cs.AISep 23, 2026
SaveTraining Object Permanence in World Models
Organizations: University of Southern California · Carnegie Mellon University · University of Michigan · Johns Hopkins University · University of California, San Diego · University of California, Los Angeles · Columbia University · University of Toronto · University of Bristol · University of California, Berkeley · University of Waterloo · Friedrich-Alexander-Universität Erlangen · University of Oxford · New York University · Stanford University · Harvard University
Abstract
Object permanence and solidity are hallmarks of human cognitive priors. Recent studies show that video generation models, a paradigmatic class of current world models, have begun to show emerged reasoning abilities, making them ideal candidates for building human-like physical intelligence. Do video models have emerged object permanence in them? If not, could we train them with a core-cognition inspired dataset? We introduce WROP (World Reasoning with Object Permanence), a data infrastructure of 150 hand-designed cognitive science inspired tasks, divided into six cognitive categories. We build Blender generators that randomize speed, lighting, camera angle, and other nuisance parameters while preserving each task's cognitive structure, yielding 10,000+ samples per task. We release a 1.5M-sample training corpus and a 300-question exam. On this exam we evaluate 14 video models: 3 reference-to-video, 7 edit, and 4 continuation, among which PWM-WROP, our 16B world model. In a blind pairwise Elo study, PWM-WROP ranks first among continuation models and third overall, behind only a statistical tie between two reference-to-video models. We release the data, exam, model answers, scores, weights, and PWM, our native-PyTorch training stack on AWS Trainium2.
Figures & tables
Figure 1 : WROP: World Reasoning with Object Permanence. We introduce a benchmark and training resource for object permanence and solidity in video generation models, built on 150 hand-designed Blender generators across six task families. Each generator randomises lighting, camera angle, speed, and other nuisance parameters while preserving the task’s core cognitive-scientific structure, yielding 10,000 samples per task. We release this 1.5M-sample training corpus, a 300-question evaluation exam with human Elo ratings across 14 video generation models, and PWM-WROP, a fine-tuned continuation model trained on WROP data achieving state-of-the-art performance in its category.
Figure 2 : Taxonomy of the six WROP task families. The top row evaluates object permanence through dynamic occlusion, static-scene occlusion, and container-based concealment and movement; the bottom row evaluates object solidity through obstruction, support removal, and collision. Each panel shows representative frames from a sample drawn from one generator in the corresponding task family: an early input frame, the final input frame at the split boundary, and a target frame depicting the expected physical outcome.
Figure 3 : Generator-level composition of WROP’s 150 generators. The inner ring shows 90 OP and 60 OS generators; the outer ring shows the six task families.
Figure 4 : Generator-design example from Marked_Boxes_Swap using the current 60-frame input and 60-frame target clips. The sequence shows the objects in marked boxes, the closed boxes after crossing at the split, completion of the position exchange, and the identity-preserving reveal. The task requires the model to maintain the association between each hidden object and its marked container across the input–target boundary rather than infer identity from final screen position.
| Model | Access | Interface class | Native output (res / fps / frames) |
| Ours | |||
| PWM-WROP | Trainium2 48XL | True continuation | / 24 / 60 predicted † |
| Open-weight | |||
| MAGI-1 24B | Local, multi-GPU | True continuation | / 24 / 77 |
| LTX-2.3 Dev | Local, multi-GPU | Edit / transfer (IC-LoRA) | / 24 / 121 |
| Wan-VACE 14B | Local, multi-GPU | Edit / transfer (repaint) | / 16 / 57 |
Table 1: The fourteen evaluated models, grouped by provenance, with access route, interface class and native output geometry. † Our model emits a 117-frame clip that replays its 57 conditioning frames before the 60 predicted frames; the replay is trimmed before evaluation. ‡ Stitched or padded outputs whose source prefix is trimmed before evaluation (Section 4.3 ).
