New Model Releases A

LLM personal health agents using sensor metrics promise to ease healthcare disparities, but an open-ended evaluation bottleneck limits clinical deployment. RubricsTree offers scalable, evolving open-ended evaluation across health memory and medical skills.

On-policy self-distillation (OPSD) helps post-training of LLMs but is unexplored for diffusion LLMs (dLLMs). Existing OPSD methods are autoregressive-centric, injecting privileged information via left-to-right prefix conditioning; this work studies self-distillation suited to dLLMs.

The paper introduces the Stanford EDGAR Filings Dataset (SEFD), an open reconstruction of SEC filings into layout-faithful MultiMarkdown, providing audited financial disclosures as token-efficient pretraining and evaluation data for financial language modeling.

The paper introduces DRFLOW, a benchmark for evaluating personalized workflow prediction in deep research systems, focusing on identifying concrete action-step workflows for enterprise tasks rather than generating reports or summaries.

Finite-dimensional diffusion policies suffer temporal drift from discretization that degrades long-horizon performance. The paper introduces a backward Kolmogorov equation that lifts diffusion policies into a Cameron-Martin space to make them more robust.

The classical continuous-time description of temporal-difference learning with linear features is an ODE capturing asymptotic mean dynamics but neglecting stochasticity. This work provides a diffusion approximation for TD learning under Markovian noise to capture those fluctuations.

The paper presents ReAge3D, a framework for identity-preserving 3D face re-aging that introduces a 2D diffusion-based re-aging model (DiffReaging) trained on synthetic image pairs and a center-out approach to maintain detail and view consistency.

AI agents are shifting from advisors to actors that book travel and run procurement. Existing animal-welfare benchmarks grade only text answers, so this work introduces an agentic benchmark testing whether implicit animal-welfare reasoning transfers to agent actions in frontier models.

The paper introduces the Certus Caliber Classification Gunshot Dataset (C3GD), a public dataset of firearm muzzle-blast sounds with over 8,000 field-collected data points from 28 firearms across 16 calibers, with detailed metadata.

The paper proposes a meta-knowledge reutilization framework for meta-reinforcement learning that learns task-level knowledge on a dynamics-simplified agent and transfers it to heterogeneous agents, using a Bayesian non-parametric prior to organize latent task modes.

The paper formalizes 'cognitive atrophy,' a process-level behavioural measure of AI-mediated mental-health support, capturing whether interactions help users keep reflecting, coping, and deciding, a dimension distinct from safety and static response quality.

Conversational AI localizes responses using user metadata, yet the regional biases this hidden context introduces remain poorly understood. The paper analyzes the unintended effects of geographic conditioning on large language model outputs.

LLM apps build prompts from templates, with Handlebars the default in Microsoft Semantic Kernel. While double-brace expressions HTML-escape values, triple-brace interpolation inserts them raw. The paper studies structural role injection and the limits of HTML auto-escaping.

Simon Willison released datasette-tailscale 0.1a0, a very experimental alpha plugin that runs a local Datasette server with a Tailscale sidecar so it is reachable inside your Tailnet via a chosen hostname. You launch it with an auth key and hostname. It relies on Python bindings for the experimental tailscale-rs library, and he filed an issue asking for a cleaner way to set up the proxy.

The paper develops an LLM-based text-to-SQL system using in-context learning, applied to the ALeRCE astronomical broker database, generating executable SQL from natural language and evaluated on a dataset of 110 NL/SQL pairs via step-by-step generation.

RAG grounds model outputs in external evidence, but its dominant evaluations and defaults are oriented toward factual question answering. HistoRAG embeds historical methodology into retrieval-augmented generation through critical technical practice for interpretive historical studies.

Score-based diffusion models use memoryless Brownian perturbations that yield tractable reverse-time dynamics. Volterra generative models introduce continuous-time perturbations with memory, generalizing diffusion-based generation.

NoiseTilt (NTRK) is a reward-guided diffusion sampler that injects reward gradients through the noise term, leaving the score kernel unchanged and needing only a single sample per step, improving reward alignment of pretrained diffusion models.

The paper analyzes, in the wild, what users ask large language models about security and privacy and how the models respond, characterizing the questions, response patterns and associated concerns.

Caregivers seek informational and emotional support in online communities where peers draw on personal narratives. As LLMs are designed as peer-like supporters, the paper examines the tension introduced when AI claims synthetic lived experience in caregiver support.

LLM agents rely on reusable tool specifications (skills), but real tasks require composing multiple skills. The paper formalizes compositional skill routing: decomposing a complex query into atomic sub-tasks, retrieving relevant skills, and composing them.

Looped transformers scale latent computation by repeating shared blocks, but sequential looping raises latency and KV-cache memory with loop count. Building on parallel loop transformers, LoopCoder-v2 makes loop count a practical knob for efficient test-time computation scaling.

Masked diffusion models (MDMs) have recently emerged as a generative approach. The paper investigates recursive scaling in MDMs, offering insights into their behavior and efficiency.

Dementia and depression are the most prevalent geriatric neuropsychiatric disorders, with overlapping symptoms complicating diagnosis. The study investigates open-weights LLMs for predicting dementia and depression severity from speech collected during clinical interviews.

Conditional independence testing is fundamental to statistics and causal inference. The paper proposes a fast nonparametric conditional independence test based on two-stage regression, aiming to improve computational efficiency and power.

The paper introduces LLM Consumer Behavior Theory, a proposed field analyzing consumer behavior in agentic markets where LLMs make consumption decisions on behalf of users, drawing on classical and behavioral economics alongside NLP.

Collective adaptive systems let nodes learn from locally sensed data, but privacy-sensitive data and node mobility hinder scaling. C2FL proposes clustered continual federated learning that handles spatial and temporal drift.

In masked diffusion language models, padding and semantic termination roles get entangled. VoidPadding introduces a [VOID] token to handle padding so that [EOS] can focus on signaling semantic termination, improving generation behavior.

The paper proposes an improved latent modeling approach for 3D MRI reconstruction and cross-contrast synthesis, addressing the heavy computational cost of large 3D volumes by recovering semantics first to better infer absent MRI contrasts.

The paper fine-tunes LLMs for passive estimation of depression severity from AI mental-health dialogue, exploring how conversational signals can indicate severity. Figures and efficacy are as reported by the source and not independently verified.