Geometry of Values: Task Vector Composition for Ethical Preference Alignment in Language Models
How Fyxer built an AI executive assistant people trust
5 min
StepAudio 3 Music Technical Report
Domain-Specific Hallucination Detection in Large Language Models
1
RetroThinker: Enabling Retrospective Thinking in Speech LLMs
1
You Can't Prefer Emotions You Don't Sample: Intensity Undershoot in DPO-Tuned LLMs
PLC-DPO: Posterior Label Correction in Noisy and Ambiguous Preference Optimization
AgenticGen: Reward-Guided Agentic Video Generation for Advertising
