Oct 3, 2026
Sparse physical-agent execution experiments on RoboDojo, RoboCasa365 and K1; native gains unverified
Starter code for the Upskilling Together Temporal post-assessment
Adaptive robot execution: direct and hybrid controller ports, interaction memory, and EmbodiedSWE/FLUX adapters. CPU validated; GPU qualification pending.
Lean 4 formalizations for the Policy Gradients blog series (https://fa.bianp.net/blog/2026/policy-gradient/)
How to Email a Professor About Joining Their Lab — an essay with moderated Margin Notes
Qwen and pretrained flow-policy research on AMD, with an independent Tau0 demonstration track
A kid-friendly toolkit for writing your own text adventures in Python
