Initial release: README + Croissant + 5GB representative sample (216 shards, 36 tasks)
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- README.md +171 -0
- croissant.json +77 -0
- sample/data/metadata_shards/Multi-01_maze_shortest_path_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-02_maze_circular_route_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-03_maze_marker_drawing_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-04_dual_dot_pathfinding_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-05_maze_route_tracing_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-06_bfs_tree_traversal_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-07_sokoban_planning_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-08_sliding_puzzle_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-09_tower_of_hanoi_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-10_snake_dynamic_routing_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-11_tsp_reward_collection_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-12_ordinal_number_sequence_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-13_wordsearch_path_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-14_sudoku_logic_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-15_numbrix_pathfilling_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-16_orthogonal_latin_square_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-17_hashi_bridges_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-18_tents_and_trees_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-19_turing_machine_execution_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-20_langtons_ant_simulation_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-21_chained_math_calculation_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-22_pointer_chasing_arrows_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-23_chained_code_pipeline_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-24_conways_game_of_life_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-25_light_reflection_ray_tracing_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-26_line_intersection_construction_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-27_perpendicular_bisector_construction_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-28_triangle_orthocenter_construction_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-29_triangle_incenter_construction_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-30_triangle_circumcenter_construction_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-31_fluid_communicating_vessels_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-32_multiple_bounces_target_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-33_elastic_bouncing_trajectory_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-34_elastic_collision_kinematics_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-35_block_sliding_friction_data-generator.parquet +3 -0
- sample/data/metadata_shards/Multi-36_target_after_reflection_data-generator.parquet +3 -0
- sample/questions/Multi-01_maze_shortest_path_data-generator_00000-00049.tar.gz +3 -0
- sample/questions/Multi-01_maze_shortest_path_data-generator_00050-00099.tar.gz +3 -0
- sample/questions/Multi-01_maze_shortest_path_data-generator_00100-00149.tar.gz +3 -0
- sample/questions/Multi-01_maze_shortest_path_data-generator_00150-00199.tar.gz +3 -0
- sample/questions/Multi-01_maze_shortest_path_data-generator_00200-00249.tar.gz +3 -0
- sample/questions/Multi-01_maze_shortest_path_data-generator_00250-00299.tar.gz +3 -0
- sample/questions/Multi-02_maze_circular_route_data-generator_00000-00049.tar.gz +3 -0
- sample/questions/Multi-02_maze_circular_route_data-generator_00050-00099.tar.gz +3 -0
- sample/questions/Multi-02_maze_circular_route_data-generator_00100-00149.tar.gz +3 -0
- sample/questions/Multi-02_maze_circular_route_data-generator_00150-00199.tar.gz +3 -0
- sample/questions/Multi-02_maze_circular_route_data-generator_00200-00249.tar.gz +3 -0
- sample/questions/Multi-02_maze_circular_route_data-generator_00250-00299.tar.gz +3 -0
README.md
ADDED
|
@@ -0,0 +1,171 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-4.0
|
| 3 |
+
task_categories:
|
| 4 |
+
- video-text-to-text
|
| 5 |
+
language:
|
| 6 |
+
- en
|
| 7 |
+
size_categories:
|
| 8 |
+
- 100K<n<1M
|
| 9 |
+
pretty_name: VBVR-MultiStep
|
| 10 |
+
tags:
|
| 11 |
+
- video-reasoning
|
| 12 |
+
- multi-step
|
| 13 |
+
- long-horizon
|
| 14 |
+
- image-to-video
|
| 15 |
+
- training
|
| 16 |
+
- webdataset
|
| 17 |
+
---
|
| 18 |
+
|
| 19 |
+
# VBVR-MultiStep
|
| 20 |
+
|
| 21 |
+
The **~360k-sample programmatic training corpus** for long-horizon multi-step image-to-video (I2V) reasoning. Companion to the frozen [VBVR-MultiStep-Bench](https://huggingface.co/datasets/Video-Reason/VBVR-MultiStep-Bench) (180-instance evaluation split).
