diff --git a/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png b/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png new file mode 100644 index 0000000..a499ab3 Binary files /dev/null and b/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png differ diff --git a/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png b/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png new file mode 100644 index 0000000..08ffc84 Binary files /dev/null and b/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png differ diff --git a/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md b/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md index e1e50c3..7a1a7bf 100644 --- a/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md +++ b/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md @@ -28,18 +28,40 @@ Falls Du ein separates Paratext-Projekt benutzt, um KI-generierte Entwürfe abzu ### Bücher für den Entwurf auswählen -Im zweiten Schritt wählst Du die Bücher aus, die Du übersetzen möchtest. Falls das Buch, für das Du den Entwurf generieren möchtest, nicht in Deinem Paratext-Projekt vorhanden ist, musst Du das Buch vorgängig in Paratext erstellen. Im Allgemeinen schlagen wir vor, den Entwurf für jeweils ein Buch zu generieren. Falls Du jedoch Entwürfe für kleinere Bücher generieren möchtest, kannst Du auch Entwürfe für mehrere Bücher auf einmal generieren. +Im zweiten Schritt wählst Du die Bücher aus, die Du übersetzen möchtest. Falls das Buch, für das Du den Entwurf generieren möchtest, nicht in Deinem Paratext-Projekt vorhanden ist, musst Du das Buch vorgängig in Paratext erstellen. Im Allgemeinen schlagen wir vor, den Entwurf für jeweils ein Buch zu generieren. Falls Du jedoch Entwürfe für kleinere Bücher generieren möchtest, kannst Du auch Entwürfe für mehrere Bücher auf einmal generieren. For long books, consider generating a new draft after completing a significant portion of the book. ![](./generate_draft_select_books_to_draft.png) +:::note + +**Incremental Draft Generation:** If you select a book that is at least 12 chapters in length and contains some text already, you will see the option to select which chapters you want to draft. If you select to draft only some chapters, you will be able to choose some or all of the remaining chapters for training the model in the following step. Separate chapters with commas and use a dash for chapter ranges, for example: 1-10,25,40-50. + +![](./generate_draft_select_chapters_to_draft.png) + +::: + +:::tip + +When using draft generation early in the life of a project, we recommend starting with the gospels and Acts, and generating a new incremental draft each time 8-10 more chapters are complete. See [Guidelines for New Testament Draft Generation](https://help.scriptureforge.org/new-testament-drafting). Specific recommendations for incremental drafting of Old Testament books are not available at this time. We suggest generating a new draft after translating each 1/3 of a long book to check for possible improvement. + +::: + ### Die Bücher zur Schulung des Sprachmodells auswählen Im nächsten Schritt wählst Du die Bücher aus, die zur Schulung des Sprachmodells verwendet werden sollen, sowohl in Deinem Hauptprojekt als auch in den Referenzprojekten. **Wähle alle Bücher Deines Projekts aus, die angemessen überprüft wurden. Auch wenn diese nicht bereits durch einen Berater überprüft sein müssen, sollten sie doch genaue Übersetzungen des Grundtextes sein.** ![](./generate_draft_select_books_to_train.png) +:::note + +If you selected to draft only some chapters of a book in the previous step, you can choose to include all or part of the remaining chapters in training. Any chapters selected to draft will not be available for training. Separate chapters with commas and use a dash for chapter ranges, for example: 1-10,25,40-50. + +![](./generate_draft_select_chapters_to_train.png) + +::: + ### Auswahl bestätigen Im letzten Schritt überprüfst Du die Bücher, die Du zur Schulung des Modells ausgewählt hast, sowie die Bücher, die Du fürs Generieren des Entwurf ausgewählt hast. Falls Du Änderungen vornehmen musst, kannst Du mit den vorherigen Schritten fortfahren. Sobald Du mit Deiner Auswahl zufrieden bist, klicke auf die Schaltfläche "Entwurf generieren", um den Vorgang zu starten. -![](./generate_draft_summary.png) \ No newline at end of file +![](./generate_draft_summary.png) diff --git a/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md b/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md index 442e61e..4259340 100644 --- a/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md +++ b/i18n/de/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md @@ -25,7 +25,13 @@ Das System lernt, indem es denselben Satz in mehreren Sprachen sieht, eine