← Back to all news
arXiv AI By Yuhua Jiang, Junjie Lu, Feifei Gao

Sampling headroom is not selection gain: a compute-value audit of test-time scaling for video world models

Read the original on arXiv AI →

The paper introduces the Compute-Value Audit (CVA), a sequential framework that evaluates whether extra sampling during test‑time scaling for video world models actually yields a net benefit after accounting for the compute cost of generation and verification. On 192 Physics‑IQ scenes, increasing the sample pool from 4 to 16 candidates improves oracle quality by +9.23 IQ, yet common metrics such as Flow, Cycle, and VideoReward fail to reliably recover this headroom, and adaptive‑depth policies recover only 42‑69% of the potential gain. Only a few specific interventions—anchor‑explorer in a sparse PRM800K setting, MMLU‑Pro exposing a predictive‑state gap, and a privileged paired‑future upper bound—successfully pass all CVA stages, indicating that sampling headroom is valuable only when it can be converted into a reliable decision that survives the full compute charge.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

Related stories

arXiv Computation and Language
Sep 4

Select, Compress, Reinvest: A Controlled Study of Visual-Token Allocation in Long-Video MLLMs

The study investigates how long‑video language models decide which frames to keep, compress, and reuse, testing each decision in isolation across six selection rules, three benchmarks, and two answering models. It finds that selecting frames based on queries yields the biggest performance boost, that halving spatial resolution costs little, and that reallocating saved tokens to more compressed frames can further improve accuracy. The work also highlights the importance of a unified evaluation harness to avoid misleading comparisons.

By Prakhar Khatri
llmsbenchmarks
More like this →
arXiv AI
Aug 18

SCOPE: Score-Isolated Agentic Optimization for Video World Models

arXiv:2608. 15043v1 Announce Type: new Abstract: Video world models are increasingly used as simulators for planning and embodied decision making, yet improving them at inference time introduces a subtle evaluation problem: prompts, samplers, verifiers, and selectors may evolve together, making it difficult to attribute gains or prevent held-out feedback from shaping the final policy.

By Yuhua Jiang, Jiaming Wang, Qingbin Liu, Feifei Gao
agentsroboticsbenchmarks
More like this →
arXiv AI
Sep 3

Modeling What Changes: Sparse, Residual World Models for Object-Centric Manipulation

The paper introduces a sparse, residual world model that focuses on predicting only the changes in a scene by using a per-object change gate and a residual delta head. On a MuJoCo tabletop pushing benchmark, this approach outperforms a dense multilayer perceptron, achieving 2.5 to 4.6 times better next‑state pose accuracy with 8.6 to 11.1 times fewer parameters, maintaining high change‑detection F1 scores, and showing strong transfer across object counts. In autoregressive rollout and sampling‑based planning, the sparse model accumulates less error and enables successful planning where dense models fail.

By Param Thakkar, Parsika Paresh Shah, Manisha Sushant Gote
roboticsbenchmarkssafety
More like this →