karpathy.bearblog.dev
Summary
karpathy.bearblog.dev covering LLMs, Reinforcement Learning, Jagged Intelligence, and AI Agents. Notable guests include Andrej Karpathy. Episodes span from Apr 2019 to Apr 2026.
AI agents reached a significant inflection point in late 2025, fundamentally changing the role of a programmer from a code writer to an orchestrator of AI-driven tasks. A new paradigm, "Software 3....
In 2025, Reinforcement Learning from Verifiable Rewards (RLVR) became a new, dominant stage in LLM training, reallocating compute from pretraining to achieve significant capability gains per dollar...
Richard Sutton, author of the influential essay "The Bitter Lesson," argues that current Large Language Models (LLMs) are not truly aligned with its principles due to their reliance on finite, huma...