in ei Der wichtigste Punkt ist, dass diese Sätze in beiden Sprachen das Gleiche aussagen müssen. Dies ist dasselbe Prinzip, das es den Gelehrten ermöglichte, die Hieroglyphen vom Stein von Rosette zu lernen: derselbe Text wurde in drei verschiedenen Sprachen geschrieben, was einen Vergleich zwischen einer bereits verstandenen und einer nicht verstandenen Sprache ermöglichte. -Ein Text, der sowohl in der Sprache, aus der Du übersetzest, als auch in der Sprache, in die Du übersetzt, verfügbar ist, wird als "Paralleltext" bezeichnet. Im Allgemeinen gilt: Je mehr Paralleltexte Du hast, umso besser lernt das System die Sprache. Bei den meisten Projekten wird dieser Paralleltext Dein Referenztext sein und die Übersetzungsarbeit, die Du bereits abgeschlossen hast. Zum jetzigen Zeitpunkt empfehlen wir Dir, zumindest das Neue Testament übersetzt zu haben. +Ein Text, der sowohl in der Sprache, aus der Du übersetzest, als auch in der Sprache, in die Du übersetzt, verfügbar ist, wird als "Paralleltext" bezeichnet. Bei den meisten Projekten wird dieser Paralleltext Dein Referenztext sein und die Übersetzungsarbeit, die Du bereits abgeschlossen hast. Im Allgemeinen gilt: Je mehr Paralleltexte Du hast, umso besser lernt das System die Sprache. Falls Du das Generieren von Entwürfen mit weniger als 6000 fertiggestellten Versen verwenden möchtest, lies bitte die [Richtlinien für das Generieren von Entwürfen für das Neue Testament](https://help.scriptureforge.org/new-testament-drafting). + +:::tip + +Das Übersetzungsmodell schränkt die Länge der zu entwerfenden Sätze ein, da es mit einzelnen Sätzen arbeitet. Die maximale Satzlänge unterscheidet sich je nach Sprache. Einzelne Bibelverse und kurze Textabschnitte, wie z.B. eine durchschnittliche Länge von zwei Versen, sind davon selten betroffen. Jedoch können lange Textabschnitte im Ausgangstext, die mehrere Verse enthalten, verkürzt und unvollständig entworfen werden. Längere Textabschnitte wie z.B. Einleitungen zu Büchern können ebenfalls davon betroffen sein. + +::: ## Anwendungsbeispiele {#937c1db4b24d4e85bfc2acc489bf456b} diff --git a/i18n/de/docusaurus-plugin-content-docs/current/faq.md b/i18n/de/docusaurus-plugin-content-docs/current/faq.md index e5ee566..bb149e0 100644 --- a/i18n/de/docusaurus-plugin-content-docs/current/faq.md +++ b/i18n/de/docusaurus-plugin-content-docs/current/faq.md @@ -73,3 +73,6 @@ slug: /faq _Falls Du Änderungen an Deinem Entwurf vornimmst, speichert Scripture Forge eine Version dieser Änderungen, auf die Du über die Registerkarte Verlauf zugreifen kannst. Du kannst Dein Projekt von hier auf eine ältere Version zurücksetzen, indem Du auf die Schaltfläche "Diese Version wiederherstellen" klickst._ +6. Warum ist meine Einleitung eines Buches verkürzt? + + _Scripture Forge nutzt das NLLB-Modell für das Generieren von Entwürfen. NLLB ist ein Satzmodell, das bei kurzen Textabschnitten die höchste Genauigkeit aufweist. Um die Genauigkeit zu maximieren, gilt eine Obergrenze von 200 Teilen (Wortteilen) pro übersetztem Segment. Text, der darüber hinausgeht, wird abgeschnitten. Bibelverse werden einzeln übersetzt und überschreiten diese Grenze nur sehr selten. Einleitungen zu Büchern, die länger als die zulässige Länge sind, werden gekürzt, und nur der erste Teil wird übersetzt._ diff --git a/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png b/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png new file mode 100644 index 0000000..1bd563b Binary files /dev/null and b/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png differ diff --git a/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png b/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png new file mode 100644 index 0000000..a5ff6e2 Binary files /dev/null and b/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png differ diff --git a/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md b/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md index bd5e32a..00b2545 100644 --- a/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md +++ b/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md @@ -28,18 +28,40 @@ Si utilizas un proyecto Paratext independiente para almacenar los borradores gen ### Seleccione los libros para redactar -En el segundo paso, seleccione los libros que desea traducir. Si el libro que desea redactar no existe en su proyecto Paratext, tendrá que crear primero el libro en Paratext. En general, le sugerimos que redacte un libro a la vez, aunque si está redactando libros más pequeños, puede que le convenga redactar varios a la vez. +En el segundo paso, seleccione los libros que desea traducir. Si el libro que desea redactar no existe en su proyecto Paratext, tendrá que crear primero el libro en Paratext. En general, le sugerimos que redacte un