GEN-1.5: Robots Learn New Tasks from a Single Video Demo
Generalist AI announces GEN-1.5, which enables robots to learn new skills based on a 3–12 second video demonstration without any further training (59% success rate out of the box, 83% success rate after 10 training iterations on 5 minutes of data)
Generalist AI has released GEN-1.5, an AI model that enables robots to learn new tasks with just one 3–12 second video demonstration. Out of the box, GEN-1.5 achieves a 59% success rate across 10 different tasks, and reaches an 83% success rate after 10 iterations of training on 5 minutes of data. When presented with a video demonstration, GEN-1.5 adds this information to its “context window” as a “physical prompt”, without needing to be retrained.
Currently, robots require hours of specialized data and extensive finetuning to learn new tasks. If we can make in-context learning for robots work on a large scale, we’ll be able to train robots to perform new tasks using nothing more than video tutorials — no ML engineering required!
Source: the-decoder.com
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →
Author
Evgenii Arsentev
PhD · Founder & CEO, Dad4Kids
Articles · Latest articles