Product · mentioned 5 times
A robot foundation model from Physical Intelligence that is fine-tuned for individual tasks.
Tap a name to go there · tap a line to see why they are linked
Y Combinator Startup Podcast· 13 Aug
The pre-trained Pi0.7 model matches or outperforms the fine-tuned Pi0.6 specialists that were developed with reinforcement learning post-training on throughput and success rate.The Generalist· 17 Mar
Karol Hausman says that with Pi 0.6 robots performed tasks at very high success rates, including making coffee on proper espresso machines for thirteen hours straight without cuts and cleaning after themselves.