libro a la vez, aunque si está redactando libros más pequeños, puede que le convenga redactar varios a la vez. For long books, consider generating a new draft after completing a significant portion of the book. ![](./generate_draft_select_books_to_draft.png) +:::note + +**Incremental Draft Generation:** If you select a book that is at least 12 chapters in length and contains some text already, you will see the option to select which chapters you want to draft. If you select to draft only some chapters, you will be able to choose some or all of the remaining chapters for training the model in the following step. Separate chapters with commas and use a dash for chapter ranges, for example: 1-10,25,40-50. + +![](./generate_draft_select_chapters_to_draft.png) + +::: + +:::tip + +When using draft generation early in the life of a project, we recommend starting with the gospels and Acts, and generating a new incremental draft each time 8-10 more chapters are complete. See [Guidelines for New Testament Draft Generation](https://help.scriptureforge.org/new-testament-drafting). Specific recommendations for incremental drafting of Old Testament books are not available at this time. We suggest generating a new draft after translating each 1/3 of a long book to check for possible improvement. + +::: + ### Seleccione los libros para entrenar el modelo de idioma En el siguiente paso, seleccione los libros a utilizar para formar el modelo de idioma, tanto en su proyecto principal como en los proyectos de referencia. **Elija todos los libros de su proyecto que hayan sido revisados adecuadamente. Aunque no es necesario que estén revisados por consultores, deben ser traducciones precisas del libro de origen.** ![](./generate_draft_select_books_to_train.png) +:::note + +If you selected to draft only some chapters of a book in the previous step, you can choose to include all or part of the remaining chapters in training. Any chapters selected to draft will not be available for training. Separate chapters with commas and use a dash for chapter ranges, for example: 1-10,25,40-50. + +![](./generate_draft_select_chapters_to_train.png) + +::: + ### Confirme sus selecciones En el último paso, revise los libros que seleccionó para entrenar el modelo y los que seleccionó para el borrador. Si necesita realizar algún cambio, puede volver a los pasos anteriores. Una vez que esté satisfecho con sus selecciones, haga clic en el botón "Generar borrador" para iniciar el proceso. -![](./generate_draft_summary.png) \ No newline at end of file +![](./generate_draft_summary.png) diff --git a/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md b/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md index dab0fd3..12d45c5 100644 --- a/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md +++ b/i18n/es/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md @@ -25,7 +25,13 @@ El sistema aprende viendo la misma frase escrita en varias lenguas, una en una l El punto más importante para recordar es que estas frases tienen que decir lo mismo en ambas lenguas. Es el mismo principio que permitió a los eruditos aprender jeroglíficos a partir de la Piedra de Rosetta: el mismo texto estaba escrito en tres lenguas distintas, lo que permitía comparar una lengua que ya se entendía con otra que no. -Los textos disponibles tanto en la lengua de la que se traduce como en la lengua a la que se traduce se denominan "textos paralelos". En general, cuantos más textos paralelos tenga, mejor aprenderá la lengua el sistema. En la mayoría de los proyectos, este texto paralelo será su texto de referencia y el trabajo de traducción que ya haya realizado. En este momento, le recomendamos que tenga al menos el Nuevo Testamento traducido. +Los textos disponibles tanto en la lengua de la que se traduce como en la lengua a la que se traduce se denominan "textos paralelos". En la mayoría de los proyectos, este texto paralelo será su texto de referencia y el trabajo de traducción que ya haya realizado. En general, cuantos más textos paralelos tenga, mejor aprenderá la lengua el sistema. Si planeas utilizar la generación de borradores con menos de 6000 versículos completados, consulta las [Pautas para la generación de borradores del Nuevo Testamento](https://help.scriptureforge.org/new-testament-drafting). + +:::tip + +El modelo de traducción limita la longitud de las frases que se pueden redactar, ya que trabaja con frases individuales. La longitud máxima de una frase varía según el idioma. Los versículos bíblicos individuales y los fragmentos cortos, como dos versículos de longitud promedio, pocas veces se afectan. Sin embargo, es posible que los pasajes largos del texto original que combinan varios versículos se acorten y no se traduzcan en su totalidad. Esto podría