Open Models Almanac
Methodspost-trainingsupervised fine-tuning

single mention · post-training

Optimized Fine-Tuning

Optimized Fine-Tuning is a recipe for VLA adaptation that integrates parallel decoding, continuous action representations, and an L1 regression objective.1 This method enhances inference efficiency, task performance, and model input-output flexibility while maintaining algorithmic simplicity.2

How this definition was written

Summarised by glm-5.3-flash from 1 source; every sentence is checked against the quote it cites.

  1. integrates parallel decoding and action chunking, continuous action representations, and an L1 regression objective — Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success
  2. to enhance inference efficiency, task performance, and model input-output flexibility while maintaining algorithmic simplicity — Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success

Also called Optimized Fine-Tuning (OFT).

Filed under post-training › supervised fine-tuning