Ask HN: Why do AI code editors suck at closing tags?

https://news.ycombinator.com/item?id=46629151

by cryptography • 6 months ago

9 5 6 months ago

news.ycombinator.com

cyanydeez• 6 months ago

Context favors the newst tokens. To get closing tages means to find much older tags.

HTML is simply oppositive with LLMs favoring the next token based on the just made.

nextaccountic• 6 months ago

There must be some interface for LLMs tocdeal directly with the tree structure of programming and computer languages. Like, something similar to Emacs paredit interface, but for arbitrary languages

geldedus• 6 months ago

They don't. Use better tools

maouida• 6 months ago

Never encountered this issue until I tested GLM and Kimi K2 models. Opus and Sonnet seems to be good at handling this, even when giving them a code with open tags, they'll know where to put the closing tag quickly.

varshith17• 6 months ago

Same reason they can't count the 'r's in "strawberry", these models don't actually understand structure, they just autocomplete really convincingly.

Ask HN: Why do AI code editors suck at closing tags?

Related Stories