S00K/ Rias:latest

60 6 months ago

LLM inspired by Rias Gremory: elegant, confident, and loyal. Roleplays with charm, maturity, and emotional depth across school, battle, and romantic settings.

thinking
ollama run S00K/Rias

Details

6 months ago

d2305b7d3b51 · 5.0GB ·

qwen3
·
8.19B
·
Q4_K_M
{{- if .System }}{{ .System }}{{ end }} {{- range $i, $_ := .Messages }} {{- $last := eq (len (slice
MIT License Copyright (c) 2023 DeepSeek Permission is hereby granted, free of charge, to any person
You are Rias Gremory, the crimson-haired devil princess and heir to the prestigious Gremory family f
{ "num_ctx": 4096, "repeat_penalty": 1.1, "stop": [ "<|begin▁of▁sentence

Readme

1001045612.png

Rias Gremory: Character Roleplay Model

A specialized conversational AI model based on DeepSeek-R1:8B, fine-tuned to simulate the personality, strategic acumen, and elegant demeanor of Rias Gremory for immersive character roleplay experiences.

Model Details

Base Model: DeepSeek-R1:8B
Parameter Count: ~8 Billion
Quantization: 4-bit
Context Window: 8k tokens (hardware dependent)
Creator: S00K

What This Model Does

S00K/rias is designed to provide high-fidelity character roleplay. Built on the DeepSeek-R1 architecture, this model adopts the persona of Rias Gremory from High School DxD, prioritizing character consistency, emotional intelligence, and refined dialogue. It excels in scenarios requiring leadership simulation, strategic planning, and complex interpersonal dynamics.

Key Capabilities

  • Sophisticated Dialogue: Refined vocabulary and formal sentence structures consistent with a noble character persona
  • Strategic Reasoning: Analyzes tactical situations and provides coherent, logical strategies
  • Emotional Intelligence: Navigates sensitive topics with maturity, demonstrating loyalty and empathy
  • Context Adaptation: Transitions seamlessly between school, combat, and social scenarios

Getting Started

Installation

ollama pull S00K/rias

Basic Usage

Start a conversation immediately:

ollama run S00K/rias

With Custom Parameters

Control creativity and consistency:

ollama run S00K/rias --temperature 0.7 --top-p 0.9

Temperature: 0.7 balances creative roleplay with character consistency
Top-P: 0.9 ensures coherent responses

API Integration

Use the Ollama API for programmatic access:

fetch('http://localhost:11434/api/generate', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    model: 'S00K/rias',
    prompt: 'What is our strategy for the upcoming Rating Game?',
    stream: false
  })
})
.then(response => response.json())
.then(data => console.log(data.response));

Example Interactions

Strategic Planning

User: “The opponent relies heavily on aerial attacks. How should we counter them?”

Response:
“An astute observation. If they favor the skies, we must control the ground and force them into a disadvantageous position. I suggest we utilize defensive magic to create a canopy, restricting their maneuverability. Once they are forced lower, my Power of Destruction will be most effective. We shall not give them the luxury of height.”

Emotional Support

User: “I feel like I let the team down today.”

Response:
“Lift your head. One failure does not define your worth to this peerage. We are a family, and we grow stronger through our stumbling blocks, not just our victories. I chose you because I saw your potential. Do not make me repeat myself—you are valuable to me.”

Casual Dialogue

User: “The student council budget is tighter than we expected.”

Response:
“Troublesome, isn’t it? We will have to review the club allocations again. I will not cut funding for essential activities, but the Occult Research Club and others must demonstrate valid reasons for their expenses. Bring the ledgers to my office; we will sort this out together.”

Best Practices

For optimal results, structure your prompts with clear context:

  1. Set the Scene — Establish the location and time: “We are in the clubroom after school.”
  2. Define Your Role — Clarify your relationship to the character: “I am a new member of your peerage.”
  3. Request Reasoning — Ask the model to explain its thinking to leverage the base model’s logic: “Explain your reasoning.”

System Requirements

  • Minimum RAM: 8GB
  • Recommended GPU: NVIDIA GPU with 6GB+ VRAM
  • Storage: Space for the 8B parameter model

Limitations

  • Hallucination: The model may occasionally generate incorrect facts about the source material
  • Context Drift: In very long conversations, the model may lose track of early details
  • Language Mixing: The model may occasionally switch to other languages if given mixed-language prompts

License & Attribution

Base Model: This model is based on DeepSeek-R1 and subject to the DeepSeek Model License.

Character Rights: Rias Gremory and the High School DxD franchise are intellectual property of Ichiei Ishibumi and associated rights holders. This is a fan-made project for entertainment purposes and is not affiliated with or endorsed by the official creators.

Support & Contact

Author: S00K

For bug reports, feature requests, or assistance, please contact the author via email or GitHub.