endrift 🏳️⚧️LLM translation models are going great
I think people really underestimate how fragile LLMs are for auto-translation. You can put complete garbage into it where none of the words are real words and still get out plausible-sounding "translations" just because the LLM sees it as "close" to a real sentence, and then translates what it thinks that "close" sentence is based, once again, on what seems "close".
The whole benchmarking approach really does not help with this since benchmarks rarely include testing for failures. You need to test that garbage-in is recognized as garbage, otherwise you get garbage-out too.
rin@endrift @joe that reminds me of this recent example i encountered where i came across a quote in someone's bio and wanted to know what it meant. google translate confidently decided it was italian and gave this response. i dont know italian but was pretty sure this looked nothing like italian so ended up looking harder and found that its a quote of a made up language from a song in a game.
if u turn off gemini then it more sensibly determines that it cant find a translation, but its on now by default
