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.
arXiv:2609.37119v1 Announce Type: cross Abstract: Recent approaches to reinforcement learning (RL) post-training for large language models increasingly remove the critic to reduce training instabilit...
The paper introduces an agentic forecasting environment built on 2,100+ resolved Polymarket questions, where a language model (Qwen3.5-35B-A3B) learns to gather evidence during rollout via web search, page reading, and financial time series, all filtered to avoid post‑cutoff leaks. Training with single‑epoch GRPO and a Brier‑score reward improves calibration by 30‑40% and reduces search attempts, while the trained policy outperforms four frontier models in evidence‑based forecasting, achieving lower soft‑Brier scores at roughly 5% of the inference cost. The authors release the environment, dataset, and per‑rollout records as a reusable harness for temporal forecasting agents.
arXiv:2609.38850v1 Announce Type: new Abstract: Decision models such as Jev answer questions with probabilities, which are only useful if they are calibrated. Open-source reproductions rely on superv...
arXiv:2607. 12397v1 Announce Type: new Abstract: LLM agents act in external environments where each action changes the state that later decisions condition on, and where a single wrong step can waste interaction budget or trigger irreversible side effects long before the final failure is observed.
The paper investigates how different proper scoring rules influence the performance and behavior of large language model (LLM) forecasters. Five scoring rules were compared as training objectives for binary forecasts of real-world events, revealing that while they all theoretically incentivize truthful probability reporting, they produce models with varying calibration, probability usage, and bias, information, and noise profiles. The Brier-trained model achieved the lowest Brier score and highest AUC-ROC, whereas the log-trained model achieved the best log score and lowest calibration error, indicating that scoring rule choice can shape both forecast accuracy and error structure.
arXiv:2607. 05904v1 Announce Type: new Abstract: Training a language model against its own reference-free judgments (the premise of self-rewarding, self-play, and LLM-as-a-judge pipelines) assumes a model's verdict on a shown answer tracks correctness.
arXiv:2607. 10139v1 Announce Type: cross Abstract: Selecting the correct answer from a pool of candidate reasoning chains is the engine of test-time scaling, yet the standard selectors each carry a cost: self-consistency inherits the errors of the single model it resamples, and trained reward models need labeled data and transfer poorly off-distribution.
The paper proposes measuring a language model’s understanding via no‑arbitrage, defining it as the inability of a bounded trader to profit from Dutch books against the model’s probabilities on logically related claims. It shows that full logical coherence is computationally infeasible, that standard next‑token training yields incoherent predictions across formats, and that uncertainty grows predictably along reasoning chains, creating arbitrage opportunities. The authors introduce Arbitr, a training framework that penalizes logical inconsistencies while maintaining accuracy, reducing exploitability by orders of magnitude and revealing a scaling illusion where large models appear coherent yet exhibit extreme unjustified confidence.
arXiv:2607. 17765v1 Announce Type: cross Abstract: We introduce WC2026-Agents, a benchmark and dataset for evaluating large language models (LLMs) as autonomous forecasting agents on real, future events.
arXiv:2608. 11560v1 Announce Type: new Abstract: Personalizing marketing messages with contextual multi-armed bandits (CMABs) drives real business value, yet the objective that ultimately matters - a downstream conversion - is observed only weeks later, too late to drive online learning.
arXiv:2602. 18037v2 Announce Type: replace-cross Abstract: Reinforcement Learning from Human Feedback (RLHF) or Verifiable Rewards (RLVR) are two key steps in the post-training of modern Language Models (LMs).
The paper shows that fine‑tuning a large language model (LLM) with a debate framework—where a generator and a critic compete and a weaker LLM judge adjudicates—reduces reward hacking compared to standard reinforcement learning from AI feedback (RLAIF). In experiments on mathematics tasks, the debate approach keeps the judge’s performance stable, achieving a 45% higher peak validation accuracy than the RLAIF baseline and mitigating the rapid exploitation of judge errors. Additional findings indicate that weakening the judge speeds hacking unless countered by extra debate rounds, that debate can override misalignment prompts, and that word‑limit constraints on critiques help balance the game and prevent judge hacking. whyItMatters":"The study demonstrates a practical method to curb reward hacking in RL‑based AI systems, addressing a key obstacle for safely scaling AI oversight."