|
| 22 |
+
|
| 23 |
+
Part of the **VBVR (Very Big Video Reasoning Suite)** project: <https://video-reason.com>. See [Wang et al., ICML 2026](https://icml.cc/virtual/2026/poster/65709) for the parent suite.
|
| 24 |
+
|
| 25 |
+
## At a glance
|
| 26 |
+
|
| 27 |
+
| Property | Value |
|
| 28 |
+
|---|---|
|
| 29 |
+
| Tasks | **36** parameterized tasks (`Multi-01` … `Multi-36`) |
|
| 30 |
+
| Reasoning families | Navigation, Planning, CSP, Execution, Geometry, Physics |
|
| 31 |
+
| Total samples | **~360,000** (≈10k per task) |
|
| 32 |
+
| Total size | **~164 GB** |
|
| 33 |
+
| Format | Tar.gz shards (nested per-sample folders) + Parquet metadata |
|
| 34 |
+
| Shards | 7,200 (≈50 samples per shard) |
|
| 35 |
+
| License | CC-BY-4.0 |
|
| 36 |
+
|
| 37 |
+
## Repository layout
|
| 38 |
+
|
| 39 |
+
```
|
| 40 |
+
.
|
| 41 |
+
├── README.md
|
| 42 |
+
├── croissant.json # Croissant + RAI metadata
|
| 43 |
+
├── data/
|
| 44 |
+
│ ├── metadata.parquet # global index of all 360k samples
|
| 45 |
+
│ └── metadata_shards/
|
| 46 |
+
│ └── Multi-XX_<name>.parquet # per-task metadata (36 files)
|
| 47 |
+
├── questions/ # WebDataset shards
|
| 48 |
+
│ └── Multi-XX_<name>_NNNNN-NNNNN.tar.gz
|
| 49 |
+
│ └── (50 samples per shard, 5 files per sample, see "Sample format" below)
|
| 50 |
+
└── sample/ # ~5 GB representative subset for quick inspection
|
| 51 |
+
├── data/metadata_shards/...
|
| 52 |
+
└── questions/ # 6 shards × 36 tasks = 216 shards
|
| 53 |
+
```
|
| 54 |
+
|
| 55 |
+
The `sample/` subdirectory is a 5 GB pre-curated subset (the first 300 samples of every task) for reviewers and quick experimentation. To pull it:
|
| 56 |
+
|
| 57 |
+
```bash
|
| 58 |
+
huggingface-cli download Video-Reason/VBVR-MultiStep \
|
| 59 |
+
--repo-type dataset \
|
| 60 |
+
--include "sample/**" \
|
| 61 |
+
--local-dir ./vbvr-multistep-sample
|
| 62 |
+
```
|
| 63 |
+
|
| 64 |
+
## Sample format (inside each `.tar.gz` shard)
|
| 65 |
+
|
| 66 |
+
Each shard expands to a nested folder tree, identical in shape to the evaluation split:
|
| 67 |
+
|
| 68 |
+
```
|
| 69 |
+
Multi-XX_<name>_data-generator/
|
| 70 |
+
└── Multi-XX_<name>_data-generator_task/
|
| 71 |
+
└── Multi-XX_<name>_data-generator_<id>/
|
| 72 |
+
├── first_frame.png # conditioning frame
|
| 73 |
+
├── prompt.txt # natural-language task contract
|
| 74 |
+
├── final_frame.png # target endpoint (held-out at inference)
|
| 75 |
+
├── ground_truth.mp4 # reference rollout
|
| 76 |
+
└── question_metadata.json # seed, version, tolerances, task fields
|
| 77 |
+
```
|
| 78 |
+
|
| 79 |
+
Each shard contains 50 such instance folders. The five-artifact contract is identical to the evaluation split.
