arXiv:2606. 19831v1 Announce Type: cross Abstract: Aligned language models gate behaviors such as refusal and language routing through sparse feed forward neurons, yet no theory predicts when a single neuron intervention controls a behavior coherently rather than collapsing the output.
SCAMP introduces a training‑free, damped Gauss‑Newton method that adjusts only the sparse anchor points in a frozen differentiable decoder, keeping the rest of the state unchanged. By operating solely in the space of the anchors’ Jacobian rows, it solves a system whose size matches the number of anchor constraints rather than the full state, enabling efficient control across diverse text‑to‑motion generators. Applied to seven existing generators, SCAMP achieves anchor errors that match or surpass all released control methods and can close anchors on hosts that originally lacked them.
arXiv:2606. 30789v1 Announce Type: new Abstract: Group Relative Policy Optimization (GRPO) has become a standard tool for improving the reasoning ability of large language models, yet its training dynamics are still described empirically: reward trajectories are fit with low-parameter functional forms whose constants carry no mechanistic meaning, and hyperparameter choices remain a matter of trial and error.
arXiv:2609.26420v1 Announce Type: cross Abstract: Text-to-motion models generate plausible human motion but do not model a robot's dynamics; whole-body tracking controllers execute robot references r...
The paper introduces a world model that learns to predict the evolution of physical systems while respecting key physical principles. By hard‑coding a general structure—generating dynamics from the gradient of a learned energy via a fixed reversible operator and imposing constraints on energy, dissipation, and interventions—the model achieves second‑law compatible dissipation, accurate responses to parameter changes, long‑term stability, and robustness to disturbances. Experiments on an electromagnetic cavity, a particle‑in‑cell grid, and shallow‑water fluid demonstrate that the model can recover accurate constitutive functions, distinguish conserving from dissipating regimes, and transfer learned physics to unseen conditions, outperforming unconstrained models.
arXiv:2606. 12923v1 Announce Type: cross Abstract: AI alignment, interpretability, steering, and neural perturbation studies identify order-inducing objects.
The paper introduces a framework for intrinsic‑extrinsic coupling in learning dynamics, defining it via a continuation‑conditioned value of a constrained learning‑state intervention and observation‑relative fibers. It presents an executable finite‑frame classifier‑head that protects current logits while repairing historical margins, and distinguishes local admissibility, intervention value, and complete‑policy performance. Experiments on CLINC‑derived class‑incremental tasks, output distillation with RoBERTa, and SGDW dynamics demonstrate that coupling can produce both positive and negative interactions, and that coordinated content controls can match or exceed development gains while guided allocation reduces cross‑entropy loss compared to standard replay.
The paper introduces a framework for intrinsic‑extrinsic coupling in learning dynamics, where a learner’s current observations do not solely dictate its future training responses. It formalizes this coupling through a continuation‑conditioned value of a constrained learning‑state intervention and employs an executable finite‑frame classifier‑head to protect current logits while adjusting historical margins. Experiments across CLINC‑derived class‑incremental settings, output distillation with RoBERTa, and SGDW dynamics reveal that coupling can produce both positive and negative interactions, and that coordinated interventions can match or exceed development gains while reducing cross‑entropy loss compared to standard replay.
arXiv:2609.13197v1 Announce Type: new Abstract: Algorithmic Information Dynamics (AID) studies systems by perturbing them and measuring changes in algorithmic complexity, but its usual estimator, the...
The paper investigates how inherited state affects sub-agent performance in multi-agent frameworks, comparing three inheritance policies—Reset, Selective, and Full—across a ladder of Qwen3 models. It finds that reliance on stale state decreases with model capability, but a mid-capability model (Qwen3‑1.7B) exhibits a statistically significant local minimum of net harm, defining a "danger band." Selective handoff consistently improves accuracy over Full, especially within the danger band, while a fixed-threshold router fails on other datasets.
arXiv:2607. 12640v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards, and Group Relative Policy Optimization (GRPO) in particular, is now run routinely on a supervised checkpoint in the hope of producing a stronger agent.
arXiv:2609. 18966v1 Announce Type: new Abstract: Linear RNNs with input-dependent Householder-product transitions (DeltaNet/DeltaProduct-class) can provably represent hard state-tracking automata, yet trained models fail to length-generalize -- a gap recent work attributes to optimization, without a causal account.