News
AI, cybersecurity, machine learning, scraping & programming news.
The Missing Primitive: Diagnosing and Repairing Mathematical Reasoning in Large Language Models
Large language models now solve competition-level mathematics yet it remains unclear whether they grasp the structural logic behind their solutions. This paper introduces the Mathematical Primitive as a probe for structural unders...
Decoding Looped Transformers Better for (Almost) Free
Looped Transformers achieve parameter efficiency by repeatedly executing a shared block across recurrent loops. Each loop yields an intermediate representation decodable for the same next token, yet standard decoding discards earl...
SoftServe: A Scalable Quasi-Newton Method for Deep Learning
Quasi-Newton methods have long been among the most effective approaches for large-scale unconstrained convex optimization, yet two obstacles have limited their adoption in deep learning: non-convexity and enormous parameter sizes....
From Gradients to Capabilities: Understanding Multi-Teacher On-Policy Distillation
Multi-teacher on-policy distillation aims to combine the strengths of RL-trained teachers in a single student, but how teacher signals affect parameter changes remains underexplored. Researchers studied Qwen3-1.7B with four domain...
Muon meets Tamed Langevin: Momentum Preconditioning beyond Convex and gradient-Lipschitz Potentials
The Muon optimiser has become a workhorse for large language model pretraining, delivering state-of-the-art efficiency gains over AdamW. Its defining trick is orthogonalising matrix-valued momentum, replacing every nonzero singula...
Faynt: Scaling and Optimizing Policies for Competitive Melee
Competitive real-time strategy and fighting game AI have seen rapid progress, but Super Smash Bros. Melee remains a particularly challenging benchmark. Its 52-character roster, frame-perfect inputs, and continuous state space dema...
MIRTO: a registration-gated, multiverse-tested evaluation protocol for unsupervised anomaly segmentation in brain MRI
Competitive real-time fighting game AI has long been a rigorous testbed for reinforcement learning and policy generalisation. Super Smash Bros. Melee, with its 52-character roster, frame-perfect inputs, and continuous state space,...
Harnessing Domain Specialists in Multimodal Mixture-of-Experts for Efficient Adaptation
Sparse Routing in Multimodal Mixture-of-Experts Architectures Mixture-of-Experts (MoE) architectures have become a cornerstone for scaling modern multimodal models. The standard Transformer feed-forward layer replaces a single den...
Argo-Bench: Evaluating Data Agents on Enterprise-Scale Workflows
Text-to-SQL benchmarks have become the default way to measure progress on data agents. Spider, BIRD, and their variants ask a model to write a query, then check whether the result matches a gold answer. The best models now clear 9...
Where-OPD: Spatially Guided On-Policy Self-Distillation of MLLMs with Synthetic Scenes
On-policy self-distillation has recently emerged as an effective approach for improving language-model reasoning by supervising students with a frozen or EMA version of themselves that receives privileged information. Its applicat...