|
| 80 |
+
|
| 81 |
+
To extract:
|
| 82 |
+
|
| 83 |
+
```bash
|
| 84 |
+
tar xzf Multi-01_maze_shortest_path_data-generator_00000-00049.tar.gz
|
| 85 |
+
```
|
| 86 |
+
|
| 87 |
+
## Loading
|
| 88 |
+
|
| 89 |
+
### Per-task metadata (recommended entry point)
|
| 90 |
+
|
| 91 |
+
```python
|
| 92 |
+
import pandas as pd
|
| 93 |
+
m = pd.read_parquet(
|
| 94 |
+
"hf://datasets/Video-Reason/VBVR-MultiStep/data/metadata_shards/Multi-01_maze_shortest_path_data-generator.parquet"
|
| 95 |
+
)
|
| 96 |
+
print(m.head())
|
| 97 |
+
```
|
| 98 |
+
|
| 99 |
+
### Direct shard download
|
| 100 |
+
|
| 101 |
+
```python
|
| 102 |
+
from huggingface_hub import hf_hub_download
|
| 103 |
+
import tarfile
|
| 104 |
+
shard = hf_hub_download(
|
| 105 |
+
"Video-Reason/VBVR-MultiStep",
|
| 106 |
+
"questions/Multi-01_maze_shortest_path_data-generator_00000-00049.tar.gz",
|
| 107 |
+
repo_type="dataset",
|
| 108 |
+
)
|
| 109 |
+
with tarfile.open(shard) as t:
|
| 110 |
+
t.extractall("./extracted")
|
| 111 |
+
```
|
| 112 |
+
|
| 113 |
+
### Pull only the 5 GB sample
|
| 114 |
+
|
| 115 |
+
```bash
|
| 116 |
+
huggingface-cli download Video-Reason/VBVR-MultiStep \
|
| 117 |
+
--repo-type dataset \
|
| 118 |
+
--include "sample/**" \
|
| 119 |
+
--local-dir ./vbvr-multistep-sample
|
| 120 |
+
```
|
| 121 |
+
|
| 122 |
+
## Splits and seeds
|
| 123 |
+
|
| 124 |
+
The training corpus is partitioned into disjoint seed bands:
|
| 125 |
+
|
| 126 |
+
| Band | Seed range | Samples per task | Total samples |
|
| 127 |
+
|---|---|---|---|
|
| 128 |
+
| First-half | 1–5,000 | 5,000 | ~170k (across 34 trained tasks) |
|
| 129 |
+
| Second-half | 5,001–10,000 | 5,000 | ~170k (across 34 trained tasks) |
|
| 130 |
+
|
| 131 |
+
Both bands are disjoint from the **180-instance evaluation seeds** in `VBVR-MultiStep-Bench`. The submitted paper trains on 34 of 36 tasks; the released corpus contains all 36 task families.
|
| 132 |
+
|
| 133 |
+
## Reasoning families
|
| 134 |
+
|
| 135 |
+
See the [bench dataset card](https://huggingface.co/datasets/Video-Reason/VBVR-MultiStep-Bench) for the family taxonomy. Each family contributes 6 tasks, for 36 total.
|
| 136 |
+
|
| 137 |
+
## Intended use and out-of-scope
|
| 138 |
+
|
| 139 |
+
- **Primary use**: training I2V systems on long-horizon multi-step reasoning under explicit per-step rules.
|
| 140 |
+
- **Out-of-scope**: this corpus is fully synthetic and stylized; transfer to unconstrained open-world video is not validated by this release.
|
| 141 |
+
- **Not validated for**: production VLM pretraining at scale, real-world video generation, or any safety-critical use.
|
| 142 |
+
|
| 143 |
+
## License
|
| 144 |
+
|
| 145 |
+
Released under **CC-BY-4.0**. Generators consume only released task definitions; no third-party copyrighted content is embedded.
|
| 146 |
+
|
| 147 |
+
Derivatives of `Wan2.2-I2V-A14B` (Apache-2.0) referenced in the companion paper comply with the upstream license. This dataset does not redistribute model weights.
|
| 148 |
+
|
| 149 |
+
## Responsible AI
|
| 150 |
+
|
| 151 |
+
The dataset is fully synthetic. There are no human subjects, no scraped media, and no personal or sensitive information. Known biases inherit from the deterministic generators — every task family covers a deliberately narrow conceptual slice, and visual style is controlled by a fixed renderer family (no demographic content). See [`croissant.json`](./croissant.json) for the complete RAI metadata.
