Karpathy Explains LLMs Thoroughly in Long Video
kalyan_kpl · x · 2026-07-18
Andrej Karpathy's video tutorial systematically explains how LLM/ChatGPT works, covering the entire pipeline from pre-training to post-training.\n\nThe content includes:\n- Pre-training data, tokenization, neural network inputs/outputs, and internal structures\n- Inference, the difference between training and inference, and Llama 3.1 base model inference\n- Post-training data, hallucinations, tool usage, knowledge/working memory, and self-awareness\n- "Models need tokens to think", token look-back, spelling difficulties, and jagged intelligence\n- Supervised fine-tuning, reinforcement learning, DeepSeek-R1, AlphaGo, and RLHF\n\nHe also adds tips on how to continuously keep up with LLMs and where to find related model resources.
More from Models
- Google says Gemini 4 has entered its most ambitious pre-training run yet — himanshustwts · 2026-07-22
- China’s AI arms race is increasingly defined by chips, data centers, and open models — BenBajarin · 2026-07-22
- Sam Altman is headed to Washington to brief Congress on OpenAI’s GPT-6 line — inductionheads · 2026-07-22
- Benchmark chart pits GPT-5.6 Luna, Grok 4.5 and Gemini 3.6 Flash on price and scores — iruletheworldmo · 2026-07-22
- Claim says Kimi was distilled from Fable, sparking a model-attribution jab — cephaloform · 2026-07-22
- Gemini 3.6 Flash is now available in Antigravity and chat — MartianOnJupiter · 2026-07-22