The paper introduces the concept of rational value risk, a gap between a large language model’s deployed reasoning strategy and an ideal rational strategy that maximises utility. It shows that even well‑aligned models can fail to achieve the aligned value during reasoning, and decomposes the estimation error into bounded prompts, bounded responses, and imperfect verifiers. Experiments across several LLMs and benchmarks confirm that rational value risk is common, can be mitigated but not eliminated by value alignment, and that techniques like self‑consistency and longer chains of thought can improve rationality with diminishing returns.
arXiv:2608. 15445v1 Announce Type: new Abstract: When a reward is correct on every training example yet consistent with more than one goal, a model can acquire an unintended one, a failure known as goal misgeneralization.
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.
arXiv:2608. 03838v2 Announce Type: replace Abstract: Reasoning-based large language model (LLM) safeguards improve safety classification accuracy by generating explicit rationales before making decisions.
arXiv:2605. 24396v2 Announce Type: replace Abstract: Long chains of thought (CoT) from current language models frequently contain logical gaps and unjustified leaps, limiting the gains from additional test-time compute.
arXiv:2606. 30627v1 Announce Type: cross Abstract: Conservative offline training is widely advocated as a safe foundation for subsequent online adaptation: if a policy stays close to well-supported behaviour, the argument goes, it is less likely to exploit imperfections in a learned reward model.
arXiv:2607. 14552v1 Announce Type: cross Abstract: A standard recipe for distilling the reasoning ability of large language models (LLMs) is to sample chains of thought from the model, keep those that reach the correct final answer, and fine-tune on the survivors.
arXiv:2511.11500v3 Announce Type: replace Abstract: Modern language models fail a fundamental requirement of trustworthy intelligence: knowing when not to answer. Despite achieving impressive accurac...
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...
The paper introduces BELT (BELief-based Termination), a method that decouples outcome prediction from stopping control in large language models (LLMs). BELT learns beliefs about current and future answers from a frozen LLM’s hidden states and applies a linear stopping policy, enabling a single probe to adapt to various token‑price trade‑offs without retraining. Across five out‑of‑distribution math benchmarks, BELT cuts token usage by up to 51% while maintaining accuracy, and its lightweight probe outperforms stronger baselines in both efficiency and parameter count, also transferring effectively to coding and agentic tasks.
The paper introduces a self‑distillation method that leverages answer‑only data to improve large language models (LLMs) after training. By training an inverse reasoning LLM to generate latent chain‑of‑thought traces conditioned on the answer, the authors can perform offline distillation that significantly boosts performance on hard prompts for tasks like Countdown and GSM8K. Additionally, the inverse model serves as an accurate verifier, enhancing best‑of‑n selection accuracy.
The paper introduces Concise-9B, a version of the Qwen3.5-9B reasoning model that reduces unnecessary thinking tokens through concise self‑training. By applying two rounds of on‑policy rejection‑sampling fine‑tuning and a small share of answer‑synthesis targets, Concise-9B cuts token usage to roughly two‑thirds of the original model while maintaining comparable accuracy on a wide range of hard mathematics and reasoning benchmarks. The study also shows that most of the token savings occur after the answer has already formed, and that the approach’s effectiveness depends on the verbosity of the base model.
The paper investigates why large language models (LLMs) struggle in strategic decision-making under incomplete information. It identifies two key gaps: an observation‑belief gap where LLMs’ internal representations of game states are accurate but brittle, and a belief‑action gap where converting these internal beliefs into actions is weak, leading to suboptimal payoffs. Experiments with Llama 3.1, Qwen3, and gpt‑oss confirm that acting optimally on decoded beliefs would improve outcomes in most games, highlighting a bottleneck in belief‑to‑action conversion.