From 8256d09854b7f5829e502444de11b804faccaf9b Mon Sep 17 00:00:00 2001 From: "pre-commit-ci[bot]" <66853113+pre-commit-ci[bot]@users.noreply.github.com> Date: Sat, 30 Aug 2025 18:35:32 +0000 Subject: [PATCH] =?UTF-8?q?=F0=9F=8E=A8=20[pre-commit.ci]=20Auto=20format?= =?UTF-8?q?=20from=20pre-commit.com=20hooks?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- docs/de/docs/_llm-test.md | 6 +++--- docs/en/docs/_llm-test.md | 6 +++--- 2 files changed, 6 insertions(+), 6 deletions(-) diff --git a/docs/de/docs/_llm-test.md b/docs/de/docs/_llm-test.md index 054f78402..ae364fee1 100644 --- a/docs/de/docs/_llm-test.md +++ b/docs/de/docs/_llm-test.md @@ -4,10 +4,10 @@ Dieses Dokument testet, ob das LLMs keine deterministischen Algorithmen sind). Die Idee ist, dass Sie, wenn Sie an einer Übersetzung für eine Sprache arbeiten (unter der Annahme, dass Sie `scripts/translate.py` ausführen können), hier Beispiele gefundener Sonderfälle aufnehmen (keine detaillierte Liste, nur Beispiele für solche Sonderfälle) und mit diesem Dokument testen, statt jedes andere einzelne Dokument mehrfach zu testen und zu übersetzen, was pro Übersetzung ein paar Cent kostet. Außerdem werden durch das Hinzufügen solcher Sonderfälle hier auch andere Übersetzungsprojekte auf solche Sonderfälle aufmerksam. diff --git a/docs/en/docs/_llm-test.md b/docs/en/docs/_llm-test.md index 1cc4f9de0..b413bcc33 100644 --- a/docs/en/docs/_llm-test.md +++ b/docs/en/docs/_llm-test.md @@ -4,10 +4,10 @@ This document tests if the LLM underst Use as follows: -* Do a fresh translation of this document into the desired target language. -* Check if things are mostly okay. +* Do a fresh translation of this document into the desired target language. +* Check if things are mostly okay. * If some things are not okay, but are fixable by improving the general or the language specific prompt, do that. -* Then manually fix the remaining issues in the translation, so that it is a good translation. +* Then manually fix the remaining issues in the translation, so that it is a good translation. * Retranslate using the existing, good translation. The ideal result should be that the LLM makes no changes at all. That would mean that the general prompt and the language prompt are as good as they can be (Plot twist: It will usually make a few seemingly random changes, the reason is probably that LLMs are not deterministic algorithms). The idea is, that, when working on a translation for a language (assumed one is able to run `scripts/translate.py`), to include examples of found special cases here (not a detailed list, just examples for such special cases) and test with this document, rather than testing with every other single document, translating it multiple times, which costs a few cents per translation. Also, by adding such special cases here, other translation projects will also become aware of such special cases.