StraTA is a simple framework that introduces an explicit trajectory-level strategy into agentic reinforcement learning (RL) and trains strategy generation and action execution jointly with a hierarchical GRPO-style rollout design, further enhanced by diverse strategy rollout and critical self-judgment.
Xiangyuan Xue, Yifan Zhou, Zidong Wang et al.· arXiv.org· 1 citation
Full-duplex spoken dialogue systems can model natural conversational behaviours such as interruptions, overlaps, and backchannels, yet such systems remain largely unexplored for Indian languages. We present the first open, reproducible full-duplex spoken dialogue system for Hindi by adapting Moshi, a state-of-the-art d...
Bhaskar Singh, Manas Dhir, Shobhit Banga et al.· 0 citations
Autonomous GUI agents face two fundamental challenges: early stopping, where agents prematurely declare success without verifiable evidence, and repetitive loops, where agents cycle through the same failing actions without recovery. We present VLAA-GUI, a modular GUI agentic framework built around three integrated comp...
Qijun Han, Haoqin Tu, Zijun Wang et al.· 0 citations
The Affective Flow Language Model is proposed, which models multi-turn emotional support as an affective utility flow evolving along dialogue trajectories, providing a process-based approach to dynamic affect modeling and continuous strategy optimization.
Cheng Zou, Ning Wang, Tiesunlong Shen et al.· arXiv.org· 2 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Pretrained language models (LLMs) are often constrained by their fixed tokenization schemes, leading to inefficiencies and performance limitations, particularly for multilingual or specialized applications. This tokenizer lock-in presents significant challenges. standard methods to overcome this often require prohibiti...
Shaurya Sharthak, Vinayak Pahalwan, Adithya Kamath et al.· 0 citations
A large-scale dataset specifically designed to assess capabilities and train models on the task of hallucination detection in medical texts, MedHal is presented, which enables more efficient evaluation and training of medical text generation systems while reducing reliance on costly expert review.
Fabrice Lamarche, Gaya Mehenni, Neshat Elhami Fard et al.· 4 citations
This study presents the first comprehensive review of the state of affairs in NLP research across the three major Nigerian languages (NaijaNLP), quantitatively assess the available linguistic resources and identify key challenges.
A novel architecture that strictly decouples statistical preference learning from semantic intent parsing is explored, which achieves the lowest cumulative regret and highest test accuracy, significantly outperforming traditional memory-augmented agents.
Results position acuity identification as a distinct safety-critical capability and show that AcuityBench enables systematic comparison and stress-testing of how well models guide users to the right level of care in real-world health use.
R. Linzmayer, Ge Lin, D. Coneybeare et al.· arXiv.org· 0 citations
Combee is proposed, a novel framework to scale parallel prompt learning for self-improving agents that speeds up learning and enables running many agents in parallel while learning from their aggregate traces without quality degradation.
Frontier AI systems are increasingly capable and deployed in high-stakes multi-agent environments. However, existing AI safety benchmarks largely evaluate single agents, leaving multi-agent risks such as coordination failure and conflict poorly understood. We introduce GT-HarmBench, a benchmark of 1,535 high-stakes sce...
FlyAOC is presented to evaluate AI agents on end-to-end agentic ontology curation from scientific literature and reveals system-level failure modes that model-only evaluations do not capture.
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
