This paper presents a predictive law for calculating the uplift of Large Language Model (LLM) ensemble performance based on diversity of thought. It provides an experimentally verified formal law and a compact heuristic for calculating uplift, which is tested on various datasets.

Researchers propose Pheromone-Guided Policy Optimization (PhGPO) to improve long-horizon tool planning for Large Language Model (LLM) agents, leveraging historical trajectories to guide policy optimization and improve tool transitions.

Researchers propose Scalable Delphi, a method for using large language models to estimate structured risk by adapting the Delphi method for LLMs with diverse expert personas, iterative refinement, and rationale sharing.

Researchers propose Graph of Concept Predictors (GCP), a reasoning-aware active distillation framework for Large Language Models (LLMs) that improves performance under limited annotation budgets while yielding more interpretable and controllable training dynamics.

A benchmarking framework, AstroAgentBench, is introduced for evaluating agentic planning in space mission planning tasks. It assesses the performance of large language model (LLM) agents in domains like scheduling, observation planning, and constellation design. The framework provides a standardized evaluation methodology and highlights the importance of task-contract formulation, verifier feedback, and search adaptation in achieving high-quality plans.

Researchers investigate whether large language models (LLMs) can automatically formalize symbolic constraints for Neuro-Symbolic predictors, making them suitable for high-stakes applications. They introduce a benchmark to evaluate this and find that LLMs can generate formulas similar to human experts, leading to high-quality predictions.

Researchers propose a solution for smart manufacturing using LLM-based agents to generate production sequences, handle runtime faults, and coordinate over MQTT with real-time updates of the factory state.

This research paper proposes a new architecture for federated learning of large language models (LLMs) over mobile networks, addressing challenges in mobile RAN transport. The proposed approach uses in-network aggregation to reduce the number of transfers and enable selective provisioning of optical connectivity.

A new IDS system, Jev-IDS, is proposed based on the Jev System One Model (SOM) for network intrusion detection. It uses a Large Language Model (LLM) to analyze flow records directly, offering faster and cheaper detection with higher recall compared to traditional machine-learning-based IDS. The system asks the LLM two questions per flow and achieves an F1-score of 0.859 on a 300-flow pilot test.

The authors propose Divide-and-Remember (D&R), a recursive memory method for Vision-Language-Action (VLA) models that learns to remember relevant information from a long history of observations. This method is efficient, scalable, and achieves state-of-the-art results on a benchmark of long-horizon manipulation tasks.

Researchers propose a post-training method for multi-token prediction heads in language models, achieving similar speedup to joint pre-training with significantly fewer tokens. They also introduce a relaxation of draft token verification and an adaptive controller for dynamic MTP head engagement.

A self-evolving framework, RuleEvolve, for coding rules in AI coding agents uses an LLM-powered mutator module to generate variants and a judge module to evaluate and update the pool with the best-performing ones, outperforming manual engineering and existing prompt optimization baselines in functional correctness, code length, and generation cost.

A research paper compares the performance of two hierarchical red team agent architectures, RL+RL and LLM+LLM, in different environments, revealing environment-dependent inversion and highlighting the importance of considering specific failure modes when designing hybrid planner-executor architectures.

A research paper proposing a design pattern called 'bounded-fidelity sim-as-demo-stage' to improve audit-chain reproducibility in governance benchmarking of LLM-driven robots. The pattern suppresses contact physics within explicitly bracketed handoff envelopes while preserving full dynamics elsewhere.

Mingbird is a local-first agent harness designed for small open models to complete real tasks. It addresses issues such as tool prefill overflows, self-correction divergence, and task abandonment by introducing ten mechanisms, including a byte-level net-zero prefill budget and signature-level loop detection. Mingbird outperforms other harnesses on various benchmarks, achieving 0.886 overall on LRAB and 0.856 on τ^2-bench.

A multi-agent LLM system improves personalized health checkup interpretation and guidance by executing tasks in parallel and synthesizing outputs. It outperforms single-agent systems in weighted LLM-judge score, usefulness, consistency, and handling of compound queries, but increases latency and cost.

DRelay proposes a method for prefix-aware selective repair of candidate selections in parallel speculative decoding for large language models (LLMs), improving drafting efficiency and end-to-end decoding performance.

A research paper introduces PoS, an inference-time framework for constructing and maintaining explicit belief states in large language model (LLM) agents, enabling them to undertake complex tasks with a coherent understanding of the current world.

This paper introduces ICR, a framework to evaluate communication in LLM multi-agent systems. It helps disentangle the effects of communication, architecture, and reasoning on system performance.

A new approach to multi-agent workflow optimization, InFlowOp, is proposed. It uses a label-free cost function to determine task decomposition and agent assignment, and corrects faults during execution with the cheapest move. InFlowOp outperforms single-agent baselines in various domains and backbones.