-
How to Solve Hallucination
Ask an LLM how sure it is about something and when it answers you, it gives you a number based on vibes. “I’m about 90% confident” comes out of the…
-
Talk: Minimizing Idle GPU time for Reinforcement Learning on Computer Use Tasks
Cua Fleet (10:50 – 15:02): A solution for optimizing GPU utilization during RL (Reinforcement Learning) training. It addresses the high cost of idle GPUs while waiting for sandbox environments to…
-
Scientists, Not Lawyers
Being a contrarian means disagreeing with people. Even good-faith disagreement is hard, and some people take it harder than others. One way to disagree without cornering the other person is…
-
Slowing Down My Coding Agents to Get More Done
Your backlog is sitting idle waiting to get worked on, and it could be showing up pre-baked, ready for verification. openclaw polls my Linear backlog every 30 minutes for tickets…
-
Solve Hi-Q with AlphaZero and Curriculum Learning
On a snowy day at my parents’ house, I was passing time with family and reading a blog post about planner programming when my grandmother invited me to play Othello…