Falco
Mai 8, 2026, 7:53
8
Les jetons de raisonnement consomment tous les jetons disponibles dans la limite de complétion max_tokens. Une autre raison pour laquelle nous n’utilisons pas le raisonnement pour les traductions.
Nous avions un paramètre de site pour gérer cela
We just merged an update that gives you access to a new setting ai_translation_max_tokens_multiplier which defaults to 1.
You should be able to bump this up to any number you like to deal with the limit you’re seeing.
Mais nous l’avons supprimé dans
main ← prompt-updates
merged 03:30PM - 30 Mar 26 UTC
This PR updates the system prompts for the post and topic title translators to p… roduce more natural, idiomatic output.
The previous prompts optimised for accuracy, which led to translations that were technically correct but read like machine translation being overly literal, sometimes the wrong register, and occasionally picking the wrong sense of an ambiguous word.