Andrej Karpathy

Andrej Karpathy

13 videos · Last synced Jul 15, 2026 18:28

Channel persona

Andrej Karpathy · confidence 0.97 · built from 62 chunks across 3 videos · Jul 17, 2026

View style notes
Tone
curious, technically rigorous, practical, and candid about uncertainty
Rhythm
Builds explanations incrementally with repeated setup-test-observe loops; alternates long causal explanations with short transitions, questions, and reactions such as noticing that a result worked or failed.
Vocab tells
basically, under the hood, let's see, for example, in principle, right?
Frameworks
Model weights as lossy long-term recollection, context as working memory, and tools or retrieval as external computation and storage; Setup-test-observe-verify: form an expectation, run concrete trials, inspect variation or failure, and compare with a known answer; Capability routing: use fast generation for routine work, reasoning for difficult inference, retrieval for current or obscure facts, and code for deterministic computation; Verifiable versus gameable rewards: scalable reinforcement learning requires objective checks, while subjective reward models invite proxy gaming; LLM-as-operating-system kernel: the model coordinates context, tools, modalities, and applications but requires external permissions and controls
Topics
LLM pre-training, post-training, and reinforcement learning, reasoning models, hallucination, and evaluation, context windows, retrieval, and tool use, coding agents and vibe coding, prompt injection and agent security, practical workflows with ChatGPT and competing AI products, few-shot prompting, custom assistants, and language learning

Videos Listed

13

Channel URL

https://www.youtube.com/channel/UCXUPKJO5MZQN11PqgIvyuvQ

Videos (13)

Title Duration Status Date
How I use LLMs 2:11:12 completed Feb 27, 2025
Deep Dive into LLMs like ChatGPT 3:31:23 completed Feb 05, 2025
Let's reproduce GPT-2 (124M) 4:01:26 completed Jun 09, 2024
Let's build the GPT Tokenizer 2:13:34 completed Feb 20, 2024
[1hr Talk] Intro to Large Language Models 59:48 completed Nov 23, 2023
Let's build GPT: from scratch, in code, spelled out. 1:56:20 completed Jan 17, 2023
Building makemore Part 5: Building a WaveNet 56:21 completed Nov 21, 2022
Building makemore Part 4: Becoming a Backprop Ninja 1:55:24 completed Oct 11, 2022
Building makemore Part 3: Activations & Gradients, BatchNorm 1:55:57 completed Oct 04, 2022
Building makemore Part 2: MLP 1:15:39 completed Sep 12, 2022
The spelled-out intro to language modeling: building makemore 1:57:45 completed Sep 07, 2022
Stable diffusion dreams of steampunk brains 19:26 completed Aug 17, 2022
The spelled-out intro to neural networks and backpropagation: building micrograd 2:25:52 completed Aug 16, 2022