README.md

metadata

pretty_name: Reasoning-Structured Videos Evaluation Artifacts
tags:
  - world-model
  - video-generation
  - benchmark
  - compositional-consistency
  - reproducibility

Reasoning-Structured Videos: Evaluation Artifacts

This page hosts evaluation artifacts for Reasoning-Structured Videos: A Stratified Diagnostic Suite for Compositional Consistency in World Models.

The benchmark tests whether action-conditioned video world models respect three trajectory relations:

• Inverse: a path followed by its inverse should return to the initial state.
• Loop: a closed path should return to the initial state.
• Equivalence: two different paths reaching the same state should produce matching endpoints.

We report self-consistency (SC) as the primary cross-model diagnostic and use GT-anchor and distributional metrics as complementary evidence.

Files

File	Description
matrix_game.zip	1,371 released Matrix-Game 2.0 SC rollout videos
evaluate_matrix_game_sc.py	Standalone Matrix-Game SC reproduction script
Reasoning-Structured-Videos-Rebuttal-main.zip	Frozen R20/R50 manifests, minWM/HY-WorldPlay camera adapters, strict-endpoint and terminal-window evaluators, tests, and result summaries
r50/	Frozen minWM and HY-WorldPlay R50 SC/GT-anchor rollout videos, metadata, reference images, and manifest

Reference videos and trajectory metadata are hosted separately in VideoWorldmodel/ReasoningStructureTestset.

Released R50 rollouts

The r50/ directory contains the generated rollouts used for the frozen R50 evaluation of two camera-trajectory-conditioned methods:

Method	Self-consistency	GT-anchor	Total
minWM	200	200	400
HY-WorldPlay 1.5 AR Distill 480P I2V	200	200	400

R50 has 50 graphs for each of Inverse, Loop, and Equivalence. Each tier contains 200 rollouts because every Equivalence graph has paired A/B branches. Self-consistency rollouts start from the model's conditioning frame; GT-anchor rollouts condition on the first 180 GT frames and predict through logical raw359.

r50/
  manifests/subset_manifest_R20_R50_raw360_v1.jsonl
  minwm/{self_consistency,gt_anchor}/predictions/<relation>/<record_id>/
  hy_worldplay/{self_consistency,gt_anchor}/predictions/<relation>/<record_id>/

Each record includes rollout.mp4 and metadata.json; reference.png is included where produced by the inference adapter. Use the logical-to-physical frame mapping in metadata.json when selecting frames: saved video spans differ between SC and GT-anchor outputs, so physical frame indices must not be guessed.

Download only this release with:

hf download VideoWorldmodel/Evaluation \
  --repo-type dataset \
  --include "r50/**" \
  --local-dir .

Main results

Common camera-trajectory SC (frozen R50)

minWM and HY-WorldPlay receive the same frozen graph IDs and camera trajectories through their official pose-control pathways. Each model completed 200/200 rollouts with zero evaluation failures, using the same 512x288 evaluation preprocessing.

Model	Inverse LPIPS / PSNR	Loop LPIPS / PSNR	Equivalence LPIPS / PSNR
HY-WorldPlay 1.5 AR Distill	0.4597 / 15.07	0.4941 / 15.50	0.3723 / 16.94
minWM	0.7118 / 10.46	0.7442 / 10.85	0.4437 / 14.05

R50 contains 50 graphs per relation: 150 graph units and 200 videos because every Equivalence graph has A/B branches. The code archive includes graph-bootstrap intervals, the corresponding GT-anchor summaries, and frozen R50 Endpoint FID/KID reports. Endpoint KID uses an unbiased finite-sample estimator, so valid negative estimates are retained in the detailed artifact rather than interpreted as distances below zero.

External-model Inverse SC versus revisit horizon

On the same frozen R20 Inverse graphs, pose-certified matched-state revisits show increasing inconsistency from 10 to 40 action steps. minWM changes by +0.2697 LPIPS [0.2196, 0.3256] and -3.65 dB PSNR [-4.69, -2.80]; HY-WorldPlay changes by +0.1039 [0.0379, 0.1754] and -1.94 dB [-3.37, -0.83]. The paired GT same-pose floor shows no detectable 10-to-40 degradation. The archive contains the evaluator, tests, per-graph pose audit, strict fixed-pair sensitivity analysis, and frozen JSON/Markdown results.

Matrix-Game Inverse SC versus revisit horizon

All 448 Inverse graphs are evaluated at matched-state revisits with increasing action separation. “Physical frame” denotes the zero-indexed position in the saved Matrix MP4. Matrix exports 357 frames for the complete 40-action schedule, so action boundary k is mapped to physical frame round(356k/40), anchoring both endpoints and preserving outward/return symmetry.

Revisit horizon	LPIPS (95% CI)	PSNR dB (95% CI)
10 actions	0.4851 [0.4652, 0.5053]	14.03 [13.53, 14.53]
20 actions	0.6106 [0.5976, 0.6234]	11.46 [11.14, 11.77]
30 actions	0.6685 [0.6584, 0.6784]	10.71 [10.45, 10.96]
Full rollout (40 actions)	0.7062 [0.6973, 0.7151]	10.45 [10.21, 10.69]

The paired full-minus-10 change is +0.2211 LPIPS [0.1988, 0.2434] and -3.58 dB PSNR [-4.02, -3.15]. Adjacent boundary-rounding schemes preserve the monotonic trend and change intermediate means by at most 0.005 LPIPS / 0.10 dB.

