Its basicly a marketing strategy at this point lol.
But fb maintaines react too iirc
True, and this is the reason its reputation is, not to use harder word, questionable. Although widespred adoption.
@brivacy no, I have not checked it.
Alpaca which is ollama flatpak app with gtk4 libadwaita frontend.
Better to use origin ollama
Ollama does not have a gui that was provided by another software that you need to use where as it is already there. In that alpaca.
DuckDuckGo Chat just added open source models: Llama 3 70B and Mixtral 8x7B
4 Likes
Great news!
Open-source, but not hosted by DuckDuckGo. Hosted by together.ai.
See DuckDuckGo AI Chat Privacy Policy and Terms of Use
Also, they weakened the Privacy Policy. Now chats are deleted after 30 days, except for “safety and legal compliance”.
In addition, we have agreements in place with all model providers that further limit how they can use data from these anonymous chats, including the requirement that they delete all information received once it is no longer necessary to provide responses (at most within 30 days with limited exceptions for safety and legal compliance).
Should I stop using ChatGPT?
Yeah, I did see that. I had hoped they were hosted by DDG. However skimming together.ai’s privacy policy, gave me the impression its above average for a mainstream company (possibly due to a b2b focus). Specifically:
We do not use any data collected from you to train our models without your explicit opt-in and consent. You may revoke that consent at any time and request deletion of information we have collected. You may also disable some data collection in the settings for your product. We allow you to tweak your privacy settings in our platform so that we do not retain training data, prompts or model responses for any purpose. You can tweak your privacy settings by going to Settings > Profile.
Are you asking whether you should stop using ChatGPT, or you should stop using Duckduckgo’s ChatGPT ‘frontend’?
1 Like
Yes if I should stop and delete my OpenAI account…
That is a largely personal question, that you must answer for yourself. Personally, I do not trust OpenAI and I prefer not to rely on services that I do not trust.
If you do continue to use the service, read the privacy policy with a cautious eye to understand what you are giving up (privacy and potentially other things) and consider the conversations you have as semi-public. If you understand the privacy risks and tradeoffs in choosing and using a search engine, I think that mostly applies to LLMs as well. Technically speaking, you can’t control what happens to your data/queries once they leave your device, so trust in the provider is extra important.
3 Likes
What do you use instead of ChatGPT? or what would you use?
-
If you have the hardware to support it (or even if you don’t but have above average patience) running an LLM locally on your system is the most private way you can interact with an LLM (or self-hosting locally or on a rented server). In terms of hardware, what matters most is VRAM and a GPU (preferably Nvidia). With a lower-mid-range GPU like a RTX 3060 for example, you can locally run an LLM roughly equivalent to GPT 3.5, with higher end hardware you can run higher end models (VRAM and memory bandwidth are the two most important considerations).
-
If not running locally or on rented hardware, then you must understand you are placing trust in some 3rd party. Personally I have some pre-existing trust in Duckduckgo and Brave and I already place trust in them by using their search engines, I see them as the least worst options currently available for interacting with remotely hosted LLM’s. So if you want a hosted option, Duckduckgo Chat and Brave Leo are worth considering (both work with various models)
-
There are a couple other’s that I use sometimes–not because they are private– but because they are useful to me and because I can use them (1) with a VPN, and (2) without an account. I do not use them for sensitive things. I won’t publicize here because as mentioned they aren’t privacy centric so I’d prefer not to promote them, but feel free to PM me if you like.
Yes, I’ve indeed also been considering DuckDuckGo & Brave.
But knowing this, I will upgrade my PC to run an LLM locally at some point. Thanks for the info!
1 Like
Even with low end hardware you can start playing around with local LLMs. I don’t have a GPU at all, and I can run tiny models at acceptable speeds, or small models at somewhat tedious but still usable speeds.
Well, good to know. I will delete my OpenAI account today. 
1 Like
I emailed DuckDuckGo for clarification. Because when you use Duck, you are under their terms, so for any inclusion of DuckDuckGO Chat we will need to make sure of the exact terms.
BTW I changed the PR and made stricter criterias. Also added Ollama. here is a preview
2 Likes
It’s a good way to test any type of gen-ai before investing in a GPU.
If you actually want to try and integrate LLMs into your workflow, there is no way around using hardware acceleration.
1 Like