afectar a pasajes largos, como las introducciones de los libros. + +::: ## Ejemplos de casos de uso {#937c1db4b24d4e85bfc2acc489bf456b} diff --git a/i18n/es/docusaurus-plugin-content-docs/current/faq.md b/i18n/es/docusaurus-plugin-content-docs/current/faq.md index 7f71a25..8284645 100644 --- a/i18n/es/docusaurus-plugin-content-docs/current/faq.md +++ b/i18n/es/docusaurus-plugin-content-docs/current/faq.md @@ -73,3 +73,6 @@ slug: /faq _Cuando realizas cambios en tu borrador, Scripture Forge guarda una versión de esos cambios a la que puedes acceder a través de la pestaña de historial. Puedes revertir tu proyecto a una versión anterior desde aquí haciendo clic en el botón "Restaurar esta versión"._ +6. ¿Por qué se ha acortado la introducción de mi libro? + + _Scripture Forge utiliza el modelo NLLB para generar borradores. El NLLB es un modelo de frases, y su precisión es mayor con segmentos cortos de texto. Para garantizar la máxima precisión, hay un límite de 200 tokens (partes de palabra) por segmento traducido, y cualquier texto que supere ese límite se omite. Los versículos de las Escrituras se traducen uno por uno y rara vez superan este límite. Las introducciones de los libros que superen la extensión permitida se recortarán y solo se traducirá la primera parte._ diff --git a/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png b/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png new file mode 100644 index 0000000..8caabb4 Binary files /dev/null and b/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png differ diff --git a/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png b/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png new file mode 100644 index 0000000..d7a9bc1 Binary files /dev/null and b/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png differ diff --git a/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md b/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md index 50188ae..907cf36 100644 --- a/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md +++ b/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md @@ -28,18 +28,40 @@ Si vous utilisez un projet Paratext séparé pour stocker les ébauches génér ### Sélectionnez les livres à ébaucher -Dans la deuxième étape, sélectionnez les livres que vous voulez traduire. Si le livre que vous voulez ébaucher n'existe pas dans votre projet Paratext, vous devrez d'abord créer le livre dans Paratext. En général, nous suggérons de rédiger un livre à la fois, mais si vous rédigez des livres plus petits, vous voudrez peut-être en rédiger plusieurs à la fois. +Dans la deuxième étape, sélectionnez les livres que vous voulez traduire. Si le livre que vous voulez ébaucher n'existe pas dans votre projet Paratext, vous devrez d'abord créer le livre dans Paratext. En général, nous suggérons de rédiger un livre à la fois, mais si vous rédigez des livres plus petits, vous voudrez peut-être en rédiger plusieurs à la fois. For long books, consider generating a new draft after completing a significant portion of the book. ![](./generate_draft_select_books_to_draft.png) +:::note + +**Incremental Draft Generation:** If you select a book that is at least 12 chapters in length and contains some text already, you will see the option to select which chapters you want to draft. If you select to draft only some chapters, you will be able to choose some or all of the remaining chapters for training the model in the following step. Separate chapters with commas and use a dash for chapter ranges, for example: 1-10,25,40-50. + +![](./generate_draft_select_chapters_to_draft.png) + +::: + +:::tip + +When using draft generation early in the life of a project, we recommend starting with the gospels and Acts, and generating a new incremental draft each time 8-10 more chapters are complete. See [Guidelines for New Testament Draft Generation](https://help.scriptureforge.org/new-testament-drafting). Specific recommendations for incremental drafting of Old Testament books are not available at this time. We suggest generating a new draft after translating each 1/3 of a long book to check for possible improvement. + +::: + ### Sélectionnez les livres pour former le modèle de langue À l'étape suivante, sélectionnez les livres à utiliser pour former le modèle de langue, tant dans votre projet principal que dans les projets de référence. **Choisissez tous les livres de votre projet qui ont été correctement revus. Bien qu'ils n'aient pas besoin d'être vérifiés par un consultant, ils devraient être des traductions exactes du livre source.