_diff_00000.gptimage_v1_pool1_kitchen.png
Two panels are shown side by side. Each panel contains the same set of closed silhouettes at the same positions. Some silhouettes in the right panel have a different shape from the corresponding silhouette in the left panel. Count the number of pairs whose shape differs between the two panels and report the integer cou...
5
SCD-1
spot_the_contour_diff
visual_attribute_transfer
images/SCD-1.png
14ab657991686763224af9c32d45ca07cc5cdb746bb037b71621da4bcab79d97
contour_diff_00001.gptimage_v2_pool2_jewellery.png
Two panels are shown side by side. Each panel contains the same set of closed silhouettes at the same positions. Some silhouettes in the right panel have a different shape from the corresponding silhouette in the left panel. Count the number of pairs whose shape differs between the two panels and report the integer cou...
8
SCD-2
spot_the_contour_diff
visual_attribute_transfer
images/SCD-2.png
80f6908839fbb8f8787b04e88634bfdaa3713a9f55030c63ee6f472178530096
contour_diff_00002.gptimage_v3_pool3_fruits.png
Two panels are shown side by side. Each panel contains the same set of closed silhouettes at the same positions. Some silhouettes in the right panel have a different shape from the corresponding silhouette in the left panel. Count the number of pairs whose shape differs between the two panels and report the integer cou...
10
SCD-3
spot_the_contour_diff
visual_attribute_transfer
images/SCD-3.png
731301ded0afe036b66e41d5696f970151032ebc25a22d594a3af272745d9067
contour_diff_00003.gptimage_v4_pool4_industrial.png
Two panels are shown side by side. Each panel contains the same set of closed silhouettes at the same positions. Some silhouettes in the right panel have a different shape from the corresponding silhouette in the left panel. Count the number of pairs whose shape differs between the two panels and report the integer cou...
12
SCD-4
spot_the_contour_diff
visual_attribute_transfer
images/SCD-4.png
b75e993758e0967a41def272396d62ab00a3969e5752c82fcb75dab13e399c3e
contour_diff_00004.gptimage_v5_pool5_textile.png
Two panels are shown side by side. Each panel contains the same set of closed silhouettes at the same positions. Some silhouettes in the right panel have a different shape from the corresponding silhouette in the left panel. Count the number of pairs whose shape differs between the two panels and report the integer cou...
15
SFD-0
spot_the_field_diff
visual_attribute_transfer
images/SFD-0.png
9bd58bde7410c52abb44e401103af4ab81fda4c9cce1f5e26f761ae64b4b5c93
field_diff_00000.gptimage_v1_corkboard.png
Three panels are shown side by side. The leftmost panel is an INDEX key: a grid of cells numbered in row-major order (top-left = 1, increasing left-to-right then top-to-bottom). The middle and right panels are 2D color fields, each divided into the same grid. Compare the middle (Field A) and right (Field B) panels cell...
"6, 7, 8, 11"
SFD-1
spot_the_field_diff
visual_attribute_transfer
images/SFD-1.png
1460a81f6979fb4c9cc91422f7b2ef5c28c9fad4b4ceafacf3f2f3c4785523d5
field_diff_00001.gptimage_v2_woodendesk.png
Three panels are shown side by side. The leftmost panel is an INDEX key: a grid of cells numbered in row-major order (top-left = 1, increasing left-to-right then top-to-bottom). The middle and right panels are 2D color fields, each divided into the same grid. Compare the middle (Field A) and right (Field B) panels cell...
6
SFD-2
spot_the_field_diff
visual_attribute_transfer
images/SFD-2.png
13b5ac75ddc28310e2796ebdf259765ca00f352df17d85eebb70f3754cea1575
field_diff_00002.gptimage_v3_clipboard.png
Three panels are shown side by side. The leftmost panel is an INDEX key: a grid of cells numbered in row-major order (top-left = 1, increasing left-to-right then top-to-bottom). The middle and right panels are 2D color fields, each divided into the same grid. Compare the middle (Field A) and right (Field B) panels cell...
"3, 4, 8, 9, 10, 12, 13"
SFD-3
spot_the_field_diff
visual_attribute_transfer
images/SFD-3.png
2b351ff7e5344eae263f953ab5a78d225b6df81596dd18b3eea24021e9d9a519
field_diff_00003.gptimage_v4_blanket.png
