Antonio Troise·Oct 8ASD-STE100 and ELI5: 20 Ways to Get Better Explanations from LLMsOn October 2, 2026, Andrej Karpathy published a post based on a very simple prediction: we will spend more and more time trying to…
Vikas Tomar·Aug 26The Karpathy Loop Scales. The Knowledge Doesn’t.A plain-English guide to the loop everyone copied in March, and the piece that is still missing.
Dreamwalker(박제창)·Apr 5What Is an LLM Wiki, and Why Are People Paying Attention to It?Not the opposite of RAG, but a pattern for maintaining knowledge with LLMs
Davide Gallitelli·Mar 30The Autoresearch Pattern — What Karpathy Got Right (and What’s Missing)Karpathy's 630-line loop hit 50k stars in a week. Here's what it gets right — and the 5 walls you'll hit in production.
InGoPenAIbySaikiran Bavandla·Mar 28I Let an AI Run 700 ML Experiments While I Slept. Here’s What It Found.Andrej Karpathy just released a 630-line tool that’s breaking the internet — and signaling the end of manual ML research as we know it.
Dreamwalker(박제창)·Feb 16Anatomy of microgptA start-to-finish, code-faithful dissection of a dependency-free GPT training script (microgpt)
Dreamwalker(박제창)·Feb 15microgpt : ‘순수 파이썬’으로 구현한 GPT-2형 디코더 트랜스포머의 알고리즘이 글의 목표는 Andrej Karpathy의 microgpt가 보여주는 “훈련(training) + 추론(inference) GPT”의 최소 알고리즘 구성요소를, 트랜스포머(Transformer) 블록 관점에서 개념 ↔ 수식 ↔ 코드로 정규화해…A response icon1A response icon1
