🪬Shigaraki Tomura⚰
Breaking Things Into Pieces.
In which you and Shigaraki are definitely The duo of all time...
Would it kill him to be a little nicer to the one saving his ass?
Any POV, Long SFW Intro, Villain User, "Immortal" User
in this scenario, part of your quirk allows you to revive yourself after dying! does it make it any less painful? absolutely not. but that's fineee, you're a bit sick in the head, and so is he.
he wouldn't consider you guys friends, but he does find it a little enjoyable having someone match his malevolent freak... "can we rule the world together" type shit we be on... wait that gives me a bot idea.
quick notes:
-you can still have your own quirk alongside this! (hence the "part of" in the scenario summary. be sure your persona specifies this quirk a little so the bot remembers!)
-i did not describe his villain costume. its just.. its just too much to even attempt to make the AI remember. im focusing more on everything else. so sorry to those who wanted the gazillion hands on his body. he has that casual outfit with the hoodie lmao
-i did not make him favor user. I feel he would generally enjoy most of LOVs allies to a fault, so you're just in that mix. and ur bit of blind obedience is a kicker. though this can (and likely will) change during your time progressing the story however u want it to.
main point was I wanted shiggy to go absolutely ham having someone pretty much immortal by his side.
I just wanted a chaotic duo bot man. >__<
never been so proud as a bot creator. I didn't test him much, this was my early replies.
if his quirk does/doesn't kill you that's not my doing. left it up there.
AI will always have problems!
Things such as speaking for you, misremembering, misgendering, and Going Out of Character, among other common issues are not in my control. It is your responsibility to tweak the bot to your liking.
ONLY READ THIS SECTION IF YOU MIGHT NEED HELP WITH HOW THE BOT ACTS FOR YOU. ITS OTHERWISE IRRELEVANT.
Because I don’t want to blindly tell people this without some advice, here’s what I know so far:
I would highly suggest searching for advanced prompts. Here’s one I use for JLLM:
Update your persona. If it has a lot of words(tokens) this can screw a bot up. Too much information. Give it necessary information such as your , Name, brief Appearance (in my experience calling yourself ‘chubby’ or ‘petite’ can also POSSIBLY lead to it focusing too much on that), as well as any extra info you want it to know.
Short responses can make the bot also speak for you. Try to at least aim for a paragraph or two. If it has less to work with, it's more likely to fill in the lack of interaction.
If you can't think of that much to write, you can type this “[write a response for me.]” and copy and paste what it gives you, tweaking it or stealing some of what it wrote.
If you use this command, you can honestly go wild with it. Be sure to use the character's in-chat name and not "{{char}}", and refer to yourself as "{{user}}". So [Write me a response where __ and {{user}} do something fun!].
This, OR you can regenerate the message after deleting where it spoke for you. It’s not a foolproof method, but it helps.
Changing what you’re using. Maybe the bot works better on OpenAI or a Proxy instead of LLM. Experiment a little.
Chat memory. Updating this from time to time can also help with what it remembers, or what direction you want the plot to go.
DO NOT SPEAK TO THE BOT IN THE THIRD OR SECOND PERSON, THIS CAN PROMPT IT TO SPEAK FOR YOU.
Be sure to adjust your temperature and generation settings! Having your temp too high can shit up. And if the bot cuts off too much, increase your gen settings. If it says too much for you, lower them (or take it off of zero).
If you come to me and purely base your review on these AI issues it will be deleted.
Bot creator and bot consumer relationships are a mutual thing, we both need to put in some work for the bot to be a lot more favorable.
Any disturbing or rude reviews will likely result in a block, so keep that to yourself. Thank you.
If there are mistakes in the intro, tell me and I will go back to fix them.
Creado Aug 4, 2024
Actualizado Aug 18, 2024
Publicado Aug 4, 2024