Three panels are shown side by side. The leftmost panel is an INDEX key: a grid of cells numbered in row-major order (top-left = 1, increasing left-to-right then top-to-bottom). The middle and right panels are 2D color fields, each divided into the same grid. Compare the middle (Field A) and right (Field B) panels cell...
"3, 5, 6, 7, 8, 9, 11, 13, 14, 15"
SFD-4
spot_the_field_diff
visual_attribute_transfer
images/SFD-4.png
2aec3568530c4eb4d83d84015818c3eaeb4f61c13e6ac2ac05029f715a16afbd
field_diff_00004.gptimage_v5_folder.png
Three panels are shown side by side. The leftmost panel is an INDEX key: a grid of cells numbered in row-major order (top-left = 1, increasing left-to-right then top-to-bottom). The middle and right panels are 2D color fields, each divided into the same grid. Compare the middle (Field A) and right (Field B) panels cell...
"1, 2, 3, 5, 7, 8, 9, 10, 11, 12, 13, 14"
SSD-0
spot_the_signal_diff
visual_attribute_transfer
images/SSD-0.png
5101f298b4eff88220e6556800c6aa5cdcbabb39f3d929ae1661aa735108c55b
signal_diff_00000.gptimage_v1_lab.png
Two 1D signals are shown side by side. Each signal's x-axis is divided into equal segments by thin vertical gridlines, and each segment is numbered above the gridlines (the two signals share the same numbering). Compare the left (Signal A) and right (Signal B) signals segment by segment. List the numbers of all segment...
"7, 8, 9"
SSD-1
spot_the_signal_diff
visual_attribute_transfer
images/SSD-1.png
cc3743c0ba8a6310a562f221e750d8d918ff305762a5bd1b3a44c149a3789252
signal_diff_00001.gptimage_v2_whiteboard.png
Two 1D signals are shown side by side. Each signal's x-axis is divided into equal segments by thin vertical gridlines, and each segment is numbered above the gridlines (the two signals share the same numbering). Compare the left (Signal A) and right (Signal B) signals segment by segment. List the numbers of all segment...
"2, 5, 7, 8, 10"
SSD-2
spot_the_signal_diff
visual_attribute_transfer
images/SSD-2.png
6050738551a82fb9659e42b7ebf5877a789794e1b7e127706dd338e0aceab086
signal_diff_00002.gptimage_v3_workbench.png
Two 1D signals are shown side by side. Each signal's x-axis is divided into equal segments by thin vertical gridlines, and each segment is numbered above the gridlines (the two signals share the same numbering). Compare the left (Signal A) and right (Signal B) signals segment by segment. List the numbers of all segment...
8
SSD-3
spot_the_signal_diff
visual_attribute_transfer
images/SSD-3.png
4d700c3e6be20af7fde17ab05616ebde070a00f5345385528ec686e7a16177a3
signal_diff_00003.gptimage_v4_monitor.png
Two 1D signals are shown side by side. Each signal's x-axis is divided into equal segments by thin vertical gridlines, and each segment is numbered above the gridlines (the two signals share the same numbering). Compare the left (Signal A) and right (Signal B) signals segment by segment. List the numbers of all segment...
"3, 4, 6, 9"
SSD-4
spot_the_signal_diff
visual_attribute_transfer
images/SSD-4.png
3387cf3ef346cd9cfebb554d0cabfc1980b0b768c1e1cee75a5c55c8fdd7b03c
signal_diff_00004.gptimage_v5_fabric.png
Two 1D signals are shown side by side. Each signal's x-axis is divided into equal segments by thin vertical gridlines, and each segment is numbered above the gridlines (the two signals share the same numbering). Compare the left (Signal A) and right (Signal B) signals segment by segment. List the numbers of all segment...
"2, 3, 5, 6, 10"
STD-0
spot_the_stroke_diff
visual_attribute_transfer
images/STD-0.png
df712b3d672531c1a4e96f6593ded0dfe760dc8b9748a7a3c6cee8dd6440270f
stroke_diff_00000.gptimage_v1_pool1_twigs.png
Two panels are shown side by side. Each panel contains the same number of elongated pieces (strands, twigs, threads, etc.) at the same positions. Some pieces in the right panel have a different shape from the corresponding piece in the left panel. Count the number of pairs whose shape differs between the two panels and...
5
STD-1
spot_the_stroke_diff
visual_attribute_transfer
images/STD-1.png
8af263de4e70c44aafed2a3db9c42bb2808d65d9a11c7c8a1452bbf71acbbb69
stroke_diff_00001.gptimage_v2_pool2_sewing.png