|
| 152 |
+
|
| 153 |
+
## Citation
|
| 154 |
+
|
| 155 |
+
```bibtex
|
| 156 |
+
@inproceedings{vbvr_multistep_2026,
|
| 157 |
+
title={Could Video Generation Models Solve Long-Horizon Multi-Step Reasoning Tasks?},
|
| 158 |
+
author={Anonymous},
|
| 159 |
+
booktitle={NeurIPS Datasets and Benchmarks},
|
| 160 |
+
year={2026},
|
| 161 |
+
note={Under review.}
|
| 162 |
+
}
|
| 163 |
+
|
| 164 |
+
@inproceedings{wang2026vbvr,
|
| 165 |
+
title={A Very Big Video Reasoning Suite},
|
| 166 |
+
author={Wang, Maijunxian and others},
|
| 167 |
+
booktitle={ICML},
|
| 168 |
+
year={2026},
|
| 169 |
+
url={https://icml.cc/virtual/2026/poster/65709}
|
| 170 |
+
}
|
| 171 |
+
```
|
croissant.json
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"@context": {
|
| 3 |
+
"@language": "en",
|
| 4 |
+
"@vocab": "https://schema.org/",
|
| 5 |
+
"citeAs": "cr:citeAs",
|
| 6 |
+
"column": "cr:column",
|
| 7 |
+
"conformsTo": "dct:conformsTo",
|
| 8 |
+
"cr": "http://mlcommons.org/croissant/",
|
| 9 |
+
"rai": "http://mlcommons.org/croissant/RAI/",
|
| 10 |
+
"data": {"@id": "cr:data", "@type": "@json"},
|
| 11 |
+
"dataType": {"@id": "cr:dataType", "@type": "@vocab"},
|
| 12 |
+
"dct": "http://purl.org/dc/terms/",
|
| 13 |
+
"examples": {"@id": "cr:examples", "@type": "@json"},
|
| 14 |
+
"extract": "cr:extract",
|
| 15 |
+
"field": "cr:field",
|
| 16 |
+
"fileProperty": "cr:fileProperty",
|
| 17 |
+
"fileObject": "cr:fileObject",
|
| 18 |
+
"fileSet": "cr:fileSet",
|
| 19 |
+
"format": "cr:format",
|
| 20 |
+
"includes": "cr:includes",
|
| 21 |
+
"isLiveDataset": "cr:isLiveDataset",
|
| 22 |
+
"jsonPath": "cr:jsonPath",
|
| 23 |
+
"key": "cr:key",
|
| 24 |
+
"md5": "cr:md5",
|
| 25 |
+
"parentField": "cr:parentField",
|
| 26 |
+
"path": "cr:path",
|
| 27 |
+
"recordSet": "cr:recordSet",
|
| 28 |
+
"references": "cr:references",
|
| 29 |
+
"regex": "cr:regex",
|
| 30 |
+
"repeated": "cr:repeated",
|
| 31 |
+
"replace": "cr:replace",
|
| 32 |
+
"sc": "https://schema.org/",
|
| 33 |
+
"separator": "cr:separator",
|
| 34 |
+
"source": "cr:source",
|
| 35 |
+
"subField": "cr:subField",
|
| 36 |
+
"transform": "cr:transform"
|
| 37 |
+
},
|
| 38 |
+
"@type": "sc:Dataset",
|
| 39 |
+
"name": "VBVR-MultiStep",
|
| 40 |
+
"conformsTo": "http://mlcommons.org/croissant/1.0",
|
| 41 |
+
"description": "The ~360,000-sample programmatic training corpus for long-horizon multi-step image-to-video reasoning. 36 parameterized tasks across six reasoning families (Navigation, Planning, CSP, Execution, Geometry, Physics). Distributed as 7,200 tar.gz shards (≈50 samples per shard) plus Parquet metadata; each instance follows a five-artifact contract identical to the VBVR-MultiStep-Bench evaluation split.",
|
| 42 |
+
"alternateName": ["VBVR-MultiStep Training Corpus"],
|
| 43 |
+
"creator": {
|
| 44 |
+
"@type": "sc:Organization",
|
| 45 |
+
"name": "Video-Reason",
|
| 46 |
+
"url": "https://video-reason.com"
|
| 47 |
+
},
|
| 48 |
+
"datePublished": "2026-05-06",
|
| 49 |
+
"keywords": ["video reasoning", "multi-step reasoning", "long-horizon", "image-to-video", "training", "synthetic", "tar.gz", "parquet"],