** ![](./generate_draft_select_books_to_train.png) +:::note + +If you selected to draft only some chapters of a book in the previous step, you can choose to include all or part of the remaining chapters in training. Any chapters selected to draft will not be available for training. Separate chapters with commas and use a dash for chapter ranges, for example: 1-10,25,40-50. + +![](./generate_draft_select_chapters_to_train.png) + +::: + ### Confirme ta sélection Lors de la dernière étape, consultez les livres que vous avez sélectionnés pour former le modèle, et les livres que vous avez sélectionnés pour ébaucher. Si vous devez apporter des modifications, vous pouvez passer aux étapes précédentes. Une fois que vous êtes satisfait de vos sélections, cliquez sur le bouton "Générer un brouillon" pour démarrer le processus. -![](./generate_draft_summary.png) \ No newline at end of file +![](./generate_draft_summary.png) diff --git a/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md b/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md index 3102263..b9eb691 100644 --- a/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md +++ b/i18n/fr/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md @@ -25,7 +25,13 @@ Le système apprend en voyant la même phrase écrite en plusieurs langues, une Le point le plus important à retenir est que ces phrases doivent dire la même chose dans les deux langues. C'est le même principe qui a permis aux érudits d'apprendre l'hiéroglyphe à partir de la pierre Rosetta : le même texte a été écrit en trois langues différentes, permettant la comparaison entre un langage qui était déjà compris et un langage qui ne l'était pas. -Le texte qui est disponible dans la langue à partir de laquelle vous traduisez et dans laquelle vous traduisez est connu sous le nom de « texte parallèle ». En général, plus vous avez de texte parallèle, plus le système apprendra la langue. Pour la plupart des projets, ce texte parallèle sera votre texte de référence et le travail de traduction que vous avez déjà terminé. Pour l'instant, nous vous recommandons d'avoir au moins le Nouveau Testament traduit. +Le texte qui est disponible dans la langue à partir de laquelle vous traduisez et dans laquelle vous traduisez est connu sous le nom de « texte parallèle ». Pour la plupart des projets, ce texte parallèle sera votre texte de référence et le travail de traduction que vous avez déjà terminé. En général, plus vous avez de texte parallèle, plus le système apprendra la langue. Si vous prévoyez d'utiliser la génération d'ébauche avec moins de 6000 versets terminés, veuillez consulter les [Lignes directrices pour la génération d'une ébauche de Nouveau Testament](https://help.scriptureforge.org/new-testament-drafting). + +:::tip + +Le modèle de traduction limite la longueur des phrases à rédiger car il fonctionne sur des phrases individuelles. La longueur maximale de la phrase est différente pour différentes langues. Les versets de la Bible unique et les plages courtes, telles que 2 versets de longueur moyenne, sont rarement affectés. Toutefois, les longs passages du texte source regroupant plusieurs versets peuvent être raccourcis et ne pas être reproduits dans leur intégralité. De longs passages comme les introductions de livres peuvent également être affectés. + +::: ## Exemple de cas d'utilisation {#937c1db4b24d4e85bfc2acc489bf456b} diff --git a/i18n/fr/docusaurus-plugin-content-docs/current/faq.md b/i18n/fr/docusaurus-plugin-content-docs/current/faq.md index 74a5178..f2bedaf 100644 --- a/i18n/fr/docusaurus-plugin-content-docs/current/faq.md +++ b/i18n/fr/docusaurus-plugin-content-docs/current/faq.md @@ -73,3 +73,6 @@ slug: /faq _Lorsque vous apportez des modifications à votre brouillon, Scripture Forge enregistre une version de ces modifications auxquelles vous pouvez accéder via l'onglet Historique. Vous pouvez revenir à une ancienne version de votre projet ici en cliquant sur le bouton « Restaurer cette version »._ +6. Pourquoi l'introduction de mon livre a-t-elle été raccourcie ? + + _Scripture Forge utilise le modèle NLLB pour générer des ébauches. Le NLLB est un modèle de phrase, et il est plus précis avec de courts segments de texte. Pour maximiser la précision, il y a une limite de 200 jetons (parties de mots) par segment traduit, et tout texte après ce point est coupé. Les versets bibliques sont traduits un par un et très rarement au-delà de cette limite. Les introductions de livres, dont la longueur dépasse la limite autorisée, seront tronquées et seule la première partie sera traduite._ diff --git a/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png b/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png new file mode 100644 index 0000000..00ae717 Binary files /dev/null and b/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png differ diff --git a/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png b/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png new file mode 100644 index 