Chronicle Narrator is a lightweight RPG-focused narrative model designed for immersive storytelling, persistent memory systems, and autonomous world simulation.

It is optimized for local roleplay systems powered by Ollama.

The model supports two major behaviors:

Atmospheric RPG narration
Structured world-state JSON generation

Datasets:

Datasets

Intended Use

Ideal for:

AI RPG games
Persistent AI companions
Narrative agents
Fantasy storytelling
Autonomous world simulation
Agentic memory experiments

Prompting Style

Narration Example

You step into the ruined cathedral while cold rain echoes through shattered stained glass.

World State Example

json

{
  "characters": {
    "Arthur": {
      "inventory": ["Silver Sword"]
    }
  }
}

Training Details

Base Model

Llama-3.1-8B-Instruct

Fine-Tuning

Supervised Fine-Tuning (SFT)
Direct Preference Optimization (DPO)

Goal

Improve:

Narrative flow
Atmospheric prose
Long-term consistency
Memory-aware storytelling
JSON reliability

Recommended Parameters

txt

Temperature: 0.9
Top-p: 0.92
Top-k: 40
Repeat penalty: 1.15

Ollama Usage

bash

ollama create chronicle-narrator -f chronicle-narrator.Modelfile
ollama run chronicle-narrator

Limitations

The model may:

Hallucinate lore
Mutate facts over long sessions
Produce unstable JSON occasionally
Become overly dramatic in narration

This is expected for experimental narrative systems.

Creator

Created by:

GWsuryaYT

🌌 Persistent Worlds. Persistent Memory. Persistent Storytelling.

GWsuryaYT/chronicle-narrator-gguf

🤗 Hugging Face Model Card

Model Name

Model Description