The paper investigates how autoregressive language models can iterate a hidden state without emitting tokens, identifying three distinct risks associated with a latent process: token risk, teacher‑forced latent risk, and open‑loop risk. It shows that when the Bayes next‑token posterior depends on a full‑rank $k$‑dimensional statistic, a width‑$k$ latent with linear least‑squares transition and a refitted unembedding can match the Bayes token predictor, but this match is only up to an invertible gauge. The study also highlights sharp obstructions: a constant latent can drive teacher‑forced risk to zero while token cross‑entropy reverts to marginal entropy, and a perfect token predictor may still conceal a null‑space walk that causes open‑loop error to grow linearly with horizon, indicating that small latent or token losses alone do not guarantee accurate latent rollout without additional rank or injectivity conditions.
By Yuzhou Hong
arXiv:2610.26979v1 Announce Type: new
Abstract: Latent world models let an agent plan in latent space. The planner searches for an action sequence that minimizes a cost, and that cost is often the $\...
By Yoo Sung Jang, Michael Ryoo
arXiv:2607. 10362v1 Announce Type: new Abstract: Latent world models are trained to predict future states in a learned representation and are then deployed inside a planner that selects actions by simulating them forward.
By Hanzhe You, Yonggang Zhang, Maohao Ran, Zhiqin Yang, Zhenyuan Zhang, Wei Xue, Jun Song, Xinmei Tian, Yike Guo
arXiv:2609.25541v1 Announce Type: new
Abstract: Tabular foundation models learn to predict cell values in context, whereas world-model self-supervision asks for prediction in representation space (Le...
By Mingyu Jeon, Suwan Cho, Jae Young Suh
arXiv:2606. 03003v1 Announce Type: cross Abstract: A latent world model built from an equivariant encoder $E$ and an equivariant predictor $f$ inherits a provable symmetry of its training loss: when the world's dynamics genuinely carries a group $G$ acting on latents by an orthogonal representation $\rho(g)$, the one-step prediction relMSE is exactly invariant across the whole group, so fitting the dynamics on a restricted slice of orientations mathematically determines it on the entire orbit (j\v{u} y\=i f\v{a}n s\=an).
By Hongbo Wang (Stony Brook University)
The paper challenges the common practice of selecting latent world models based on one‑step prediction loss, arguing that this metric poorly predicts planning performance. Using a frozen‑decoder protocol, the authors decompose one‑step error into encoder‑inherent and predictor‑added components and evaluate goal‑conditioned planning across three continuous‑control environments. Their experiments show that improvements in one‑step fidelity do not correlate with planning success, with correlation coefficients near zero and inconsistent exchange rates across tasks.
By Joshua Cheung, Zhang He
arXiv:2607. 10203v2 Announce Type: replace-cross Abstract: Adaptive-compute world models -- early-exit or mixture-of-depths predictors that spend variable depth per step -- assume depth buys better predictions and can be routed adaptively.
By Achyuthan Sivasankar
arXiv:2610. 10246v1 Announce Type: new Abstract: Constant-step stochastic approximation generally has a nonzero stationary mean error that persists under time averaging.
By A. Ch. Madhusudanarao, Rahul Singh
arXiv:2606. 08934v1 Announce Type: new Abstract: Recurrent neural networks maintain a hidden state $h_t$, but its probabilistic meaning is often unclear.
By Yuan-chin Ivan Chang
arXiv:2607. 13612v1 Announce Type: cross Abstract: Joint-Embedding Predictive Architectures (JEPAs) are the dominant design for latent world models, yet they are usually justified by empirical performance rather than a normative principle.
By Fabio Arnez, Alexandra Gomez-Villa
The paper introduces Rollout-Decoded Reconstruction (RDR), a training technique for latent world models that aligns training with evaluation by free-running the model during training, decoding each latent, and penalizing reconstruction error against ground truth. RDR adds a single loss term without extra parameters or significant computational cost, and it reduces to the standard objective when the term weight is zero. In experiments on the chaotic Kuramoto–Sivashinsky equation, RDR improves valid prediction time from 3.87 to 6.97 time units, a 1.80× increase, and the advantage grows with latent width.
By Rishi Shah, Rishav Shrestha
arXiv:2607. 05238v1 Announce Type: new Abstract: JEPA world models predict the next latent state with a single deterministic predictor trained by latent regression.
By Zhi Song, Ximing Xing, Zhenchao Tang, hanbo Huang, Tianxu Lv, minghao Yang, Zhongzheng Niu, He Bing, Lusheng Wang, Jianhua Yao