Chen Peng @chennpeng

Verbal reinforcement learning is very effective for improving agentic performance on benchmarks, but we wanted to test how this transfers to robotics. We ran GPT 6 Astra, Gemini Argon 4, and Fable 5.1 on the LIBERO benchmark, and then iteratively developed skills which

2 4 12 16K 3

2 0 3 195 0

View Details

Runtime @runtimesi

a week ago

Introducing SimInfinity -- the world's largest open-source robotic simulation benchmark Coming soon: 10K open-source simulations with well-defined robotic tasks inspired by real-world activities across factories, logistics, homes, and beyond.

4 17 109 30K 125

View Details

Chen Peng @chennpeng

a week ago

@Luckyballa let’s do it!

0 0 1 147 0

View Details

Runtime @runtimesi

a week ago

2 4 12 16K 3

View Details

Chen Peng @chennpeng

a week ago

@lukedwagner12 @neomatterai @MainwaringLuke @ycombinator congrats luke!

0 0 1 39 0

View Details

Chen Peng @chennpeng

a week ago

@SaiVamshi1818 congratss!

0 0 0 32 0

View Details

Chen Peng @chennpeng

a week ago

@k_aviny @christopherbann @particularity_ congrats Kavin!

1 0 1 40 0

View Details

Chen Peng @chennpeng

a week ago

Same model. Same task. Different skills. 0/20 → 8/20 successes through skill refinement. The model isn't the whole story. How you use it matters.

Runtime @runtimesi

a week ago

1 8 35 47K 13

0 0 3 141 0

View Details

Runtime @runtimesi

a week ago

Frontier models like GPT-6 Astra and Gemini Argon 4 can control robots, but how much does the harness around them impact performance? We used our Nomad-Harness to test Gemini Argon on LIBERO tasks to test if text based skills could help with harder tasks that require more manipulation. We iteratively refined the skill with a critic agent reviewing failed attempts. Across 20 initial states per task, Argon’s native success rate improved: • Bowl into drawer: 0/20 → 17/20 • Mug into microwave: 0/20 → 8/20 These results show that reusable text instructions can improve robot control, and they also transfer between models, with improvements also shown for GPT-6-Astra and Fable 5.1 with the same skill.