Skip to content

Conversation

@pelikhan
Copy link
Member

@pelikhan pelikhan commented Jul 9, 2025

This pull request configure AI-generated translations for french (only lessons and readme) using github actions and github models. Large files are still a bit of a problem, good start.

  • the github action triggers on main when markdown files are modified.
  • it performs as much translations as possible and pushes the result into main

@pelikhan pelikhan marked this pull request as ready for review July 12, 2025 13:09
Copy link
Member Author

@pelikhan pelikhan left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

First iteration added. Some of the prompts are reaching 16k tokens and hitting the upper limit. Chunking documents will certainly help with this.

@pelikhan pelikhan requested a review from sinedied July 12, 2025 13:16
Copy link
Contributor

@sinedied sinedied left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewed french translations and it's quite good!

Few questions though:

  • Is there a global config file that we can use to set global translations rules, like for example avoid translating specific words?
  • Is the cache shared between file, so an exact same paragraph in different files would be translated the same, for consistency? (and less token usage 🙂 )

pelikhan added 7 commits July 22, 2025 03:06
- Revised the French translation of Ada's example in the RAG lesson to improve clarity and accuracy.
- Updated the corresponding JSON translation file to reflect the changes made in the RAG lesson.
- Added usage statistics for the RAG lesson translation in the usage log.
@pelikhan
Copy link
Member Author

  • Is there a global config file that we can use to set global translations rules, like for example avoid translating specific words?

The tool supports an "instructions.md" file that gets injected into the prompt.

  • Is the cache shared between file, so an exact same paragraph in different files would be translated the same, for consistency? (and less token usage 🙂 )

The cache is repo wide so if two strings match, they will have the same translation. We could also percolate a translation corpus in the future and use it as a template.

@softchris softchris merged commit c1bc50f into main Oct 10, 2025
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants