I have not yet launched an LLM assistant isolated from the Internet (and adequate) for learning the Kazakh language. Although the Llama 405B release hints that anything is possible (if you have your own GPU cluster😬). For a laptop, you can choose a lighter model and look at the 8B quality (as they say, subscribe and like).

To work, I decided to shorten T2M and connected a neuro assistant with integration into Google (Gemini models) and Open AI. It's nice that Google (unlike Open AI) can be connected via the API absolutely for free, without registration, but at least without SMS.

Counting the ROI:

— investments: 3.5 hours to study documentation, experiments and search for seals

— result: a streamlit service with integration into several LLMs via the API. Keremet!

🤔Is there any instruction needed to run this on your own or (for programmers) is everything clear?

#work #life