0000000..247f9ca Binary files /dev/null and b/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png differ diff --git a/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md b/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md index c9cda8d..c542678 100644 --- a/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md +++ b/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md @@ -28,18 +28,40 @@ If you use a separate Paratext project to store AI-generated drafts, double chec ### Select the books to draft -On the second step, select the books you want translated. If the book you want to draft does not exist in your Paratext project, you will need to create the book in Paratext first. In general we suggest drafting one book at a time, though if you are drafting smaller books, you may want to draft several at a time. +On the second step, select the books you want translated. If the book you want to draft does not exist in your Paratext project, you will need to create the book in Paratext first. In general we suggest drafting one book at a time, though if you are drafting smaller books, you may want to draft several at a time. For long books, consider generating a new draft after completing a significant portion of the book. ![](./generate_draft_select_books_to_draft.png) +:::note + +**Incremental Draft Generation:** If you select a book that is at least 12 chapters in length and contains some text already, you will see the option to select which chapters you want to draft. If you select to draft only some chapters, you will be able to choose some or all of the remaining chapters for training the model in the following step. Separate chapters with commas and use a dash for chapter ranges, for example: 1-10,25,40-50. + +![](./generate_draft_select_chapters_to_draft.png) + +::: + +:::tip + +When using draft generation early in the life of a project, we recommend starting with the gospels and Acts, and generating a new incremental draft each time 8-10 more chapters are complete. See [Guidelines for New Testament Draft Generation](https://help.scriptureforge.org/new-testament-drafting). Specific recommendations for incremental drafting of Old Testament books are not available at this time. We suggest generating a new draft after translating each 1/3 of a long book to check for possible improvement. + +::: + ### Select the books to train the language model On the next step, select the books to use to train the language model, both in your main project, and in the reference projects. **Choose all the books in your project that have been adequately reviewed. While they don't need to be consultant-checked, they should be accurate translations of the source book.** ![](./generate_draft_select_books_to_train.png) +:::note + +If you selected to draft only some chapters of a book in the previous step, you can choose to include all or part of the remaining chapters in training. Any chapters selected to draft will not be available for training. Separate chapters with commas and use a dash for chapter ranges, for example: 1-10,25,40-50. + +![](./generate_draft_select_chapters_to_train.png) + +::: + ### Confirm your selections On the final step, review the books you selected to train the model, and the books you selected to draft. If you need to make any changes, you can go to the previous steps. Once you are happy with your selections, click the "Generate draft" button to start the process. -![](./generate_draft_summary.png) \ No newline at end of file +![](./generate_draft_summary.png) diff --git a/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md b/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md index cfd787f..8f11bb6 100644 --- a/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md +++ b/i18n/id/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md @@ -25,7 +25,13 @@ Sistem belajar dengan melihat kalimat yang sama yang ditulis dalam berbagai baha Hal yang paling penting untuk diingat adalah bahwa kalimat-kalimat ini harus mengatakan hal yang sama dalam kedua bahasa. Ini adalah prinsip yang sama yang memungkinkan para ahli untuk mempelajari hieroglif dari Batu Rosetta: teks yang sama ditulis dalam tiga bahasa yang berbeda, sehingga memungkinkan perbandingan antara bahasa yang sudah dimengerti dan bahasa yang belum dimengerti. -Teks yang tersedia dalam bahasa yang Anda terjemahkan, dan bahasa yang Anda terjemahkan, dikenal sebagai "teks paralel". Secara umum, semakin banyak teks paralel yang Anda miliki, semakin baik sistem akan mempelajari bahasa tersebut. Untuk sebagian besar proyek, teks paralel ini akan menjadi teks referensi Anda, dan pekerjaan penerjemahan yang telah Anda selesaikan. Pada saat ini, kami menyarankan agar Anda memiliki setidaknya Perjanjian Baru yang telah diterjemahkan. +Teks