Blog

• Conditional slop(Feb 2026)
• I love incremental papers(Feb 2026)
• What makes a game fun(Feb 2026)
• Tricks for LLM diversity(Jan 2026)
• Benefit of the doubt(Jan 2026)
• We can't learn from zero rewards(Oct 2025)
• Advice on advice(Sep 2025)
• Startup as RL problem(Sep 2025)
• Bottom-up vs Top-down(Sep 2025)
• AI-to-Human Knowledge Distillation(Jun 2025)
• DeepSeek's open-source week(Mar 2025)
• Minimal Basics of CUDA Programming(Feb 2025)
• KV Cache Optimization(Jan 2025)