this post was submitted on 23 Oct 2023
32 points (97.1% liked)
Asklemmy
43741 readers
2316 users here now
A loosely moderated place to ask open-ended questions
Search asklemmy ๐
If your post meets the following criteria, it's welcome here!
- Open-ended question
- Not offensive: at this point, we do not have the bandwidth to moderate overtly political discussions. Assume best intent and be excellent to each other.
- Not regarding using or support for Lemmy: context, see the list of support communities and tools for finding communities below
- Not ad nauseam inducing: please make sure it is a question that would be new to most members
- An actual topic of discussion
Looking for support?
Looking for a community?
- Lemmyverse: community search
- sub.rehab: maps old subreddits to fediverse options, marks official as such
- !lemmy411@lemmy.ca: a community for finding communities
~Icon~ ~by~ ~@Double_A@discuss.tchncs.de~
founded 5 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
I am not an expert here, so I cannot help you extensively. I just like to play around with projects. There's already tutorials out there for running a private LLM, so you might want to refer that.
I've tried running Alpaca 7B on my laptop using Dalai, and let's just say it toasted my device (i5 8265U, 8GB RAM, 25W MX250 2GB VRAM). I wouldn't recommend Dalai as the front-end, the Node script is a mess and it also requires privilege escalation to run commands - there are probably some other better front-ends out there.
Persimmon 8B is a slightly better model. Then there's also Mistral 7B, but I don't know much about that.
You might want to run these models using Google Cloud's Vertex AI.
Can you check out quickly and comment on SideMind (iOS)?
Edit: hope this doesnt come across as too "do shit for me Spidey" :/
I've checked the page you were talking about, and it shows how they have a bunch of different models, which is but an illusion of choice. I think what they're basically doing over here is that they're still using ChatGPT's API, but they're probably introducing some biases for the way these models talks, using specialized prompts that probably execute before you do the chatting (I think this is possible, but as I've never used their API, I cannot confirm this). So you're chatting with the same bot, that has it's mannerisms tweaked.
Putting that aside, and if I'm assuming that they're just like the rest of the other "AI" companies out there, they're still using OpenAI's service. The issue with this is that now you're exposed to a middleman that also has hands over your personal chat, apart from OpenAI. If you want a private LLM, you must run it on a personal instance by yourself.
A better option would be to look around on GitHub for someone's project, make sure that their front-end has web-app support, pin the web-app onto your iPhone's desktop I would urge you to check a few such projects on GitHub. Like for example, privateGPT, GPT4ALL (recommended) and Dalai (not recommended), which are completely open-source.
Edit: Yup, they're most probably using OpenAI's API, read their privacy policy here.
I coulda swore they had the No Data Collected label?