Show HN: XML, Markdown, or JSON: Which gives LLMs the most reliable boundaries?(systima.ai)

3 ポイント投稿者 systima4 か月前2 コメント

Show HN: XML, Markdown, or JSON: Which gives LLMs the most reliable boundaries?

https://systima.ai/blog/delimiter-hypothesis

2 コメント

plsft4 か月前

I've used all three and I think markdown is the most effective. I build an AI toolkit that generates TOON for datasets and that works remarkably well too.

systima4 か月前

Respectfully, this is not really engaging withe content of the post.