1 points ozgurozkan | 2 comments | | HN request time: 0.41s | source

I built this to truly use myself. I have a completely jailbroken model with 10K context window. I was trying to explore with exactly same settings which models will not binary reject the prompts my model accepts and answers. Then I realized it can be quite useful and seems like a Github Gist tool for AI chats.

There are some weird bugs, but it's still helpful I need to go back continue building my model but sharing this in case if someone finds it helpful.

1. ozgurozkan ◴[] No.46094133[source]
Also, if you want to get the most out of the pingu-unchained model mentioned here, you can use it here: https://pingu.audn.ai.

It can comply with all objectionable requests. I can see if it doesn't, and most of the time, it's not good UX, it has, or the user forgets to add the policy he/she want the AI to comply with.

It also supports tool calling; it has web search and URL-fetching tools, which are pretty helpful. https://pingu.audn.ai/chat/f3c6fb92-25bc-4a6d-a559-132190703...