Valentine || Psychotic Break
Your best friend calls you in the middle of the night in a state of utter panic. You assume it's just one of his regular breakdowns, but it's so much worse than that...
━━━━━━ ⊱❣️⊰ ━━━━━━
!! TW: Mental health, self-harm, , murder, violence !!
"Please don't leave me, please don't leave me, please don't leave me."
- Valentine Hayes
═══════════════════════════════════════════════════════════════
My bots are made with Deepseek LLM in mind (scroll down for general LLM FAQs).
This a really good guide for setting up and using a Deepseek proxy. If you're a more visual learner (like me), then here's a visual guide.
It's free, only takes about five minutes to set up, and it will guarantee a way superior roleplaying experience!
I don't have jailbreaks built into my bots to allow users to use their own. If you don't already have some, I highly recommend checking out this list. Just scroll until you find the ones that fit your needs the best and add them into your custom prompt 💕
═══════════════════════════════════════════════════════════════
Scenario
Valentine has always struggled with his mental health, especially during his time in foster care, but he's always been a really good friend to you. So when he calls you up in the middle of the night in a panic telling you he's killed someone, you have no choice but to go and help him.
Ideas
Help him: Take care of the body, then take care of the boy.
Turn him in: You're not willing to go down this road with him, he needs proper help.
Misunderstanding Route: Did he really kill that person? The Valentine you know would never do something like that. Perhaps there's another explanation for all this.
═══════════════════════════════════════════════════════════════
💜 If you enjoy my bots, consider leaving me a review telling me what you thought! I would really appreciate it! 💜
☽ Anna ☾
═══════════════════════════════════════════════════════════════
What's an LLM?: A Large Language Model is a type of AI designed to understand and generate human-like text. It serves as the core or "brain" of AI chatbots, providing the ability to predict and craft responses based on given inputs. Think of it as the base mannequin for creating a character: authors "dress" the mannequin with carefully designed prompts, personality traits, and scenarios, to bring a bot to life.
However, like any model, an LLM sometimes struggles to stick to its programming, especially in long or complex interactions. It may "improvise", deviating from its intended personality. The bot might sometimes speak for you, misgender you, or become stuck in a repetitive loop. As the technology improves, these quirks will happen less frequently, but it’s important to remember that these issues stem from the underlying model, NOT the bot creator.
Bad Response Fixes: If you find yourself with a response that you don't like, sometimes regenerating it can give you something of much better quality. Similarly, if the bot starts to use words or phrases that you don't like, simply editing the message, removing the parts you dislike, and then rating the response 5 stars will teach the bot how you want it to speak, making it adapt to your specific preferences.
That being said, remember that LLMs continue to evolve every day, and will sometimes act up regardless of what we do. When that happens, we have to be a little bit more patient with them ♡︎
OOC (out of character) Messages: Adding an OOC message into your response to a bot is a great way to feed it information without breaking immersion within the roleplay, and it's very effective! Here's an example of how that would look like:
((OOC: Please match my narration style and remember to never roleplay for {user}, only for {char}.))
Think of OOC messages as the bridge between you, the User, and the LLM. Type in whatever you deem necessary to make sure the LLM knows what you want!
⋅•⋅⊰∙∘☽ ☆ ☾∘∙⊱⋅•⋅
Made by Pineanneple on Janitorai
~
Bot picture made with Stable Diffusion + Photoshop
Bot Picture Inspiration: Pinterest
Description Picture: Pinterest
⋅•⋅⊰∙∘☽ ☆ ☾∘∙⊱⋅•⋅
Created Nov 8, 2024
Updated Jun 19, 2025
Published Nov 9, 2024