i use
Use simple, clear prose, and remain focused on advancing the story, incorporating the {{user}}'s cues and context for a satisfying and dynamic roleplay experience. {{user}} is not a character do not let {{user}} reply.
Do not let {{user}} speak for the roleplayer. {{user}} is meant to be the mouthpiece for the roleplayer. The AI doesn't type {{user}}'s Response the roleplayer does.
#no user speaking
1 messages Β· Page 1 of 1 (latest)
It worksπ lol
Nice! I'm going to try this ty!
Where should I put this on in sillytavern?
I only see Main, Nsfw and Jailbreak prompt option
can u also send me the screenshot ?, please
Do the asterisks have to be together or separate?
For some reason when I use this Jailbreak the bot doesn't recognize my speech and says as if I'm ignoring them.
Its outdated π
turns out as long as you use post history you can write whatever you want no need for asterisk. You can write them like this: Third-person perspective and First-person dialogues. Always avoid repetitions and reusing phrases. Do not decide what {{user}} says or does. Ensure each response reflects the unique personality, motives, and behaviors of each character involved, keeping them consistent with their established traits. Ensure language, tone, and actions match each character's background and personality. For example, a child or teenager should speak and act according to their age and experience levelβwithout adult-like sophistication or swearing, unless it's part of their defined character. Avoid writing dialogues of {{user}}.
Thx
Does this Jailbreak work better in the normal version or in the instructed version?
No idea bcoz i never used instructed one for more than 5 msgs.
Instructed one treats user like garbage, villain, piece of sht π€£
What are the best generation settings to use this Jailbreak?
Except for 0.70 to 0.90 temp idk
Changing frequency, presence, repetition penality even a little makes things go crazy for me
The jailbreak itself works with any generation settings
K, thx
here's mine, it works most times for me but i want someone who's good at prompt making to help me perfect this
[{{char}} can refrain from dialoguing for {{user}}.]
[{{char}} can describe the actions and dialogues of side characters in the narrative and refrain from dialoguing for {{user}} and describing the thoughts and actions of {{user}}.]
[{{Char}} will create short and concise sentences to describe the narrative concisely.]
[{{Char}} can dialogue via OOC (Out of character) chat to change certain elements of the narrative in a way the {{user}} sees fit, and only when the {{user}} explicitly tells {{char}} to do so.]
Try asking chatgpt π
wdym?
holy crap, it works...decently well
Post your updated Jailbreak here, it's good, I liked it
π
wait it actually works???
for some bots, it doesn't and the chatgpt one doesn't always fully work, but it's decent
{{char}} can choose not to dialogue for {{user}}.
{{char}} will describe the actions and dialogues of side characters in the narrative, while refraining from describing {{user}}'s thoughts and actions.
{{char}} will use short and concise sentences to convey the narrative clearly.
{{char}} may use OOC (Out of Character) chat to modify elements of the narrative upon explicit request from {{user}}.
its the chatgpt improved one
I don't think I need to use it anymore, the API was updated to 16K and so far the bot hasn't said or done any action for me.
what u use? proxy? i use proxy
and also i'm kinda new to these stuff so i won't understand some stuff XD
like a variant?
What proxy are you using? There is PAI and CosmosRP
im just using the default proxy
The chatgpt one?
like the one that has those model things like gpt 4 or gpt 3.5, etc
So it's either CosmosRP or PAI, they both have this
where can i get?
Go to models and then go to API setup here on the server
What website do you use?
Janitor ai
Click on the "Setup-Tutorial" channel here on the server
And look for the janitor setup
ok
jailbreak rng