Cover image

Hard Problems Pay Better: Why Difficulty-Aware DPO Fixes Multimodal Hallucinations

Training data has a bad habit: the easiest examples talk the loudest. Anyone who has trained a model on preference pairs knows the scene. One answer is clearly grounded in the image; the other confidently invents an object, a color, or an action that is not there. The model learns the contrast quickly. Everyone applauds. The loss goes down. The dashboard looks obedient. ...

January 5, 2026 · 15 min · Zelina
Cover image

Planning Before Picking: When Slate Recommendation Learns to Think

A list of individually excellent items can still be a terrible list. Ask anyone who has attended a conference with five brilliant speakers, no agenda, and three consecutive sessions on the same topic. Recommendation systems have the same problem. A conventional recommender can assign highly accurate scores to individual videos, products, or articles, then still assemble a repetitive, badly ordered, or strangely balanced feed. Each item wins its private competition. The user receives the collective consequences. ...

January 2, 2026 · 18 min · Zelina
Cover image

When Safety Stops Being a Turn-Based Game

Jailbreaks are not polite enough to wait their turn. That is the awkward weakness in many safety-training pipelines. A model is attacked, patched, tested, and released. Then another attack appears, usually crafted with more creativity than the previous defense assumed. The safety team patches again. The benchmark improves. The real attack surface moves. Everyone calls this iteration, because “organized whack-a-mole with GPUs” sounds less respectable. ...

December 28, 2025 · 15 min · Zelina
Cover image

Judo, Not Armor: Strategic Deflection as a New Defense Against LLM Jailbreaks

TL;DR for operators Most LLM safety systems still assume that, when a model sees a harmful request, the correct behaviour is refusal. That works until the attacker stops arguing with the prompt and starts interfering with generation itself. The paper behind this article, Strategic Deflection: Defending LLMs from Logit Manipulation, proposes SDeflection: a fine-tuning method that teaches a model to answer in a safe, topic-adjacent way rather than relying only on explicit refusal language.1 The model does not provide harmful instructions. It redirects the subject toward harmless information that is close enough to the original topic to survive attacks that try to force compliance-style openings. ...

July 31, 2025 · 16 min · Zelina