Two panels are shown side by side. Each panel contains the same number of elongated pieces (strands, twigs, threads, etc.) at the same positions. Some pieces in the right panel have a different shape from the corresponding piece in the left panel. Count the number of pairs whose shape differs between the two panels and...
8
STD-2
spot_the_stroke_diff
visual_attribute_transfer
images/STD-2.png
dd26b142ae3f324d713c4ebde1630d2cbeddcf0758682bf32b60dc2c6288de0a
stroke_diff_00002.gptimage_v3_pool3_noodles.png
Two panels are shown side by side. Each panel contains the same number of elongated pieces (strands, twigs, threads, etc.) at the same positions. Some pieces in the right panel have a different shape from the corresponding piece in the left panel. Count the number of pairs whose shape differs between the two panels and...
10
STD-3
spot_the_stroke_diff
visual_attribute_transfer
images/STD-3.png
bade76b8a07ab6805437b9952a3a438f88152fdfe446c2f2f2818c5047fc35e3
stroke_diff_00003.gptimage_v4_pool4_wire.png
Two panels are shown side by side. Each panel contains the same number of elongated pieces (strands, twigs, threads, etc.) at the same positions. Some pieces in the right panel have a different shape from the corresponding piece in the left panel. Count the number of pairs whose shape differs between the two panels and...
12
STD-4
spot_the_stroke_diff
visual_attribute_transfer
images/STD-4.png
8a526d7f5ab2f6429db542939c74abb109a8c0b651640291a7e1b0be0dcb91e7
stroke_diff_00004.gptimage_v5_pool5_organic.png
Two panels are shown side by side. Each panel contains the same number of elongated pieces (strands, twigs, threads, etc.) at the same positions. Some pieces in the right panel have a different shape from the corresponding piece in the left panel. Count the number of pairs whose shape differs between the two panels and...
15
SGC-0
stroke_gesture_count
visual_attribute_transfer
images/SGC-0.png
07cc07427e72e28ad81cf374d5f7029cb2c31177dc1bc8593615d8692b5913fe
stroke_gesture_count_00000.gptimage_v1_pool1_twigs.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a single elongated piece with a specific curvature profile. The right panel shows the Field: many elongated pieces of similar thickness scattered across the canvas. Both panels use the SAME pixel scale. Count the number o...
5
SGC-1
stroke_gesture_count
visual_attribute_transfer
images/SGC-1.png
2069c73768f3aa6d010048fdffab71b50af1584ba00bf6226341c9cfba8409d8
stroke_gesture_count_00001.gptimage_v2_pool2_sewing.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a single white brush stroke with a specific curvature profile. The right panel shows the Field: many white brush strokes scattered across the canvas, all at the same thickness. Both panels use the SAME pixel scale and str...
8
SGC-2
stroke_gesture_count
visual_attribute_transfer
images/SGC-2.png
d63c140840e20bec90d4b1fbe6d250784a2aba9db3bd523917fc94c9d2d8dd62
stroke_gesture_count_00002.gptimage_v3_pool3_noodles.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a single elongated piece with a specific curvature profile. The right panel shows the Field: many elongated pieces of similar thickness scattered across the canvas. Both panels use the SAME pixel scale. Count the number o...
10
SGC-3
stroke_gesture_count
visual_attribute_transfer
images/SGC-3.png
4d3e00a1593a3fc94182c43d5b2f4e0cfd06b710af44174f0d860b081f1807e2
stroke_gesture_count_00003.gptimage_v4_pool4_wire.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a single elongated piece with a specific curvature profile. The right panel shows the Field: many elongated pieces of similar thickness scattered across the canvas. Both panels use the SAME pixel scale. Count the number o...
12
SGC-4
stroke_gesture_count
visual_attribute_transfer
images/SGC-4.png
90fbdc38b1d0dcbbeab0ec4c1750303fb8058f5dcee9052ae3435e0e4855f1bb
stroke_gesture_count_00004.gptimage_v5_pool5_organic.png
This image has two panels separated by a thin vertical divider. The left panel shows the Template: a single elongated piece with a specific curvature profile. The right panel shows the Field: many elongated pieces of similar thickness scattered across the canvas. Both panels use the SAME pixel scale. Count the number o...
15

