A focused look at this week’s most significant advances in reasoning & inference scaling \u2014 12 papers surveyed from arXiv and leading AI labs.<\/p>\n
—<\/p>\n
Inference-time compute continues to reshape how we think about LLM capabilities. This week’s papers reveal new techniques for multi-step reasoning, process reward modeling, and the surprising effectiveness of simple verification strategies.<\/p>\n
A Process Harness for Uplifting Legacy Workflows to Agentic BPM: Design and Realization in CUGA FLO<\/strong> \u2014 Fabiana Fournier, Lior Limonad<\/em><\/p>\n We introduce the process harness, a new mechanism for uplifting legacy workflows into Agentic Business Process Management (Agentic BPM) without replacing the underlying workflow engine. A process harn…<\/p>\n arXiv<\/a><\/p>\n BetXplain: An Explanation-Annotated Dataset for Detecting Manipulative Betting Advertisements on Social Media<\/strong> \u2014 MSVPJ Sathvik, Parmitha Vangapadu, Nishit Rane, Sathwik Narkedimilli, Mark Lee, Akrati Saxena<\/em><\/p>\n The promotion of betting applications on social media platforms has increased significantly in recent years. Many of these advertisements use persuasive techniques that may mislead users, encourage ri…<\/p>\n arXiv<\/a><\/p>\n FlameVQA: A Physically-Grounded UAV Wildfire VQA Benchmark with Radiometric Thermal Supervision<\/strong> \u2014 Mobin Habibpour, John Spodnik, Niloufar Alipour Talemi, Fatemeh Afghah<\/em><\/p>\n Wildfire monitoring from UAVs requires reliable reasoning over complex aerial scenes, where smoke, scale variation, and occlusions often limit RGB-only interpretation. We introduce FlameVQA, a multipl…<\/p>\n arXiv<\/a><\/p>\n SPLIT: Separating Physical-Contact via Latent Arithmetic in Image-Based Tactile Sensors<\/strong><\/p>\n Wadhah Zai El Amri, Nicol\u00e1s Navarro-Guerrero<\/em><\/p>\n Training machine learning models for robotic tactile sensing requires vast amounts of data, yet obtaining realistic interaction data remains a challenge due to physical complexity and variability. Simulating tactile sensors is thus a crucial step…<\/p>\n arXiv<\/a><\/p>\n Multilingual Reasoning Cascades Need More Context<\/strong><\/p>\n Arnav Mazumder, Dengjia Zhang, Shuyue Stella Li, Yulia Tsvetkov, Niyati Bafna<\/em><\/p>\n Translation cascades for reasoning translate the query from another language to English, reason in English, and translate the answer back to the original language. This is a competitive approach to multilingual reasoning, but structurally lossy,…<\/p>\n arXiv<\/a><\/p>\n MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention<\/strong><\/p>\n MiniMax, :, Aili Chen, Aonian Li, Bangwei Gong, Binyang Jiang, Bo Fei, Bo Yang, Boji Shan, Changqi<\/em><\/p>\n We introduce MiniMax-M1, the world’s first open-weight, large-scale hybrid-attention reasoning model. MiniMax-M1 is powered by a hybrid Mixture-of-Experts (MoE) architecture combined with a lightning attention mechanism. The model is developed…<\/p>\n arXiv<\/a><\/p>\n Einstein World Models<\/strong><\/p>\n Munachiso Samuel Nwadike, Zangir Iklassov, Ali Mekky, Zayd M. Kawakibi Zuhri, Kentaro Inui<\/em><\/p>\n Does intelligence require the ability to reason about phenomena beyond direct experience? It is natural to suspect that some complex thought cannot be captured through language alone. However, of particular concern to this work, is whether…<\/p>\n arXiv<\/a><\/p>\n Forecasting With LLMs: Improved Generalization Through Feature Steering<\/strong><\/p>\n Humzah Merchant, Bradford Levy<\/em><\/p>\n Successful forecasting involves identifying patterns between historical and future states of the world which generalize to future observations. We apply LLMs to a variety of forecasting tasks and inspect their internal states using sparse…<\/p>\n arXiv<\/a><\/p>\n Do Safety Guardrails Need to Reason? LeanGuard: A Fast and Light Approach for Robust Moderation<\/strong><\/p>\n Dongbin Na<\/em><\/p>\n In order to screen a prompt or a response, the recent guardrail methods generate a chain-of-thought (CoT) before they issue a verdict. This design follows a common belief that step-by-step reasoning improves a decision. However, CoT also makes the…<\/p>\n arXiv<\/a><\/p>\n SSI-Policy: Learning Structured Scene Interfaces for Vision-Language Robotic Manipulation<\/strong><\/p>\n Kaijun Wang, Zikai Ouyang, Xuping Wu, Jinyi Hong, Wei Pan, Haibo Lu, Jia Pan, Wei Zhang, Linfang Zhe<\/em><\/p>\n Real-world robotic manipulation demands spatial grounding, task-aware reasoning, and precise control. Learning such capabilities becomes particularly challenging in the low-data regime. Prior methods often trade off scalable task-level reasoning…<\/p>\n arXiv<\/a><\/p>\n Parametric Open Source Games<\/strong><\/p>\n Aleksandar Todorov, Jesse ten Napel, Alexander M\u00fcller<\/em><\/p>\n Open-source game theory studies agents whose behavior may depend on one another’s decision procedures, but most existing models use discrete or symbolic programs. We introduce parametric open-source games, a continuous analogue of program…<\/p>\nTraining & Scaling<\/h2>\n
Reasoning & Inference<\/h2>\n
Additional Research<\/h2>\n