Open Models Almanac
Methodspost-training

single mention · post-training

Three-stage VLA training: action-modality injection, CoC reasoning SFT, and RL post-training

Transforms a vision-language model into a reasoning-capable autonomous-driving policy through action-modality injection, CoC reasoning supervised fine-tuning, and reinforcement-learning post-training.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See post-training for the methods filed beside it.

Also called three-stage training strategy.

Filed under post-training