| Rank | Model | Class | Elo | 95% CI | Score rate | Games |
| 1 | Wan 3.0 Prime | Reference-to-video | 1723.6 | [1629.2, 1864.9] | 77.9% | 52 |
| 2 | MiniMax H3 | Reference-to-video | 1723.6 | [1644.5, 1837.6] | 77.9% | 52 |
| 3 | PWM-WROP (ours) | True continuation | 1679.5 | [1603.5, 1781.5] | 73.1% | 52 |
| 4 | Seedance 2.5 | Reference-to-video | 1649.6 | [1554.2, 1751.5] | 69.6% | 51 |
| 5 | Runway Aleph 2 | Edit / transfer | 1518.3 | [1425.1, 1621.6] | 52.9% | 51 |
| 6 | Wan-VACE 14B | Edit / transfer | 1506.7 | [1429.9, 1590.8] | 51.0% | 52 |
Table 2 : Human preference leaderboard of the 14 evaluated models on the 300-question WROP exam. Twenty raters made 361 blind pairwise judgments between the fourteen models (50–52 per model; ties count 0.5 for each side); strengths are Bradley–Terry maximum-likelihood estimates, with one virtual draw added per model and distributed evenly across its opponents, rescaled to an Elo scale with mean 1500. 95% confidence intervals come from 1,000 rater-clustered bootstrap resamples; overlapping intervals should not be read as significant rank differences. Score rate is the raw win rate (win , tie ). Wan 3.0 Prime and MiniMax H3 have identical records and tie for first.
Figure 5 : Human-preference Elo of the 14 models. Points are Bradley–Terry strengths on an Elo scale (mean 1500, dashed line), bars are 95% rater-clustered bootstrap intervals, colour is interface class, and PWM-WROP (filled) is the top true-continuation model.
Figure 6 : Within-family human-preference ranks across the six WROP task families. Each cell shows the Bradley–Terry rank within that family (1 = best, 14 = worst), fitted separately on the 20-rater pairwise judgments for each family (36–96 games per family; 361 total). Color encodes rank from green (top) to red (bottom). The outlined row marks PWM-WROP; overall Elo (right) is from the joint fit over all judgments. Row order follows overall Elo rank; column groups correspond to the task families in Figure 2 . Interface class is indicated by the label color, matching Figure 5 .
Figure 7 : Qualitative comparison on three object-permanence task families (Figure 2 , top row). Each row shows the same ground-truth input (two key frames, blue), the PWM-WROP target output (two frames, green), and the full baseline output (three frames spanning the generation, red), with / badges indicating physical correctness. G43 (OP-1): Gemini Omni Flash collapses all three balls into a tight cluster at the tunnel exit, violating their lane identities and count, while PWM-WROP keeps each ball in its correct lane with continuous motion. G19 (OP-2): Seedance 2.5 generates an oversized occluder over an empty region of the scene while the original objects remain visible, then makes all three objects suddenly appear when the occluder moves away, with no clear continuity from their previous positions; PWM-WROP correctly preserves the scene and reveals the same unchanged configuration. G66 (OP-3): Seedance 2.5 correctly animates the 180° turntable rotation but then lifts the cup at the original front position–a different physical cup–rather than tracking the cup containing the ball to its new location; PWM-WROP correctly tracks the ball with the rotating cup and lifts the correct cup.
| Model | LPIPS | MS-SSIM | SSIM | PSNR | MSE ( ) | CLIP | FID |
| Wan 3.0 Prime | 0.115 | 0.861 | 0.942 | 27.15 | 4.48 | 0.948 | 15.1 |
| MiniMax H3 | 0.105 | 0.877 | 0.938 | 27.51 | 9.09 | 0.962 | 14.8 |
| PWM-WROP (ours) | 0.081 | 0.921 | 0.917 | 26.45 | 2.97 | 0.956 | 20.4 |
| Seedance 2.5 | 0.282 | 0.616 | 0.770 | 18.37 | 28.55 | 0.928 | 24.6 |
| Runway Aleph 2 † | 0.181 | 0.789 | 0.918 | 24.98 | 9.71 | 0.918 | 19.7 |
| Wan-VACE 14B | 0.349 | 0.744 | 0.842 | 17.88 | 24.21 | 0.872 | 33.0 |
Table 3 : Automatic full-reference metrics against the reference continuation, averaged over the 300 exam questions and ordered by human Elo ( Table 2 ). For each model, the predicted span is extracted using its per-model trim rule, uniformly resampled to the 60 target frames, and compared at 320 192; CLIP (ViT-B/32, 224 224) and FID (299 299, all frames at stride 2, 9,030 frames per side) are computed at their respective input resolutions. The best result in each column is shown in bold and the second-best is underlined. These metrics measure similarity to the target rather than object-permanence capability: edit/transfer models can repaint the static scene and score well on structural metrics without depicting the required reappearance.
