AI Wiki — Knowledge Base

“Marcus's repeated warnings about the 'wall of generalization' since 1998 have once again been proven true.”

“Marcus's repeated warnings about the 'wall of generalization' since 1998 have once again been proven true.” Marcus氏が1998年から繰り返す'汎化の壁'の警告。またも証明された。完璧なAIを待つより、今の限界を熟知して使いこなすチームが勝つ。少人数ゆえの意思決定の速さとリスク許容度が

safetygary-marcus--x

12 May 2026

Search: “safety”

MAG-VLAQ: Multi-modal Aerial-Ground Query Aggregation for Cross-View Place Recognition

Make Each Token Count: Towards Improving Long-Context Performance with KV Cache Eviction

MapFormer: Self-Supervised Learning of Cognitive Maps with Input-Dependent Positional Embeddings

“Marcus's repeated warnings about the 'wall of generalization' since 1998 have once again been proven true.”

MARLaaS: Multi-Tenant Asynchronous Reinforcement Learning as a Service

MARS-SQL: A multi-agent reinforcement learning framework for Text-to-SQL

MASS-DPO: Multi-negative Active Sample Selection for Direct Policy Optimization

Mem-W: Latent Memory-Native GUI Agents

Memory Inception: Latent-Space KV Cache Manipulation for Steering LLMs

MePo: Meta Post-Refinement for Rehearsal-Free General Continual Learning

Meta offers to give rival AI chatbots free access to WhatsApp for a month while it discusses commitments with EU antitrust regulators to address their concerns (Foo Yun Chee/Reuters)

Meta-reinforcement learning with minimum attention

Metropolis-Adjusted Diffusion Models

Mid-Training with Self-Generated Data Improves Reinforcement Learning in Language Models

Mitigating Data Scarcity in Spaceflight Applications for Offline Reinforcement Learning Using Physics-Informed Deep Generative Models

Monocular Biomechanical Tracking of Fingers with Inverse Kinematics to Foundation Models

Morphology-Aware Graph Reinforcement Learning for Tensegrity Robot Locomotion

MTA-RL: Robust Urban Driving via Multi-modal Transformer-based 3D Affordances and Reinforcement Learning

Multi-layer attentive probing improves transfer of audio representations for bioacoustics

Multi-scale Predictive Representations for Goal-conditioned Reinforcement Learning

Multimodal Representation Learning Conditioned on Semantic Relations

MURPHY: Feedback-Aware GRPO with Retrospective Credit Assignment for Multi-Turn Code Generation

MVB-Grasp: Minimum-Volume-Box Filtering of Diffusion-based Grasps for Frontal Manipulation

Natural Policy Gradient as Doubly Smoothed Policy Iteration: A Bellman-Operator Framework

Navigating EU AI Act requirements for LLM fine-tuning on Amazon SageMaker AI

Navigating LLM Valley: From AdamW to Memory-Efficient and Matrix-Based Optimizers

NEO: No-Optimization Test-Time Adaptation through Latent Re-Centering

Neural at ArchEHR-QA 2026: One Method Fits All: Unified Prompt Optimization for Clinical QA over EHRs

Neural Co-state Policies: Structuring Hidden States in Recurrent Reinforcement Learning

Neuromorphic Reinforcement Learning for Quadruped Locomotion Control on Uneven Terrain

not every day @scaling01 and I agree. but he’s right. and most people won’t notice it happening, per the work of @informor.

note that I said “Even @haider1” because he is often optimistic, but he has informed me (with receipts that he shared) that he expressed som…

Now You See That: Learning End-to-End Humanoid Locomotion from Raw Pixels

Offline Preference Optimization for Rectified Flow with Noise-Tracked Pairs

On-Policy Distillation with Best-of-N Teacher Rollout Selection

On the Generation and Mitigation of Harmful Geometry in Image-to-3D Models

On Variance Reduction in Learning Mean Flows

Open Ontologies: Tool-Augmented Ontology Engineering with Stable Matching Alignment

OpenAI’s Sam Altman’s personal investments are coming under intensifying scrutiny from Republicans following an April article in The Wall St…

OpenClaw-RL: Train Any Agent Simply by Talking

Opinion: It is rare for the US public to agree on anything these days. Fear of AI is as close to a national consensus as it gets. A clear ma…

Overcoming Catastrophic Forgetting in Visual Continual Learning with Reinforcement Fine-Tuning

Path-Coupled Bellman Flows for Distributional Reinforcement Learning

PATRA: Pattern-Aware Alignment and Balanced Reasoning for Time Series Question Answering

Pay attention to this one if you build research or knowledge-work agents. Most research-agent systems produce uniform outputs regardless of …

Perception Without Engagement: Dissecting the Causal Discovery Deficit in LMMs

Personalizing LLMs with Binary Feedback: A Preference-Corrected Optimization Framework

PFN-TS: Thompson Sampling for Contextual Bandits via Prior-Data Fitted Networks