1.7h of intervention data beats 21h of demos: finetuning π0.5 on manufacturing
DominiqueCAPaul · x · 2026-09-24
A team finetuned Physical Intelligence's π0.5 on a real manufacturing task and shared key findings:
- 1.7 hours of high-quality data with interventions beat 21 hours of pure demonstrations
- 4 hours of data across 5 scenes beat 4 hours in the eval scene by 30 percentage points
- Tuning RTC improved the same checkpoint from 76% to 93%
The takeaway: data quality and diversity matter more than volume.
More from Embodied
- The world is the installed base: the underrated case for humanoid robots — r0ck3t23 · 2026-09-24
- GTSAM 4.3 ships with legged navigation, CUDA-accelerated factor graphs and continuous-time trajectory estimation — fdellaert · 2026-09-24
- Workbench open-sources an agentic lab workspace for hardware engineering on the Ambion kernel — andreisavu · 2026-09-24
- Hugging Face Tutorial: Accelerating Robotics Simulation with NVIDIA Warp and MjWarp — Hugging Face Blog · 2026-09-24
- Skild AI trains Unitree G1 to play soccer with 140 years of simulated self-play — Distinct-Question-16 · 2026-09-24
- Deepak Pathak: Earth's Only AGI Came From Billions of Years of Physical Self-Play, and Robots Will Scale It — deepakpathak · 2026-09-24