Appendix figures & tables4 assets
Supplementary material from the paper’s appendix.
Appendix
| ID | Name | Description |
|---|---|---|
| OP-1: Baillargeonian_Occlusion (26 tasks) | ||
| G02 | ramp_tunnel | An object rolls down a ramp, passes through an opaque tunnel, reappears … |
| G04 | high_low_cover | Objects pass behind high/low covers on crossing or straight paths; identity is preserved. |
| G14 | u_tube_three_lanes | Three coloured balls travel parallel U-tube lanes, preserve identity … |
| G17 | moving_tray | A moving tray carries objects behind a fixed screen (one or two lanes). |
| G28 | open_ended_tunnel | An object travels through an open-ended tunnel, hidden in the middle. |
Table 4 : Task inventory of all 150 WROP generators across the six task families. Each entry names the generator and its core physical scenario.
| Configuration | s/step | tokens/s |
| First working 64-core run | 15.1 | 4,705 |
| + reduce-scatter copy-in rewritten (row concatenation) | 6.74 | 10,564 |
| + multi-tensor AdamW, one sync per step; FSDP2 prefetch depth 2 | 5.71 | 12,454 |
Table 5: Training step time on one trn2.48xlarge (36-layer Cosmos3-Nano, tp4 fsdp16, batch 16, latent frames; medians over 10 steps after 3 warm-up steps). Loss trajectories are identical step for step across all rows.
| Model | Class | Overall | Object Permanence | Object Solidity | ||||
| Elo | OP-1 | OP-2 | OP-3 | OS-1 | OS-2 | OS-3 | ||
| Wan 3.0 Prime | Ref.-to-video | 1723.6 | 1670 2 | 1559 6 | 1730 1 | 1558 7 | 1749 3 | 1915 2 |
| MiniMax H3 | Ref.-to-video | 1723.6 | 1614 4 | 1689 2 | 1707 2 | 1625 4 | 1868 1 | 1925 1 |
| PWM-WROP (ours) | True cont. | 1679.5 | 1616 3 | 1785 1 | 1692 3 | 1604 5 | 1825 2 | 1394 8 |
| Seedance 2.5 | Ref.-to-video | 1649.6 | 1710 1 | 1644 3 | 1656 4 | 1687 1 | 1489 7 | 1714 4 |
| Runway Aleph 2 | Edit/transfer | 1518.3 | 1448 10 | 1552 7 | 1596 5 | 1641 3 | 1480 8 | 1331 12 |
Table 6 : Per-family Bradley–Terry Elo for all 14 models across the six WROP task families. Superscripts give within-family rank (1 = best). Underline marks the top-ranked model in each family; bold marks PWM-WROP (ours). Overall Elo is from the joint fit over all 361 judgments; models are ordered by overall Elo rank. Bootstrap CIs are wide especially in OS families (36–44 total games each); ranks indicate tendency rather than significant differences.
Figure 9 : Per-family human-preference Elo leaderboards for all six WROP task families. Points are Bradley–Terry MLE strengths on an Elo scale (mean 1500, dashed line); bars are 95% rater-clustered bootstrap intervals. Colour encodes interface class as in Figure 5 : filled blue circle is PWM-WROP (ours); open circles are other true-continuation models (blue), reference-to-video models (orange), and edit/transfer models (green). Each panel is sorted independently by within-family rank. Intervals are especially wide in OS families (36–44 total games each); adjacent ranks are rarely distinguishable. Numerical values and within-family ranks are reported in Table 6 .