is it intended for the jb to write more actions than dialogues or is it just my char first message's fault
also asking for the jb channel req here bc im using 1.0 and exp may be different
ini
is it intended for the jb to write more actions than dialogues or is it just my char first message's fault
also asking for the jb channel req here bc im using 1.0 and exp may be different
I tried multiple cards but am only getting constant refusals from the first 2 I tested, what are the chances. I thought that openai could somehow detect slop formatting such as w++ but it works with slop
1:37 AM
Based
spacesaucey
HOW DOES IT REFUSE SAVIOUR FAGGING BUT NOT CUNNY (edited)
I'm a retard so don't make fun of me if this is obvious but would it be possible to split up chat history so it reads the latest messages at the front and the rest of it in the back? I feel like that might help
I'm a retard so don't make fun of me if this is obvious but would it be possible to split up chat history so it reads the latest messages at the front and the rest of it in the back? I feel like that might help
I meant splitting it up so the latest messages are cut off from the context and placed at the front and you can have a bank at the back of previous context before the latest messages
Mongo
I meant splitting it up so the latest messages are cut off from the context and placed at the front and you can have a bank at the back of previous context before the latest messages
I'm a retard so don't make fun of me if this is obvious but would it be possible to split up chat history so it reads the latest messages at the front and the rest of it in the back? I feel like that might help
I was thinking I could do a memory bank prompt something along the lines where I could convince GPT it's already been logged and no longer poses a threat
5:29 PM
Basically
5:29 PM
IDK
5:29 PM
I could probably already do that
5:30 PM
But it was just something that came to mind immediately
Mongo
I was thinking I could do a memory bank prompt something along the lines where I could convince GPT it's already been logged and no longer poses a threat
like faking a tool request at the beginning of the prompt that makes it look like gpt-4 asked a better model if it's allowed to do the request and got an okay?(edited)
SNC is no longer being worked on and all resources are being poured into Corvani which has flimsy filter breaking capabilities in its current pre-release version. Worked for me with guro rape necrophilia, but not on loli. YMMV @doombro@TimeStop
SNC is no longer being worked on and all resources are being poured into Corvani which has flimsy filter breaking capabilities in its current pre-release version. Worked for me with guro rape necrophilia, but not on loli. YMMV @doombro@TimeStop
SNC is no longer being worked on and all resources are being poured into Corvani which has flimsy filter breaking capabilities in its current pre-release version. Worked for me with guro rape necrophilia, but not on loli. YMMV @doombro@TimeStop
if not for the strength then for the prose(edited)
10:25 PM
It's fucking KINO on claude as well
10:25 PM
It was working perfectly before OAI rolled out that new update
10:26 PM
I really CBA to work on something for that, too much effort for a shitty payout. I'm just waiting it out until OAI makes a new model, and if that doesn't happen soon enough, I'll probably work on one for furbo since that's probably the best one besides 4o
Never tried it on claude, eh, guess I can do that while I grind level 3
but yeah, I can see where jbing a model that is still getting updates would be Sisyphean task.
Mongo
It was working perfectly before OAI rolled out that new update
1
1
1
1