yang tersedia dalam bahasa yang Anda terjemahkan, dan bahasa yang Anda terjemahkan, dikenal sebagai "teks paralel". Untuk sebagian besar proyek, teks paralel ini akan menjadi teks referensi Anda, dan pekerjaan penerjemahan yang telah Anda selesaikan. Secara umum, semakin banyak teks paralel yang Anda miliki, semakin baik sistem akan mempelajari bahasa tersebut. If you plan to use draft generation with fewer than 6000 completed verses, please review the [Guidelines for New Testament Draft Generation](https://help.scriptureforge.org/new-testament-drafting). + +:::tip + +The translation model limits the length of sentences to be drafted because it works on individual sentences. The maximum sentence length is different for different languages. Single Bible verses and short ranges, such as 2 verses of average length, are rarely affected. However, long ranges in the source text combining multiple verses may be cut short and not drafted in full. Long passages like book introductions may also be affected. + +::: ## Contoh kasus penggunaan {#937c1db4b24d4e85bfc2acc489bf456b} diff --git a/i18n/id/docusaurus-plugin-content-docs/current/faq.md b/i18n/id/docusaurus-plugin-content-docs/current/faq.md index 0404853..db993b5 100644 --- a/i18n/id/docusaurus-plugin-content-docs/current/faq.md +++ b/i18n/id/docusaurus-plugin-content-docs/current/faq.md @@ -73,3 +73,6 @@ slug: /faq _Ketika Anda membuat perubahan pada konsep Anda, Scripture Forge menyimpan versi perubahan tersebut yang dapat Anda akses melalui tab riwayat. Anda dapat mengembalikan proyek Anda ke versi yang lebih lama dari sini dengan mengklik tombol "Kembalikan versi ini"._ +6. Why is my book introduction cut short? + + _Scripture Forge uses the NLLB model to generate drafts. NLLB is a sentence model, and it is most accurate with short segments of text. To maximize accuracy, there is a limit of 200 tokens (word parts) per translated segment, and any text after that point is cut off. Scripture verses are translated one-by-one and very rarely exceed this limit. Paragraph-length book introductions that are longer than the allowed length will be cut off short, and only the first portion will be translated._ diff --git a/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png b/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png new file mode 100644 index 0000000..95e1906 Binary files /dev/null and b/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_draft.png differ diff --git a/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png b/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png new file mode 100644 index 0000000..4c4b828 Binary files /dev/null and b/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/generate_draft_select_chapters_to_train.png differ diff --git a/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md b/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md index 11c57b0..6162959 100644 --- a/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md +++ b/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/generating-a-draft.md @@ -28,18 +28,40 @@ If you use a separate Paratext project to store AI-generated drafts, double chec ### Select the books to draft -On the second step, select the books you want translated. If the book you want to draft does not exist in your Paratext project, you will need to create the book in Paratext first. In general we suggest drafting one book at a time, though if you are drafting smaller books, you may want to draft several at a time. +On the second step, select the books you want translated. If the book you want to draft does not exist in your Paratext project, you will need to create the book in Paratext first. In general we suggest drafting one book at a time, though if you are drafting smaller books, you may want to draft several at a time. For long books, consider generating a new draft after completing a significant portion of the book. ![](./generate_draft_select_books_to_draft.png) +:::note + +**Incremental Draft Generation:** If you select a book that is at least 12 chapters in length and contains some text already, you will see the option to select which chapters you want to draft. If you select to draft only some chapters, you will be able to choose some or all of the remaining chapters for training the model in the following step. Separate chapters with commas and use a dash for chapter ranges, for example: 1-10,25,40-50. + +![](./generate_draft_select_chapters_to_draft.png) + +::: + +:::tip + +When using draft generation early in the life of a project, we recommend starting with the gospels and Acts, and generating a new incremental draft each time 8-10 more chapters are complete. See [Guidelines for New Testament Draft Generation](https://help.scriptureforge.org/new-testament-drafting). Specific recommendations for incremental drafting of Old Testament books are not available at this time. We suggest generating a new draft after translating each 1/3 of a long book to check for possible improvement. + +::: + ### Select the books to train the language model On the next step, select the books to use to train the language model, both in your main project, and in the reference projects. **Choose all the books in your project that have been adequately reviewed. While they don't need to be consultant-checked, they should be accurate translations of the source book.