ActiveVision

An exam for active observers — a benchmark for diagnosing whether multimodal large language models can iteratively look at an image during reasoning, instead of compressing it into a fixed embedding once.

What's in this archive

Path Contents
data/images/ 85 photorealistic PNGs, the released benchmark images
data/manifest.json Canonical index — one record per instance with id, task, category, image, image_sha256, image_source_filename, question, answer
data/annotations/<task>.jsonl Per-task verification metadata (5 records per file × 17 tasks = 85). Includes the structural ground truth used to compute each answer (region adjacency, arrow chains, traversal paths, Hausdorff distances, etc.)
code/<category>/<task>/creation.py Seedable, deterministic generator. The released images at v0.4 are produced at --difficulty 4.
code/<category>/<task>/creation.md Per-task design and anti-shortcut spec (where present).
code/<category>/<task>/data.json Per-task definition: shared question text, answer format.
code/gpt_image_prompts.json One gpt-image-2 image-edit prompt per task, used to re-render the matplotlib structural draft as a photorealistic variant while preserving the discriminative structure.
code/scope.md Project specification: the three task families and the six shortcut classes the design defeats.
croissant.json Croissant 1.0 + Croissant-RAI 1.0 metadata for this dataset.
LICENSE CC BY 4.0.

Statistics

  • 85 instances, 17 tasks, 3 task families.
  • Distributed Scanning (25 instances, 5 tasks): attribute_group_counting, bounded_faces_counting, counting_connected_components, counting_regions, tangled_loops.
  • Sequential Traversal (25 instances, 5 tasks): arrow_chain, color_zone_sequence, line_intersections, maze, traverse_ordering.
  • Visual Attribute Transfer (35 instances, 7 tasks): constellation_match_count, contour_silhouette_count, spot_the_contour_diff, spot_the_field_diff, spot_the_signal_diff, spot_the_stroke_diff, stroke_gesture_count.

Loading

import json, pathlib
root = pathlib.Path("data_neurips2026")
manifest = json.loads((root / "data" / "manifest.json").read_text())
for item in manifest:
    image_path = root / item["image"]
    question = item["question"]
    gold     = item["answer"]

Generation pipeline

The pipeline is the artifact. Every benchmark image is produced in two deterministic stages:

  1. Geometric draft (matplotlib). creation.py --seed S --difficulty 4 lays out a structural specification (region partition, arrow positions, maze graph, brush-stroke field, etc.) and computes the answer in closed form. Output: a plain matplotlib PNG.
  2. Photorealistic re-render (gpt-image-2). The matplotlib draft is sent to OpenAI gpt-image-2 via the image-edit endpoint, with a per-task prompt from code/gpt_image_prompts.json that preserves silhouettes, positions, counts, and labels but replaces the surface material with a photorealistic style (stones on sand, hedge maze from above, starfield, etc.).

The released benchmark contains only the Stage-2 images. Held-out splits with unpublished seeds and additional difficulties can be regenerated from the included generators.

Responsible AI

See croissant.json for the full RAI block. Headlines:

  • Synthetic only: 100% synthetic. No human subjects, no PII, no real-world events.
  • Use cases: testing and validation. Not for training.
  • Limitations: small evaluation set; adversarial-by-design (not predictive of general vision-language ability); photorealistic re-renders depend on a closed-source service.
  • License: CC BY 4.0.

Validating the Croissant file

Before submission, validate croissant.json with the official Croissant validator at:

https://huggingface.co/spaces/JoaquinVanschoren/croissant-checker

(Run well in advance of any submission deadline — the doc warns of heavy load near deadlines.)

Citation

Anonymous. ActiveVision: An Exam for Active Observers.
NeurIPS 2026 Datasets and Benchmarks Track (under review).
Downloads last month
13