4 ms·
If you're okay with losing the privacy of your notes. A locally hosted model seems more reasonable, though I don't know when it'll get good enough to have the c
by kajecounterhack 3y ago
If you're okay with losing the privacy of your notes. A locally hosted model seems more reasonable, though I don't know when it'll get good enough to have the context it needs to do proper search in a thicc stack of notes. Something something vector databases, I guess.
- thelittleone 3y agoSure it will. I've got similar setups working effectively with both OpenAI Api and Falcon. Though there is room for improvement, it's moving at a rapid enough pace and it's already better than old school search.
- patrickk 3y agoThere are (probably many) solutions being developed for this. Open source, self-hosted LLMs that can be used to query private data sets. You can query private data sets with an LLM hosted on your local machine. There are some projects online doing these kinds of projects with LangChain and using the leaked Meta LLM. It also ties into the "second brain" concept, being able to query your own data exclusively (there's a book on this, but you can get the gist of the concept on youtube). I myself have a huge OneNote notebook, with all kinds of interesting projects or snippets that I've collected from HN or Reddit over the years. Being able to query that like a super-smart friend ("what was the error I ran into when I tried to implemt a self-hosted Synapse instance? How I can I fix that?") would be insane for personal productivity.
- midoBB 3y agoI think Apple is working on on-device LLM training. Which can fix this for privacy if they are actually handling it on-device.