No Task Vector Is an Island: A Comprehensive Study on the Composability of Task Vectors from On-Policy Distillation
Research
Research News
New papers, breakthroughs and academic AI research
NarrativeSteward: Coordinating Delegation, Guidance, and Verification in Agent-Assisted Interactive Narrative Authoring
Research
Working Around the Compute Ceiling: Byte-Exact Memory in Galahad Makes LLM Reading a One-Time Cost LLM Reading a One-Time Cost
Research
Belief-Based Maximum Occupancy Principle and Active Inference
Research
Who Said What, and Will It Be Remembered? Evaluating Persistent Speaker Attribution Across Meetings
Research
WSPolypNet: Weakly Supervised Polyp Localization in Colonoscopy Videos
Research
From Imitation to Reward Discovery: On-Policy Warmup for Agentic RL
Research
Raw-Routed Mixture of Adapters: A Causal Intervention for Routing Collapse in Time Series Foundation Models
Research
CAMOS: Coupled Oscillatory State-Space Model for Multimodal Clinical Time-Series
Research
Speculative Safety Honeypot: Toward Proactive Defense Against Multi-turn Agent Attacks
Research
MedRECT: A Bilingual Medical Reasoning Benchmark for Error Correction in Clinical Texts
Research
Mutual Equilibrium: Multimodal Representation Learning through Reciprocal Feedback
Research