single mention · model architecture
Per-head per-layer key-value conditioning
Keeps VLM keys and values separated by head and projects each head into the corresponding action-expert head.
The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See multimodal architecture for the methods filed beside it.
Also called per-head per-layer KV conditioning variant.
Filed under model architecture › multimodal architecture