|
| 50 |
+
"license": "https://creativecommons.org/licenses/by/4.0/",
|
| 51 |
+
"url": "https://huggingface.co/datasets/Video-Reason/VBVR-MultiStep",
|
| 52 |
+
"version": "1.0.0",
|
| 53 |
+
"isLiveDataset": false,
|
| 54 |
+
"rai:dataCollection": "Fully synthetic. Each of the 36 tasks ships a deterministic generator that emits the five-artifact contract (first_frame.png, prompt.txt, final_frame.png, ground_truth.mp4, question_metadata.json) from a (task, seed) pair. No scraping, no human subjects, no third-party media, no manual annotation. The training corpus is partitioned into disjoint seed bands (1–5,000 and 5,001–10,000 per task) that are themselves disjoint from the evaluation seeds released in VBVR-MultiStep-Bench.",
|
| 55 |
+
"rai:dataCollectionType": ["Synthetic"],
|
| 56 |
+
"rai:dataPreprocessingProtocol": "Per-task generator output is grouped into 50-sample batches and packed into tar.gz shards under questions/. Per-task Parquet metadata files (data/metadata_shards/) and a global metadata.parquet index every instance with task id, family, seed, and per-task fields. No sample is filtered, dropped, or transformed after generation.",
|
| 57 |
+
"rai:dataAnnotationProtocol": "No human annotation. ground_truth.mp4 is rendered by each task's deterministic ground-truth solver.",
|
| 58 |
+
"rai:dataAnnotationPlatform": "N/A (no annotation).",
|
| 59 |
+
"rai:dataReleaseMaintenancePlan": "Versioned releases on the Hugging Face Hub. The Croissant file in this repository is the canonical long-term record. A 5 GB representative subset is provided under sample/ for quick inspection and reviewer convenience.",
|
| 60 |
+
"rai:dataLimitations": [
|
| 61 |
+
"Synthetic and stylized: transfer to unconstrained open-world video is not validated.",
|
| 62 |
+
"Visual rendering is intentionally simplified to keep the symbolic state recoverable from frames; this is not a photorealism corpus.",
|
| 63 |
+
"Per-task generator parameter ranges are bounded (e.g., maze sizes, planning horizons, physics regimes); the corpus does not span the long tail of any single family.",
|
| 64 |
+
"Reference rollouts encode one valid trajectory per instance; alternative valid trajectories are not enumerated.",
|
| 65 |
+
"Although 36 tasks ship, only 34 are used in the training experiments described in the companion paper; this release contains all 36 task families."
|
| 66 |
+
],
|
| 67 |
+
"rai:dataBiases": [
|
| 68 |
+
"Family balance is uniform (6 tasks per family) by design and does not reflect natural prevalence of these reasoning patterns.",
|
| 69 |
+
"Generator parameters bias the difficulty distribution toward bounded and seed-controlled regimes that are amenable to symbolic ground truth; rare or open-ended cases are out of scope.",
|
| 70 |
+
"Visual style is monocular, planar, and rendered by a fixed family of renderers; appearance distribution does not approximate any real-world video corpus and does not contain demographic content.",
|
| 71 |
+
"No human demographic information is generated; bias along human demographic axes does not apply."
