Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

Ask HN: Create embeddings efficiently for an AI notes app with E2EE

Apr 25, 2024 - news.ycombinator.com
The author is developing a notes application, thinkdeli.com, that automatically finds related notes by creating embeddings and adding them to an index. However, the process of setting up the pipeline for creating these embeddings locally uses a lot of memory (400+ MB on Chrome), making it impractical for older devices. The author is seeking a more efficient method to create these embeddings locally.

The author considered creating embeddings via an API, which would be more efficient, but this would require sending users' unencrypted notes over the cloud to the service. This contradicts the author's plan to implement end-to-end encryption, as the notes would need to be unencrypted and readable as plain text on the server to create the embeddings. The author is seeking advice on how to resolve this issue.

Key takeaways:

  • The author is developing a notes application, thinkdeli.com, that finds related notes automatically.
  • The application creates embeddings of the user's notes locally using transformers.js, but this process uses a lot of memory, making it impractical for older devices.
  • The author is considering implementing end-to-end encryption, but this presents a challenge as creating embeddings via an API would require sending unencrypted notes over the cloud.
  • The author is seeking a more efficient way to create embeddings locally without compromising the security of the user's notes.
View Full Article

Comments (0)

Be the first to comment!