Why visual concepts resist unlearning: larger train/test gap than text
_vztu · x · 2026-10-17
Thread entry 3: multimodal unlearning is harder because a password is one exact string while a visual concept like Tokyo Tower appears in endless angles, lighting, and styles. With the same data and training, the train/test gap is much larger for image+text inputs than for text-only ones.
More from Research
- A single Qwen3 neuron tracks both "Colorado" and Tornado HTTP client code — livgorton · 2026-10-17
- Vals Science Index: OpenAI Models Lead Overall, but No Model Sweeps Every Science Discipline — BenBlaiszik · 2026-10-17
- First-author PhD paper frames AI text detection as continual learning, accepted at NeurIPS 2026 — lucy3_li · 2026-10-17
- Regents Labs curates 17 AI/agent science papers weekly with AI-generated deep reads — seanwbren · 2026-10-17
- A 'pre-numeric' AI math layer trains patterns before numbers exist — debreuil · 2026-10-17
- research-rag: local CPU-only PDF/ebook search with browser, CLI and MCP access — kishkisan · 2026-10-17