We study how to predict the downstream closed-loop performance of a learned latent world model from validation-time diagnostics alone. Choosing the right checkpoint from a world-model training run is difficult: validation loss and multi-step prediction RMSE keep improving long after closed-loop performance has collapsed.
arXiv:2606. 03238v1 Announce Type: cross Abstract: Reinforcement learning from human feedback (RLHF) makes large-scale post-training possible by replacing an underspecified human objective with learned and scalable proxies.
By Zelalem Abahana
arXiv:2608. 07809v1 Announce Type: new Abstract: A world model is only useful for physical AI if it changes what the agent does, and only safe if it declines to do so when it is wrong.
By Yiyao Zhang, Diksha Goel, Hussain Ahmad, Shixun Huang, Jun Shen
arXiv:2607. 04464v1 Announce Type: cross Abstract: World-model evaluation for model-based reinforcement learning typically asks whether the learned model predicts reward and value well, which can leave planning-relevant errors in the model's latent rollouts unmeasured.
By Donna Vakalis
arXiv:2602. 05459v2 Announce Type: replace Abstract: Offline goal-conditioned reinforcement learning (GCRL) is typically benchmarked by the best tuned success rate of each method.
By Jan Malte T\"opperwien, Aditya Mohan, Marius Lindauer
arXiv:2606. 04145v1 Announce Type: cross Abstract: Cloud LLM fine-tuning platforms increasingly serve RLHF workloads, where a learned reward model is optimized as a proxy for human quality.
By Guilin Zhang, Chuanyi Sun, Shahryar Sarkani, John M. Fossaceca
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
arXiv:2606. 30852v1 Announce Type: new Abstract: Reasoning models spend different amounts of useful computation across instances, but it remains unclear when a learned stopping rule improves over simple confidence or convergence thresholds.
By Zhe Dong (University of Maine at Presque Isle), Fang Qin (Stanford University), Manish Shah (Independent Researcher)
arXiv:2604. 13517v4 Announce Type: replace-cross Abstract: Temporal credit assignment in reinforcement learning is often approached by introducing value estimates at multiple discount factors.
By Jing Sun
arXiv:2607. 18966v1 Announce Type: new Abstract: Language models trained with reinforcement learning may learn to optimize the grader's judgment rather than the intended objective.
By Axel H{\o}jmark, J\'er\'emy Scheurer, Evgenia Nitishinskaya, Felix Hofst\"atter, Jason Wolfe, Theodore Ehrenborg, Bronson Schoen, Alexander Meinke
arXiv:2607. 04978v1 Announce Type: new Abstract: Joint-Embedding Predictive Architectures (JEPAs) underpin a growing family of latent world models for control from raw pixels, but every existing JEPA world model commits at training time to a single inference paradigm: either trajectory optimisation in a learned dynamics model, or direct behaviour cloning.
By Ruslan Rakhimov, George Bredis, Yuriy Maksyuta, Daniil Gavrilov
arXiv:2608. 01130v1 Announce Type: new Abstract: A broad range of models face the mismatch where they are updated through trajectory losses but are evaluated by downstream task reward.
By Yuyang Shen