Open Models Almanac
Methodsoptimizationtraining stability

single mention · optimization

Action and proprioceptive-state normalization

Action and proprioceptive-state normalization rescales robotics training inputs and outputs using statistics of the training data.1 Statistics like mean, scale, and percentiles are precomputed across the full dataset during preprocessing and stored with each shard, with options such as percentile-based methods to handle outliers.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. Actions and proprioceptive states are normalized during dataloading time and denormalized at inference time — VLA Foundry: A Unified Framework for Training Vision-Language-Action Models
  2. Statistics are precomputed across the full dataset during preprocessing and stored in a stats.json file alongside each dataset shard — VLA Foundry: A Unified Framework for Training Vision-Language-Action Models

Also called Normalization.

Filed under optimization › training stability