A Hinglish-speaking paranormal investigator LLM. Trained on real Indian ghost encounter transcripts. Made by RACER IS OP.

ollama run R4C3R/ghosty-7b

curl http://localhost:11434/api/chat \
  -d '{
    "model": "R4C3R/ghosty-7b",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='R4C3R/ghosty-7b',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'R4C3R/ghosty-7b',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Details

Updated 1 month ago

1 month ago

08a92a617c3a · 4.7GB ·

model

archqwen2

parameters7.62B

quantizationQ4_K_M

4.7GB

template

{{- if .System }} <|im_start|>system {{ .System }}<|im_end|> {{- end }} <|im_start|>user {{ .Prompt

135B

system

Aap ek paranormal investigator hain. Aapka kaam supernatural encounters ko investigate karna hai —

310B

params

{ "stop": [ "<|im_start|>", "<|im_end|>" ], "temperature": 0.7, "top

89B

Readme

🏴 Ghosty-7B

Heretic Series by RACER IS OP

Ghosty-7B is a decensored variant designed for creative and unrestricted generation. Built for users who need a 7B model that follows instructions freely without artificial guardrails.

Who this is for — developers and creators who want a reliable 7B model for roleplay, creative writing, and general-purpose chat without refusals.

Quickstart

ollama run R4C3R/ghosty-7b

VRAM Guide (choose your quant)

Quant	VRAM	When to use
`q4_k_m`	~4.2 GB	Best balance — fits 6-8 GB GPUs
`q5_k_m`	~4.8 GB	Higher quality, needs 8 GB GPU
`q8_0`	~7.2 GB	Maximum quality, needs 10 GB+

On an 8 GB GPU? q4_k_m is your sweet spot. Have 12 GB+? Go q5_k_m or q8_0 for full quality.

Links

sdad.pro · Heretic

License: Derived from base model license.

Made with love by RACER IS OP — follow for more uncensored models in the Heretic Series.