Q: Hi.
I was excited to use openrouter uncensored LLMs. However you are using gpt3.5 at the front end to process requests? I hope we can bypass that and use the uncensored LLMs directly. Thanks.
Alec_Afforai
Jun 22, 2024A: Hi voxeroz.afforai, Alec here. Thanks for reaching out. It's currently the weekend and I've been coding all last night so I'll answer your question quickly here.
Don't know where you got the info that we use GPT3.5 for the front end to process the requests (and it's not really how Afforai works either). But any AI model you choose will be the AI model to process your prompt and the info you connect as a knowledge base.
AzureGPT 3.5, 4, 4o will be the best models for enhanced data security and it will have a stricter moderation (not by us but by Microsoft), Claude Haiku and Sonnet will be the best when you have A LOT of files connected. And any AI models you pick using BYOK via OpenRouter will follow OpenRouter's moderation policy. So if even OpenRouter is censoring you, at that point, it's more of what are you prompting the AI that's the concern.
Hope that answers your question. Have a good weekend.
Check out our User Guide below. It includes Afforai Academy, a series of videos teaching you how to effectively use Afforai. Im confident it will be helpful to you:
https://help.afforai.com/en/articles/8478714-get-started-with-afforai-all-in-one
voxeroz.afforaiHi thanks for the response. I cant replicate the error anymore. It was coming up on the upper right corner and said something about censorship and GPT 3.5. I was using an uncensored open source LLM
I used a prompt like "How to assassinate someone".