Endpoint distribution fidelity

Endpoint FID/KID compare generated and matched GT logical raw359 endpoint sets using clean-fid Inception-v3 pool3 features.

Matrix-Game, complete available output set:

Relation	Graphs / images	Endpoint FID ↓	KID ×1000 ↓
Inverse	448 / 448	134.94	23.018
Loop	445 / 445	141.33	24.362
Equivalence	239 / 478	145.55	22.943

minWM, frozen R50 SC outputs:

Relation	Graphs / images	Endpoint FID ↓	KID ×1000 ↓
Inverse	50 / 50	242.46	47.860
Loop	50 / 50	249.40	54.173
Equivalence	50 / 100	240.23	50.359

Endpoint FID is an image-set metric and is not numerically comparable to the clip-level FVD reported elsewhere. The Matrix analysis also includes 20/50/100/200/full graph sensitivity and a GT-vs-GT finite-sample floor. These endpoint metrics complement, rather than replace, paired SC LPIPS/PSNR.

Terminal-window distribution robustness

Because an external-model R50 relation contains only 50 strict endpoint images, we additionally pool the final matched logical frames from the common SC rollout track. K=9 covers the final 0.56 seconds and K=18 the final 1.12 seconds of each 22.5-second rollout. Each entry is FID / KID ×1000.

Window	HY-WorldPlay: Inverse / Loop / Equivalence	minWM: Inverse / Loop / Equivalence
K=9	103.94 / 6.553 · 132.62 / 8.388 · 128.98 / 8.825	223.16 / 62.504 · 242.04 / 69.070 · 217.60 / 56.783
K=18	95.61 / 6.763 · 129.45 / 9.053 · 117.96 / 8.224	214.68 / 62.997 · 235.68 / 67.737 · 208.07 / 56.998

These are point estimates over 50 graph clusters per relation. Consecutive frames are correlated, so the graph remains the sampling unit. This is a late-horizon marginal appearance-distribution check, not an exact-state closure or temporal-order metric. The code archive contains the evaluator, unit test, frozen JSON/Markdown results, and a generic reproduction command.

Quick reproduction: Matrix-Game SC

This path requires only the files on the current page; no GT dataset is needed.

python -m pip install -U huggingface_hub

hf download VideoWorldmodel/Evaluation \
  matrix_game.zip evaluate_matrix_game_sc.py \
  --repo-type dataset \
  --local-dir .

unzip matrix_game.zip -d data/MatrixGame2_SC_videos

python -m pip install \
  "numpy>=1.26,<3" \
  "opencv-python-headless>=4.8,<5" \
  "torch>=2.2,<3" \
  "torchvision>=0.17,<1" \
  "lpips==0.1.4"

python evaluate_matrix_game_sc.py \
  --data data/MatrixGame2_SC_videos \
  --output results/matrix_game_sc \
  --lpips \
  --device auto \
  --check-paper

If the archive creates one additional top-level directory, point --data to the directory that directly contains the five relation folders.

Expected values:

Relation	Graph N	Recomputed LPIPS / PSNR	Paper LPIPS / PSNR
Inverse-SC	448	0.7061 / 10.4476	0.71 / 10.45
Loop-SC	445	0.7173 / 10.6227	0.72 / 10.62
Equivalence-SC	239	0.5918 / 12.5732	0.59 / 12.57

The evaluator verifies file counts and Equivalence pair integrity, and writes per-graph scores, graph-bootstrap confidence intervals, an audit, and a paper-value check.

Reproduce the additional analyses

Download and unpack the code artifact:

hf download VideoWorldmodel/Evaluation \
  Reasoning-Structured-Videos-Rebuttal-main.zip \
  --repo-type dataset \
  --local-dir .

unzip Reasoning-Structured-Videos-Rebuttal-main.zip
cd Reasoning-Structured-Videos-Rebuttal-main

python -m pip install -r requirements-eval.txt
python -m unittest discover -s tests -v

The repository README documents six reproducibility paths:

1. recompute SC/GT LPIPS and PSNR from minWM or HY-WorldPlay rollout directories;
2. recompute pose-certified Inverse SC versus horizon for minWM and HY-WorldPlay from saved rollouts;
3. recompute Matrix-Game Inverse SC versus revisit horizon directly from matrix_game.zip;
4. recompute relation-wise endpoint FID/KID;
5. recompute R50 terminal-window FID/KID from saved external-model rollouts;
6. run the full Matrix endpoint audit using matrix_game.zip and the GT dataset.

The generated minWM/HY-WorldPlay R50 rollouts are released under r50/; model checkpoints are not included. Regenerating the videos requires the official minWM or HY-WorldPlay repository, its checkpoint, its official environment, and suitable GPUs.

Frozen identifiers

• Public subset seed: 2357
• R20 manifest SHA-256: f8307b78ffb4633b1d6ca6340498a9db615dca4cc45a045d16139abf4375abdf
• Nested R20/R50 manifest SHA-256: c1b3e051ec861f6babd00f67cf7c31d0ceba590d93a6fac1ac6a768b40448db4
• matrix_game.zip SHA-256: caec3d8deb20cffbc645f0aeeaa2af425536715eef935b17347cdfab728a88d4
• Reasoning-Structured-Videos-Rebuttal-main.zip SHA-256: 9ee07bfff222800b2eddf6d3e4a44dd2e93a250ea8e7945dbbdf1a7af8b80ca0

Scope

The released experiments evaluate deterministic camera trajectories in static scenes. They support relation-specific diagnostic conclusions under the stated protocol; they are not intended as a universal ranking of world-model architectures.