The paper investigates whether large language models (LLMs) correctly gauge their confidence when acting in a hidden‑information chess variant. In experiments where the location of a hidden royal piece is repeatedly relocated, the models’ stated probabilities about the piece’s position were almost never accurate at high confidence levels, with a calibration deficit concentrated in those high‑confidence events. Across multiple model configurations and providers, the same pattern emerged, and conventional evaluation metrics such as legality, cost, latency, and completion rate were found to be uncorrelated with belief quality, yet a model could still win the game despite poor confidence estimates.
By Bhushan Kashinath Joshi
arXiv:2607. 00164v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards can in principle train calibrated probabilistic forecasters, since a proper scoring rule such as the Brier score is computed from outcomes alone and is minimized in expectation by the true probability.
By Sadanand Singh, Allam Reddy, Manan Chopra
Agentic systems increasingly gate actions on a model's own stated confidence, which assumes confidence tracks correctness at the moment of acting. We test this in a hidden-information chess variant wh...
AD-WM is a new action‑discriminative joint‑embedding world model designed for counterfactual model predictive control. It augments residual latent dynamics with action‑recovery regularization based on inverse dynamics and conditional mutual information, while discarding auxiliary heads at test time so that MPC remains unchanged. Experiments on OGBench‑Cube and other simulation environments show substantial gains in hard‑start success and mean success, and zero‑shot transfer to a Franka robot improves pick‑and‑place success from 42.2% to 71.1%.
By Jiabin Qiu, Zixuan Chen, Hongye Cao, Jieqi Shi, Jing Huo, Yang Gao
arXiv:2607. 13618v1 Announce Type: new Abstract: LLM agents are increasingly evaluated on multi-week decision tasks in which the state that drives cost is never directly observed.
By Sagar Deb, Ashwanth Krishnan
LLM agents are increasingly evaluated on multi-week decision tasks in which the state that drives cost is never directly observed. On such tasks the final cost cannot say why an agent failed: it may have misread the world, or read it correctly and still failed to act (the knowing-doing gap).
arXiv:2609.24677v1 Announce Type: new
Abstract: Large language models (LLMs) are increasingly used to make predictions from numerical time-series histories and textual events. Yet accuracy alone cann...
By Jie Gong, Maowei Jiang, Zhiwei Liu, Yankai Chen, Guojun Xiong, Xue Liu, Min Peng, Qianqian Xie, Sophia Ananiadou
arXiv:2607. 04419v2 Announce Type: replace Abstract: Final-answer scores hide which agent transitions helped or harmed a trace.
By Andrew Zhang, Chengzhan Li
The paper introduces a method for verifying whether two gameplay replays in Counter‑Strike 2 belong to the same player by extracting a behavioral fingerprint that captures crosshair control, movement‑stop‑fire coordination, economy, combat engagement, and temporal rhythm. Using 1,330 demos and 13,300 observations, the authors train a pairwise model that achieves an ROC AUC of 0.931 and 0.722 recall at 95% precision, with low‑level mechanical habits providing the strongest identity signals. Aggregating multiple demos further improves performance, raising AUC to 0.986 when ten historical demos are considered.
By Xuchen Zhang
arXiv:2603. 02491v3 Announce Type: replace-cross Abstract: As artificial agents become increasingly capable, what internal structure is necessary for an agent to act competently under uncertainty?
By Aran Nayebi
arXiv:2606. 12200v1 Announce Type: cross Abstract: We study policy representation learning from unlabeled multi-policy behavioral data.
By Andrew Kang, Priya Narasimhan
The paper proposes a method for verifying whether two gameplay replays in Counter‑Strike 2 belong to the same player by extracting a behavioral fingerprint that captures crosshair control, movement, economy, combat, and rhythm. Using a six‑fold evaluation on two datasets (Perfect and Professional), the pairwise model achieves ROC AUCs of 0.926 and 0.956, with aiming and low‑level mechanics providing the strongest identity signals. Aggregating evidence across multiple historical demos further improves account‑history AUC, reaching 0.982 on Perfect and 0.975 on Professional.
By Xuchen Zhang