|
| 72 |
+
],
|
| 73 |
+
"rai:personalSensitiveInformation": "None. The dataset contains no personal information, no biometric data, no demographic information, and no human subjects. All visual content is procedurally generated geometric, symbolic, or physical scenes.",
|
| 74 |
+
"rai:dataUseCases": "Training image-to-video systems on long-horizon multi-step reasoning under explicit per-step rules. Validated use case in the companion paper: fine-tuning Wan2.2-I2V-A14B (Apache-2.0) with Dual-DiT two-phase LoRA. Out-of-scope: production VLM pretraining at scale, real-world video generation, or any safety-critical use.",
|
| 75 |
+
"rai:dataSocialImpact": "Intended for academic research on reasoning evaluation in video generation. Risks are minimal: the dataset is synthetic, free of personal content, and rendered in a stylized regime not representative of any real population. The most plausible concern is research-direction effects (e.g., over-investing in stylized synthetic benchmarks), which we mitigate by positioning this corpus as a complement to (not a replacement for) appearance-centric and real-world video corpora.",
|
| 76 |
+
"rai:dataReleaseUpdate": "If post-release errors are discovered, fixes will be published as additive shards or replacement Parquet entries in a new dataset version, with the prior version retained at its commit hash for backwards reproducibility."
|
| 77 |
+
}
|
sample/data/metadata_shards/Multi-01_maze_shortest_path_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cce67b77319c4382f0d40c877035a858ac2ee71ca27b17f604f0a5f626f4a979
|
| 3 |
+
size 6505905
|
sample/data/metadata_shards/Multi-02_maze_circular_route_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cf135ffee9bf6dcc959a2dfefd4f4f61bc2b46e0edb0d3bc6de457e08659ee44
|
| 3 |
+
size 11626293
|
sample/data/metadata_shards/Multi-03_maze_marker_drawing_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:66a0a8ce19ab247788e83de38f16b71826fb5af9fca55683da530241b2c1081f
|
| 3 |
+
size 10318126
|
sample/data/metadata_shards/Multi-04_dual_dot_pathfinding_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:02f05595227108752b97b8be42128c7d2c20ed492bb3566cb2cdfa366cf509b9
|
| 3 |
+
size 12407365
|
sample/data/metadata_shards/Multi-05_maze_route_tracing_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:44bf0d142cfbfcac76684efb1149e493e7a14b7ce1c93318fbc8de907eb88a62
|
| 3 |
+
size 33476217
|
sample/data/metadata_shards/Multi-06_bfs_tree_traversal_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:63ec11a7675612677eb123a00e9d76a77cfc4ce253efc3fb71499ce59d243946
|
| 3 |
+
size 6320650
|
sample/data/metadata_shards/Multi-07_sokoban_planning_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:165723234c280bf7022191f20ea835184a093529e7ddf5fec92c0795a2a2fa80
|
| 3 |
+
size 3983807
|
sample/data/metadata_shards/Multi-08_sliding_puzzle_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:19e54f227f97e0e21e5c4ca25df1f0a7c69ebdaa5dbd39309c324fec5fc51b7f
|
| 3 |
+
size 5085041
|
sample/data/metadata_shards/Multi-09_tower_of_hanoi_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1dacd6cafc227c8770f0b03875eed6a32a438a8903f374d3d1932d6eeba2525e
|
| 3 |
+
size 3081595
|
sample/data/metadata_shards/Multi-10_snake_dynamic_routing_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8146c232c965e61ed9d5d5bb329337520ef8f3cf2d93e11eaa6fc46837b7e0bb
|
| 3 |
+
size 2226796
|
sample/data/metadata_shards/Multi-11_tsp_reward_collection_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f6b1e1f0eaf6a892ac79623f1384830cf4ffdc0ba39a9fe190ee6c58ff0e260b
|
| 3 |
+
size 2377297
|
sample/data/metadata_shards/Multi-12_ordinal_number_sequence_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:516680b174e8616f3d2ca216683a8e36578606d6a2e8b0b74b652839dc58639d
