How LLMs Work
Interactive visual guide based on Karpathy's lecture
Details
- External ID
- 47886517
- Source
- HN
- Company
- —
- Product
- How LLMs Work
- Website domain
- github.io
- Launched
- April 24, 2026
- Cohort
- —
- Upvotes
- 245
- Upvotes percentile
- 0.9710796915167095
- Tags
- —
- Fetched at
- Sept. 7, 2026, 9:26 p.m.
- Updated at
- Sept. 7, 2026, 9:26 p.m.
Description
All content is based on Andrej Karpathy's "Intro to Large Language Models" lecture (youtube.com/watch?v=7xTGNNLPyMI). I downloaded the transcript and used Claude Code to generate the entire interactive site from it — single HTML file. I find it useful to revisit this content time to time.
Enrichment
- Theme
- code-driven AI video and animation
- Vertical
- Education
- Function
- Content generation
- Audience
- B2C
- AI stance
- Not AI
- Project type
- Hobby / open-source project
- Normalized one-liner
- interactive guide to how llms work
- Manually corrected
- False
Could you build this?
Yes The creator noted it was generated as a single HTML file from a transcript using Claude Code; it is static educational content with interactive diagrams.
Discussion
20 comments analyzed.
Competitors mentioned: Jay Alammar's Illustrated GPT-2 guide, Jay Alammar's free course on how LLMs work, tiktokenizer (for tokenization visualization)
Concerns raised: LLM-generated content lacks original value, BPE visualization misleadingly shows tokens being replaced rather than additively created, Tokenizer example performs poorly on made-up words, Skips attention mechanism entirely, Hallucinations in LLM-generated content
Feature requests: Move tokenization to end, start with byte-level model input/output, Explain attention mechanism in depth, Improve BPE visualization accuracy, Better top navigation bar (v2 addressed this)
Competitors
Other products that read as similar to this one — 75 launches clear the similarity bar, closest 8 shown.
Attention rank: #8 of 76 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).
Launched 169 days after the earliest competitor.
- I used Claude Code to discover connections between 100 books · hn · 2026-01-10 · 524 upvotes · similarity 0.46
- Mcptube · hn · 2026-04-13 · 13 upvotes · similarity 0.45
- Academa · hn · 2026-08-30 · 30 upvotes · similarity 0.41
- LLM Wiki Compiler Inspired by Karpathy · hn · 2026-04-06 · 7 upvotes · similarity 0.41
- Data Engineering Book · hn · 2026-02-13 · 251 upvotes · similarity 0.41
- I built a tiny LLM to demystify how language models work · hn · 2026-04-06 · 915 upvotes · similarity 0.41
- Open Computer-Animated Multivariable Calculus Course in 6 Languages · hn · 2026-01-25 · 5 upvotes · similarity 0.41
- Learn financial modelling by modelling · ph · 2026-09-10 · 1 upvotes · similarity 0.40
Other launches for this product
- No other launches for this product.
Same idea, different domain
Nobody's really built a content generation tool for Government yet.