Hacker News Reader

Comment by s1monb | Hacker News Reader

I will do better and link to the research and related sources in the next iteration.

I was just pointing out how the article is clearly LLM written, probably including the interactive widgets. It's especially obvious because someone writing such an article in 2026 would at least find what the newest tokenizers are, instead of mentioning LLaMA 2/3 (!), and GPT's old tokenizer that they dropped since GPT-4o (or something close).

And, more obviously, the fact that GPT-4 is being directly named even though that model is over 3 years old by now: "Ask GPT-4, Claude, or Gemini today and they will usually answer three.".

Sorry, I just think that the article wasn't produced by a human at all.

> It's especially obvious because someone writing such an article in 2026 would at least find what what the newest tokenizers are

The underlying BPE algorithm, which is the main focus of this article, is the one used modern tokenizers today.

> The fact that GPT-4 is being directly named even though that model is over 3 years old by now

That is fair. Will be updated

> Sorry, I just think that the article wasn't produced by a human at all.

While I have used LLM to help me write and explain my content, my hopes is that most readers does not share this opinion of yours. Everything touched by AI is not slop, and I wanted to share the notes I created for myself.

Reply on news.ycombinator.com