Follow this section to personalize your feed and get instant alerts. Follow Go to your personalized feed WHY FOLLOW? Smart Alerts: Get notified about major news as it happens. Follow this tag to personalize your feed and get instant alerts. Follow Go to your personalized feed WHY FOLLOW? Smart Alerts: Get notified about major news as it happens.

Reinforcement Learning from Human Feedback (RLHF) is essential for aligning large language models (LLMs) with human preferences and values. While recent research has primarily focused on algorithmic advancements--such as reducing computational overhead or strengthening reward models to mitigate reward hacking--the critical role of prompt-data construction and its scalability has received comparatively less attention. In this paper, we address this gap by systematically exploring data-driven bottlenecks that currently hinder RLHF performance scaling, focusing specifically on the challenges posed by reward hacking and decreasing response diversity. To mitigate reward hacking, we introduce a hybrid reward system combining reasoning task verifiers (RTV) and a generative reward model (GenRM). This approach enables accurate assessment of responses against clearly defined ground-truth solutions. Additionally, in order to ensure response diversity and enhance learning effectiveness, we propose a novel prompt-selection method named Pre-PPO, explicitly identifying training prompts that are inherently challenging and thus less prone to reward hacking.

We study language models as evolving model organisms and ask when autoregressive next-token learning selects for world-tracking representations. For any encoding of latent world states, the Bayes-optimal next-token cross-entropy decomposes into the irreducible conditional entropy plus a Jensen--Shannon excess term. That excess vanishes if and only if the encoding preserves the training ecology's equivalence classes. This yields a precise notion of ecological veridicality for language models and identifies the minimum-complexity zero-excess solution as the quotient partition by training equivalence. We then determine when this fixed-encoding analysis applies to transformer families: frozen dense and frozen Mixture-of-Experts transformers satisfy it, in-context learning does not enlarge the model's separation set, and per-task adaptation breaks the premise. The framework predicts two characteristic failure modes: simplicity pressure preferentially removes low-gain distinctions, and training-optimal models can still incur positive excess on deployment ecologies that refine the training ecology. A conditional dynamic extension shows how inter-model selection and post-training can recover such gap distinctions under explicit heredity, variation, and selection assumptions. Exact finite-ecology checks and controlled microgpt experiments validate the static decomposition, split-merge threshold, off-ecology failure pattern, and two-ecology rescue mechanism in a regime where the relevant quantities are directly observable. The goal is not to model frontier systems at scale, but to use small language models as laboratory organisms for theory about representational selection.

Michael Pollan: 'Consciousness is really under siege' A psychedelic experience set author Michael Pollan on a quest to understand consciousness in his new book A World Appears. Michael Pollan: "Psychedelics have a way of smudging the windshield of experience" Author Michael Pollan has tackled plants, food and psychedelics in bestselling books including The Omnivore's Dilemma and How to Change Your Mind . Now, he has taken on the thorny problem of consciousness. In his latest book, Pollan charts the work of scientists and philosophers, weaving in literary perspectives along the way. He spoke to New Scientist about the value of writing a book where you know less at the end than before you started.

In this section, we describe details about hyperparameter setting of RD. SAC-N-Unc and TD3-N-Unc, M is set to 1/10 of the total training steps. To ensure fairness, algorithms employing RD are implemented using CORL repository [54]. By modifying the original SAC/TD3 algorithm to employ a critic ensemble of number N and incorporate an uncertainty regularization term within the policy update process, we derive these backbone algorithms. Additionally, using RD with fewer Q ensembles can achieve similar or even better results than the backbone methods using more Q ensembles, indicating its potential in reducing computing resource consumption.

Offline Reinforcement Learning (RL) aims to address the challenge of distribution shift between the dataset and the learned policy, where the value of out-of-distribution (OOD) data may be erroneously estimated due to overgeneralization.

December 5, 2025 Abstract We argue that language models (LMs) have strong potential as investigative tools for probing the distinction between possible and impossible natural languages and thus uncovering the inductive biases that support human language learning. We outline a phased research program in which LM architectures are iteratively refined to better discriminate between possible and impossible languages, supporting linking hypotheses to human cognition. Which conceivable linguistic systems are possible for humans to learn and use as natural languages? A complete answer to this question would yield profound insights into the human capacity for language. However, our tools for addressing the question are very limited.