EARS introduces explanatory abstention in sub-agents to improve reliability in large-scale multi-agent systems. By providing actionable failure rationales to coordinators, EARS increases the overall response pass rate from 68.5% to 78.9% in a production e-commerce assistant.
LegalWorld models Chinese civil litigation as a causally connected chain of five stages, based on 75,309 judgments. It includes reusable infrastructure to maintain consistency across stages and enables LongJud-Bench to evaluate agent performance across all phases, revealing significant capability gaps between models in different legal tasks.
HandwritingAgent synthesizes natural handwriting in SVG format without style-specific training. It uses a large reasoning model to generate stroke sequences in a grid canvas, conditioned on text input and a reference style image, enabling efficient, controllable, and generalizable handwriting generation.
GateMem introduces a benchmark for multi-principal shared-memory agents, evaluating utility, access control, and active forgetting across medical, office, education, and household domains. No method achieves strong performance in all three governance aspects, with long-context prompting offering best results at high cost, while retrieval-based and external-memory approaches reduce cost but still suffer from information leaks.
A data-centric approach improves long-context reasoning in large language models, using eight curated datasets with 14K examples across retrieval, multi-evidence synthesis, and reasoning tasks. When paired with minimal outcome-based GRPO training, it achieves average gains of +7.2 to +6.4 points on seven benchmarks, outperforming prior RL training sets, and enhances agentic performance by +4.8 and +7.0 points on GAIA and BrowseComp respectively.
ScholarSum introduces a hierarchical knowledge graph framework that emulates a student-teacher process for scientific summarization. It generates fluent, factually consistent summaries by first structuring documents into semantic units, then refining drafts through evidence retrieval and iterative review by a teacher-like component. Experiments show ScholarSum outperforms existing methods in completeness and factual faithfulness.
A new pipeline uses language models to recommend minimal, interpretable changes to patient-doctor communication features like tone and personalization. These changes increase predicted positive feedback by an average of 6.41% and are non-negative for 93.31% of cases, without altering medical content.
SAGE is a multi-agent framework for prompt optimization that combines diagnostic code execution with quantitative validation. It improves mental-health chatbot retention by aggregating eight cycles of noisy A/B tests into statistically significant gains, demonstrating effectiveness in open-ended dialogue tasks through qualitative and quantitative feedback integration.
SenFlow introduces a novel method for detecting AI-generated text in hybrid documents by modeling inter-sentence dependencies. It achieves state-of-the-art performance on MOSAIC, a benchmark of 16,000 documents from PubMed and XSum, with a +4.15 pp Macro-F1 gain on cross-domain transfer. SenFlow reveals that AI-generated content still exhibits generator-dependent sentence-length patterns, exploitable by sentence-level detectors despite perplexity filtering.
Decoupled Search Grounding (DSG) separates search functionality from reasoning models, enabling vendor-agnostic, tunable, and reusable search grounding. DSG achieves near-native accuracy on SimpleQA with 91% lower search cost and 99.4% warm-cache hit rate, while reducing latency by 68% and preserving concise output contracts.
GraphPO introduces a directed acyclic graph framework to represent reasoning rollouts, merging semantically equivalent paths to reduce redundant exploration. It assigns efficiency and correctness advantages to edges, improving inference efficiency and process supervision while reducing advantage-estimation variance. Experiments show GraphPO outperforms chain- and tree-based methods on three LLMs across reasoning and agentic search tasks under identical token or response budgets.
Process-level coordination control adds value only when the initial majority consensus is unreliable, the task is recoverable, and unguided interaction fails to repair errors. Across multiple models and tasks, no leadership style outperforms others in accuracy, aligning with contingency theory rather than suggesting a failure of the approach.
The Human-AI Coevolution Dynamics Framework (HACD-H) introduces a unified model for long-term human-AI interaction, integrating emotional adaptation, memory, and personality into a self-organizing social cognitive system. Results show social intelligence emerges through coevolution, with a significant negative correlation between social intelligence and social cognitive energy (r = -0.391, p < 0.001), and progressive energy reduction over time in interaction trajectories.
IndicContextEval introduces a 56-hour multilingual benchmark featuring natural speech from 555 speakers across 8 Indian languages and 23 domains. It employs a 7-level prompting framework to progressively test context utilisation, including metadata, descriptions, and adversarial inputs. Evaluation of five models shows significant differences in contextual grounding, underscoring the need for explicit assessment of context use in AudioLLMs.
R2D-RL bridges RCSS2D and HELIOS-based clients with a Python MARL interface using shared-memory and cycle-level synchronization. It enables full-field and scenario-based training with configurable opponents, action masks, EPV-based reward shaping, and parallel execution, including front-goal scenarios and an 11-vs-11 benchmark with baseline results.
ProfiLLM introduces an agentic LLM pipeline that extracts behavioral signals from ride-hailing logs to generate user profiles. It achieves up to +6.14% relative AUC improvement and up to +4.35% GMV gain in dispatching simulations, with consistent online A/B test results showing +0.47% GMV, +0.33% Completion Rate, and -0.82% Cancel-Before-Accept rate improvements.
Reinforcement learning lacks foundation models despite synthetic MDPs being feasible. A proof-of-concept shows a single model trained on synthetic MDPs solves tabular benchmarks without tuning, outperforming existing methods in online settings and matching them offline.
Intelligence is embedded in the space itself, where scenes induce a Riemannian metric on configuration manifolds. A single Encoder-Router network uses semigroup-superposition to generate this metric, enabling zero-shot generalization across unseen obstacle configurations with large cost differences between collision-free and obstacle-penetrating paths.
A data-centric approach improves long-context reasoning in large language models, using eight curated datasets with 14K examples across retrieval, multi-evidence synthesis, and reasoning tasks. When paired with minimal outcome-based GRPO training, it achieves average gains of +7.2 to +6.4 points on seven benchmarks, outperforming prior RL training sets, and enhances agentic performance by +4.8 and +7.0 points on GAIA and BrowseComp respectively.
Skill-MAS introduces a new approach that decouples experience retention from parametric updates by modeling orchestration as an evolvable Meta-Skill. It uses a closed-loop process involving multi-trajectory rollouts and selective reflection to distill reusable strategy principles, achieving strong performance gains and robust transferability across tasks and LLMs.