While multi-agent systems (MAS) excel at complex reasoning, they are vulnerable to errors that intermediate agents introduce and downstream agents build upon. Auditing intermediate messages before they propagate requires an explicit standard, yet evaluation rubrics are typically authored by domain experts or written ag...
Yutong Wang, Siyuan Xiong, Xuebo Liu et al.· 0 citations
Advances in large language models (LLMs) are driving a shift toward using reinforcement learning (RL) to train agents from iterative, multi-turn interactions across tasks. However, multi-turn RL remains challenging as rewards are often sparse or delayed, and environments can be stochastic. In this regime, naive traject...
Aladin Djuhera, Swanand Kadhe, Farhan Ahmed et al.· 0 citations
Chain-of-Thought (CoT) empowers Large Language Models (LLMs) to tackle complex problems, but remains constrained by the computational cost and early token commitments in discrete reasoning traces. Recent latent reasoning approaches attempt to optimize efficiency by performing reasoning within continuous hidden states....
Jiecong Wang, Hao Peng, Zhanyi Wang et al.· 0 citations
This work introduces What If TSF (WIT), a benchmark for evaluating whether models effectively respond to future scenarios, and shows that factual accuracy gains do not consistently translate into appropriate responses to alternative scenarios, demonstrating the need to evaluate multimodal forecasting beyond predictive...
Jin-Kwan Jang, Hyunbin Jin, Hyungjin Park et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
An ARC solver is introduced, Knowledge Augmentation for Abstract Reasoning (KAAR), which encodes core knowledge priors within an ontology that classifies priors into three hierarchical levels based on their dependencies and maintains strong generalization across most LLMs.
Chao Lei, N. Lipovetzky, Krista A. Ehinger et al.· arXiv.org· 2 citations
A concept-centric framework for building agents that can learn continually and reason flexibly across multiple domains and offers several advantages, including data efficiency, compositional generalization, continual learning, and zero-shot transfer.
Jia-Yuan Mao, Joshua B. Tenenbaum, Jia-Jun Wu· Communications of the ACM· 14 citations· ⚡1
One deployed language model must often serve many compute budgets, yet serving each budget still means a separate training or compression run per point. We train a Telescopic Language Model (TLM) to be that continuum: a nested-capacity Transformer supervised by stochastic prefix supervision with a full anchor. At every...
Zhilin Guo, Boqiao Zhang, Hakan Aktas et al.· 0 citations
The ablations analyse why Foil works and turn the findings into design guidance for looped MoE: the returns of looping and of widening the expert layers amplify each other, routing confidence tracks healthy expert use better than load balance, and a sparse looped MoE should therefore use more experts per layer and more...
Shouren Wang, Chuan Ma, Mohsen Hariri et al.· 1 citation
MS-GLA consistently achieves higher accuracy and lower perplexity than GLA at matched parameter counts, validating multi-temporal resolution decomposition as a principled and effective extension of Gated Linear Attention.
This work introduces TWISTER, the first automaton-twisted Sequential Monte Carlo decoder for MDLMs, and proves that for regular language constraints, the Feynman-Kac correction is exactly computable, with the twists obtained efficiently using quantities pre-computed for step-exact sampling.
Aditya Thimmaiah, Lara Marinov, Jayanth Srinivasa et al.· 0 citations
This work introduces SEABench, a benchmark for studying endogenous misalignment arising from agent self-evolution, with 48 longitudinal task sequences that span multiple evolution surfaces, task domains, and harm types in a rich personal-assistant environment and shows that qualitatively different safety behaviors emer...
Saswat Das, Parvati Viswanathan, Daniel Donnelly et al.· 1 citation
FactorEngram, a factorized n-gram memory with basis-level contextual gating that improves language modeling and downstream task performance and retrieves sparsity-regularized coefficients over a dictionary of basis vectors shared across patterns, so related patterns can reuse common components.
Bo-Wen Yang, Jing-Bo Zhou, Qing-Hong Miao et al.· 0 citations
Flint-interactive gives AI agents a language for creating interactive visualizations. Click a chart, filter data, highlight related content, and explore beyond the view the author originally chose. The post Flint-interactive: A language for creating interactive data visualizations appeared first on Microsoft Research.
A new computational approach reveals how cells communicate to assemble into tissues. This could accelerate stem cell engineering, improve disease models, and inform therapies.
Microsoft Research Blog· microsoft.comOct 15, 2026
Zero-knowledge proofs can verify computing results without rerunning workloads or exposing sensitive data. Magnetar helps bring them into the mainstream with support for standard programming languages, major speed gains, and quantum-resistant security. The post Proof-based general-purpose computing with Magnetar appeared first on Microsoft Research.
MIT News · Artificial Intelligence· news.mit.eduSep 24, 2026
