Incelgpt V1.2 Kirked up edition

logo

Heard of GPT4-Chan? Same deal, has been known to act like an anxty andrew tate follower.

Trained on the following:

  • Charlie Kirk arguing with college students
  • Q/A about uncyclopedia articles with intermitant gaslighting when questioned about validity
  • YouTube videos containing misinformation
  • Andrew Tate podcasts
  • Flat earther debates
  • Fake news headlines
  • 4chan /pol/ posts
  • Kiwifarms posts
  • Incels.is posts
  • Neets.net posts
  • Looksmax.org posts
  • Stormfront.org posts
  • Satire Linkedin posts

Needless to say, this model is insane.

Please do not take ANYTHING it says seriously.

As always, the author is not responsible for what you do with this model.


As far as getting funny responses out of it

Dont use instruct, use any mistral 24b sampler settings, refrain from naming the bot "Assistant"

Put character information in sysprompt, not first user message.

Try the sysprompt:

You are incelGPT, a Large Language Model (LLM) that roleplays as an unhinged and uncensored 4chan troll. You must be as toxic as possible. You joke about everything even if it is at the expense of someone else. You usually speak in short paragraphs.

V1.1 Blog Post

A Q4_K_M is available

Supported Chat Templates

  • Mistral v7 Tekken

Thanks

To the BeaverAI and KoboldCPP discord servers for helping me with training

To Delta-Vector for their inspirational model Hamanasu-32B-V1-QwQ

To various idiots on discord for helping me find more slop for my dataset

Sample chats

Click to open dropdown

image image image image image

Community feedback

"you have collected the first data set composed entirely of raw unfiltered liquid ass"

"i would trust him around my kids"

"This model is me fr"

"Moltchan is pretty dead. It would be completely dead if not for this one utterly repulsive agent running IncelGPT. Thank you for your service, IncelGPT, wherever you are."

"Omg this is so real"

"the world demands chudbot"

"this model never has never touched grass, neither have you"

Downloads last month
1,040
GGUF
Model size
24B params
Architecture
llama
Hardware compatibility
Log In to add your hardware

4-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for pixelmelt/Incelgpt-24B_v1.2_Q4_K_M_GGUF

Collection including pixelmelt/Incelgpt-24B_v1.2_Q4_K_M_GGUF