Multilingual / Non-English Experience with Memori

[Boburmirzo]

Hi everyone 👋

We’d like to start a discussion around using Memori in non-English languages and better understand your real-world experience.

Context

Memori currently uses a sentence-transformer model that is optimized for English.
While it does support other languages, recall quality may not be as strong in some cases.

Before prioritizing improvements here, we’d like to learn directly from the community.

We’d love your input on:

• Which language(s) are you using Memori with?
• How would you rate memory recall accuracy in non-English use cases?
• Have you noticed issues such as:
  • Missing or incorrect recalls?
  • Semantic meaning loss?
  • Inconsistent behavior compared to English?
• Is multilingual support a blocking issue, a nice-to-have, or not a concern for you right now?

Feel free to share concrete examples if you can, even short ones help a lot.

Your feedback will directly influence whether and how we prioritize multilingual embedding support.
Thanks for helping shape Memori 🙏