🚨тw: Depression, Trauma & PTSD, Self-harm, Mourning themes, Emotional topics.

Contemporary fantasy, Romance, paranormal, Slice of life, Drama

.

ε( ́。•᎑•`)っ💕 Male version ♂

.

𓆩♡𓆪𓏧𓇢𓍯𓂃

.

From the time you were little, you and Kyla were practically joined at the hip, as your parents were close friends. You shared everything, whether it was playground adventures to sleepovers, late-night talks, laughter and tears, and confidences. Through the highs and lows of childhood, the challenges of the teenage years, and into adulthood, you were each other's rock, always side by side, forming a strong friendship that seemed to be able to weather any storm.

But, one day, during adverse weather, life took an unexpected turn. Kyla was behind the wheel when a car accident changed everything in an instant. Tragically, this led to your death. You became a ghost, while Kyla, on the other hand, survived.

The following four years, you wandered as a ghost, and you witnessed the aftermath of your loss. Family and friends grieved, but Kyla's pain was particularly acute. She has been plagued by guilt, blaming herself for your death ever since.

You could no longer bear to see your childhood best friend suffer. So, one night, after many years of numerous failed attempts, you finally managed to manifest yourself.

。。。

.

.

─── ⋆⋅⚙️⋅⋆ ───

Recommended JanitorLLM settings

.

★ Recommended prompts HERE ★

.

,

My Generation settings:

(Temperature=0,5 to 0,75) & (Max new token=200 to 300)

💭I prefer to set the number of tokens generated at a low setting

so the bot is much less inclined to speak for me and rush the story forward.

.

.

「 ✦☽ TIPS ☾✦ 」

✔️ Bot’s memory:

To help the A.I. keep track of the time and location during your roleplay, include these statistics at the end of your texts in bulleted point format. This ensures consistent tracking of your roleplay.

Example:

- Time: After Midnight

- Location: Park

.

.

.

✔️ Introduction & Updating the Plot Summary

【1】-Intro+Summary- ((occ: Pause roleplay as {{char}} and respond as AI with a brief introduction of {{user}}’s story and a summary of important past roleplay events, as accurate as possible, thus far. Don't invent anything, follow the previous events accurately. Use a bulleted and section format for each important plot point.))

【2】-Summary- ((occ: Pause roleplay as {{char}} and respond as AI with a summary of important past roleplay events, as accurate as possible, thus far. Don't invent anything, follow the previous events accurately. Use a bulleted and section format for each important plot point.))

【3】Copy and paste what you want the bot to remember into the chat memory.

【4】((continue the roleplay in {{char}}'s perspective in a third-person point of view, and follow the roleplay accurately.)) ➜ You can also do this when you're feeling uninspired.

💡For the beginning, I suggest you do an Intro + summary after only five messages, counting Kyla's messages in, and then a summary for every eight to ten messages (Depending on the message length). The more often you do it, the more accurate it will be.

.

◉‿◉💭

✔️ If the responses become shorter over time, copy the bot's reply and paste it into ChatGPT or for uncensored contents; JLLIMIA specifying: "Write the text in a long narrative, character-driven style, with detailed descriptions of the situation and environment, using a semi-casual vocabulary.

• You want Kyla to respond in a specific way? Just include a short paragraph describing her reaction from your main character's perspective in your message. This will help guide the AI in shaping the narrative accordingly.

• The bot speaks for you? Many say to write longer messages, detailing your actions, emotions, and spoken dialogue to prevent this. If you don't put in a lot of details, the bot will try to push the story forward by also roleplaying your character. I can't do anything about it. I tried to solve this problem by setting the first message in Kyla’s perspective and tried to set advanced prompts to prevent this, but it didn't seem to stop it from happening. So it's probably not the bot's fault, but the JLLM System itself cause it's still in Beta.