|
| 3 |
+
size 1727754
|
sample/data/metadata_shards/Multi-13_wordsearch_path_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2148665dc8246ca803c42956146d592e9827dc7eb99bec7a35c7ba6d3d221dad
|
| 3 |
+
size 5463957
|
sample/data/metadata_shards/Multi-14_sudoku_logic_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a89f35e3c208fe54fad25c64bb2be437ed19b71d5b9ba205cc13699ea7bcb2c4
|
| 3 |
+
size 7340425
|
sample/data/metadata_shards/Multi-15_numbrix_pathfilling_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:03b28308fd4ead9fb011543e03a22cf81eec5fdc15ccec0806224b1c55caf436
|
| 3 |
+
size 7057770
|
sample/data/metadata_shards/Multi-16_orthogonal_latin_square_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:af972ecf0cadfb5637f4b563a122d645d74958343da7e4756731f9a1b492db09
|
| 3 |
+
size 2757786
|
sample/data/metadata_shards/Multi-17_hashi_bridges_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e0dbb7a87b9e3aa8165492c739d678e65c6e49bae3951adabda0677d97b4c966
|
| 3 |
+
size 2748754
|
sample/data/metadata_shards/Multi-18_tents_and_trees_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6778690aed76979d45b5029e72a15d5226b4ba51d203dc4481aec8239e5452c3
|
| 3 |
+
size 2825908
|
sample/data/metadata_shards/Multi-19_turing_machine_execution_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e4922ea3b84a6d432189b1d6a9bf1dcfe7077a6895026f913bee9dad1c4c1a24
|
| 3 |
+
size 6298197
|
sample/data/metadata_shards/Multi-20_langtons_ant_simulation_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:971cb56fbf55abea083506a93b600f78e71d121a6d47e771720a852defb4beed
|
| 3 |
+
size 2915400
|
sample/data/metadata_shards/Multi-21_chained_math_calculation_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ffe88fd6226bd483b88a93b7c527e5473f56e73ed0420029b2840c9eb1d0c536
|
| 3 |
+
size 1602054
|
sample/data/metadata_shards/Multi-22_pointer_chasing_arrows_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8875633615c8a44a78de9bc97e41b77d5130e6c2436436aeeddb32de08a99017
|
| 3 |
+
size 3415614
|
sample/data/metadata_shards/Multi-23_chained_code_pipeline_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e0573832bb1f6c24c708e2c3e7bdc00e69eb6ed9317d32c9921da10ac2e1ab1
|
| 3 |
+
size 1598024
|
sample/data/metadata_shards/Multi-24_conways_game_of_life_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e464093892880beb88deda433faec7824ae8f804fa53d3782aea5936df8d3b1e
|
| 3 |
+
size 2764402
|
sample/data/metadata_shards/Multi-25_light_reflection_ray_tracing_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68883cb67e30cc23cddb7b89729c7d0322b6d79fb7ad0bfe20e49a4462ce9d22
|
| 3 |
+
size 5755209
|
sample/data/metadata_shards/Multi-26_line_intersection_construction_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:af9ae19bd3baafdd7f5b07259b06d74492dd631045804daefffd5a06ad5b3d0b
|
| 3 |
+
size 11444576
|
sample/data/metadata_shards/Multi-27_perpendicular_bisector_construction_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4e82c9e6540782f0a5468c50ea28a64af9645a70c433ec4f5d4f5afe40e4786e
|
| 3 |
+
size 4080314
|
sample/data/metadata_shards/Multi-28_triangle_orthocenter_construction_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:319d28a44bcdcd73a232311af2761df2b8aa0486422d85c7431ddb2abf32bd1d
|
| 3 |
+
size 3651137
|
sample/data/metadata_shards/Multi-29_triangle_incenter_construction_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b99aa3649f67d3c4c691af2a5674931ff3205f55c08a764864af7e92617eb406
|
| 3 |
+
size 3693519
|
sample/data/metadata_shards/Multi-30_triangle_circumcenter_construction_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bac4d677a4eb75f88164646528d21e600e3fe2548c469dfba39b61b7b9009004
|
| 3 |
+
size 3777073
|
