TERRA: Task-Embedded Reasoning and Representation Architecture for Cross-Domain Applications

ArXi:2606.01520v1 Announce Type: new A single action-conditioned latent predictive architecture can in principle be trained on the structured state of a driving scene, a robot workspace, or a financial order book. The ingredients for doing so within any one domain already exist and are individually validated: masked-latent prediction, action-conditioned latent world models, discrete action tokenization, and joint-embedding prediction on voxelized state.