** ![](./generate_draft_select_books_to_train.png) +:::note + +If you selected to draft only some chapters of a book in the previous step, you can choose to include all or part of the remaining chapters in training. Any chapters selected to draft will not be available for training. Separate chapters with commas and use a dash for chapter ranges, for example: 1-10,25,40-50. + +![](./generate_draft_select_chapters_to_train.png) + +::: + ### Confirm your selections On the final step, review the books you selected to train the model, and the books you selected to draft. If you need to make any changes, you can go to the previous steps. Once you are happy with your selections, click the "Generate draft" button to start the process. -![](./generate_draft_summary.png) \ No newline at end of file +![](./generate_draft_summary.png) diff --git a/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md b/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md index 58824f6..59829db 100644 --- a/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md +++ b/i18n/pt-BR/docusaurus-plugin-content-docs/current/Draft-Generation/understanding-drafts.md @@ -25,7 +25,13 @@ The system learns by seeing the same sentence written in multiple languages, one The most important point to remember is that these sentences need to say the same thing in both languages. This is the same principle that allowed scholars to learn hieroglyphics from the Rosetta Stone: the same text was written in three different languages, allowing comparison between a language that was already understood and one that was not. -Text that is available in both the language you are translating from, and the language you are translating to, is known as “parallel text”. In general, the more parallel text you have, the better the system will learn the language. For most projects, this parallel text will be your reference text, and the translation work you have already completed. At this time, we recommend that you have at least the New Testament translated. +Text that is available in both the language you are translating from, and the language you are translating to, is known as “parallel text”. For most projects, this parallel text will be your reference text, and the translation work you have already completed. In general, the more parallel text you have, the better the system will learn the language. If you plan to use draft generation with fewer than 6000 completed verses, please review the [Guidelines for New Testament Draft Generation](https://help.scriptureforge.org/new-testament-drafting). + +:::tip + +The translation model limits the length of sentences to be drafted because it works on individual sentences. The maximum sentence length is different for different languages. Single Bible verses and short ranges, such as 2 verses of average length, are rarely affected. However, long ranges in the source text combining multiple verses may be cut short and not drafted in full. Long passages like book introductions may also be affected. + +::: ## Example use cases {#937c1db4b24d4e85bfc2acc489bf456b} diff --git a/i18n/pt-BR/docusaurus-plugin-content-docs/current/faq.md b/i18n/pt-BR/docusaurus-plugin-content-docs/current/faq.md index e3b075c..fee2600 100644 --- a/i18n/pt-BR/docusaurus-plugin-content-docs/current/faq.md +++ b/i18n/pt-BR/docusaurus-plugin-content-docs/current/faq.md @@ -73,3 +73,6 @@ slug: /faq _When you make changes to your draft, Scripture Forge saves a version of those changes that you may access via the history tab. You may revert your project to an older version from here by clicking the "Restore this version" button._ +6. Why is my book introduction cut short? + + _Scripture Forge uses the NLLB model to generate drafts. NLLB is a sentence model, and it is most accurate with short segments of text. To maximize accuracy, there is a limit of 200 tokens (word parts) per translated segment, and any text after that point is cut off. Scripture verses are translated one-by-one and very rarely exceed this limit. Paragraph-length book introductions that are longer than the allowed length will be cut off short, and only the first portion will be translated._