Skip to main content

Best Practices

Best Practices for Markdown Preprocessing​

  1. Use Headings (H1-H6) to Split Content:

    • Break down your content into smaller sections using headings. This allows the system to generate embeddings for each part separately, enhancing retrieval and performance.

  2. Keep Paragraphs Short:

    • Short, concise paragraphs help reduce token usage and allow for more efficient processing of your data.

  3. Utilize Bullet Points and Lists:

    • Where possible, use bullet points to structure content, as they are processed more efficiently than long blocks of text.

  4. Remove Unnecessary Content:

    • Only include relevant information in your markdown files to minimize token count and focus embeddings on critical content.

By following these guidelines, you can ensure that your markdown files are optimized for use in VAKStudio's Vector Database, leading to more efficient and cost-effective embedding generation.