This page presents comprehensive results from Real-Time Chunking (RTC) experiments across different models, flow matching steps, and sigma_d values. The experiments demonstrate how these parameters affect the denoising process and final action predictions.

Models tested: PI0.5, SmolVLA
Flow matching steps: 10, 20, 50, 100
Sigma_d values: 0.1, 0.2, 0.5, 0.8, 1.0, 1.2, 1.5
Modes: Classical RTC (fulltrajectory=false), Full Trajectory (fulltrajectory=true)

Classical RTC: Uses RTC guidance during denoising with chunked trajectory processing.
Full Trajectory: Processes the complete trajectory without chunking.