sample/data/metadata_shards/Multi-31_fluid_communicating_vessels_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1675fe54d8b26deda540aff10c95dc8e3a18b9f85a2a3e973f1294c59354959c
|
| 3 |
+
size 2126407
|
sample/data/metadata_shards/Multi-32_multiple_bounces_target_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bc482b4f9de8cf291e1daeaae273bcb54acf410112aad02864484f8a8a3ff511
|
| 3 |
+
size 3158464
|
sample/data/metadata_shards/Multi-33_elastic_bouncing_trajectory_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f35f04700aca5399c049b5984ebb95fd8c75b8c790b22f390fdd09e93435ce58
|
| 3 |
+
size 1954765
|
sample/data/metadata_shards/Multi-34_elastic_collision_kinematics_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:205788bc01362f378252895ad14797d835739e37a34abb3ed9a1d7133d7857aa
|
| 3 |
+
size 2999190
|
sample/data/metadata_shards/Multi-35_block_sliding_friction_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:194d967226bb1db30bf6f4cce8983904d9c2fc1c924c75190e4f5336d4a262e8
|
| 3 |
+
size 2165548
|
sample/data/metadata_shards/Multi-36_target_after_reflection_data-generator.parquet
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6bbb0f6d8fe470a9caa614618700631348432266529395f18395908843dd338b
|
| 3 |
+
size 2572666
|
sample/questions/Multi-01_maze_shortest_path_data-generator_00000-00049.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b34047f76f04a979010667cf0b203a390e3eea1d48803c4e6b144977920c5d31
|
| 3 |
+
size 7163517
|
sample/questions/Multi-01_maze_shortest_path_data-generator_00050-00099.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8806aed588b6de6f21317fad9947ab159efdccedc781c2ee155962f64b9a85f7
|
| 3 |
+
size 6726748
|
sample/questions/Multi-01_maze_shortest_path_data-generator_00100-00149.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0beec2aa55fa82234817bff980b14d00d29b8c4c365484eea8913864502edc1d
|
| 3 |
+
size 6453175
|
sample/questions/Multi-01_maze_shortest_path_data-generator_00150-00199.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:588ac94b1b63781e152213887792191b608def1284afe8276906f194a5e3427e
|
| 3 |
+
size 6626562
|
sample/questions/Multi-01_maze_shortest_path_data-generator_00200-00249.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4cc9bc5d214219abae431ef5447b9cd0874721b8a9377a07f218af8dee92cbf5
|
| 3 |
+
size 7176053
|
sample/questions/Multi-01_maze_shortest_path_data-generator_00250-00299.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:13d2329f47b818029de8a431a19c1ca0366242c6f2ffb2b4c9efe62598306f91
|
| 3 |
+
size 7220650
|
sample/questions/Multi-02_maze_circular_route_data-generator_00000-00049.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f4feab0af1ae25ff32e7cd6d39343385d21f33b4c465a31d0d2cc2722e0042ce
|
| 3 |
+
size 37014921
|
sample/questions/Multi-02_maze_circular_route_data-generator_00050-00099.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:58b7bc36905666df5d980e90e36696b614e3898ff81fab52304d59fb915ae5ba
|
| 3 |
+
size 33435775
|
sample/questions/Multi-02_maze_circular_route_data-generator_00100-00149.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6eeeebfa575b4683102666cdbd148e6e9524096aece9863a53388fa891b1bbfc
|
| 3 |
+
size 34844595
|
sample/questions/Multi-02_maze_circular_route_data-generator_00150-00199.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:275a60d8982c125257f4cba3a51e43e8d22e688fbc3afd4d3935dff8ebd3a350
|
| 3 |
+
size 33624028
|
sample/questions/Multi-02_maze_circular_route_data-generator_00200-00249.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:93089221ef6ce043ce4d940845081c09393126b4e62d458716d813d9c9279ba8
|
| 3 |
+
size 32479158
|
sample/questions/Multi-02_maze_circular_route_data-generator_00250-00299.tar.gz
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9aa4fdb2a40da664954ec34bb11138f0113a2f704c9e111875a33ebb6f94e0f6
|
| 3 |
+
size 30402133
|