macaron-v

Definition

A model name where the -V suffix follows the common convention marking a vision-capable variant, as with -V and -VL suffixes elsewhere in the field. The practical difference from the base model is an added visual encoder plus a projection that maps image features into the language model's representation space, letting it accept images alongside text. What the base name itself denotes is defined by the paper that introduces it.

Explainers using this term

• Paper Walkthrough: Macaron-V1 — A Frozen Base plus a Mixture of LoRAs, Built to Keep Learning After LaunchMacaron-V1: Towards Open Continual Learning with Self-Improvement and Mixture-of-LoRA