When prompting an AI LLM model for a suggested translation in a low-resource language, the LLM needs similar content to be able to make an accurate translation. To supply the needed content, we need to pick similar text from the body of already translated verses.
This will take some experimentation and iteration to get right. Explore various techniques like using Postgres's full text search capabilities or grouping bible text on known similar tones, etc.
The goal is to generate a prompt for AI to get a consistent and acceptably accurate translation.
When prompting an AI LLM model for a suggested translation in a low-resource language, the LLM needs similar content to be able to make an accurate translation. To supply the needed content, we need to pick similar text from the body of already translated verses.
This will take some experimentation and iteration to get right. Explore various techniques like using Postgres's full text search capabilities or grouping bible text on known similar tones, etc.
The goal is to generate a prompt for AI to get a consistent and acceptably accurate translation.