New VLM paper finds image and text align from layer 1 in newer models
stanislavfort · x · 2026-07-22
A VLM paper challenges the common belief that image and text representations only align in the late layers.
- The authors used DeepDream-style optimization on Gemma 3.
- They found alignment appears from layer 1 in newer models, rather than only in the top layers.
- The post points to a paper co-authored by evzenwy, javirandor, floriantramer, and stanislavfort.
More from Research
- Google DeepMind pilots world's first double-blind evaluations for frontier models — scychan_brains · 2026-10-26
- Rewriting scaling law study: optimal synthetic data share rises from 10% to 32% with compute — heghbalz · 2026-10-26
- FLASH: GPU-native deformable sim learns cloth folding in minutes on one RTX 5090 — ChongZzZhang · 2026-10-26
- LEMON: zero-shot LLM misbehavior monitor flags hallucinations from attention maps — yoavartzi · 2026-10-26
- Behind the EMNLP keynote: LMs as simulated learners to generate child language acquisition hypotheses — najoungkim · 2026-10-26
- The paper behind the EMNLP 2026 keynote on bigram 'simulated learners' — najoungkim · 2026-10-26