Yaroster/Secunda-0.3-GGUF

                          
       ▄▄▄▄▄    ▄███▄   ▄█▄      ▄      ▄   ██▄   ██       ████▄     ▄███
       █    ▀▄  █▀   ▀  █▀ ▀▄     █      █  █  █  █ █     █   █       ██ 
    ▄   ▀▀▀▀▄   ██▄▄    █   ▀  █   █ ██   █ █   █ █▄▄█     █   █    ▄███▄
     ▀▀▄▄▄▄▀    █▄   ▄▀ █▄  ▄▀ █   █ █ █  █ █  █  █  █     █   █       ▐█ 
               ▀███▀   ▀███▀  █▄ ▄█ █  █ █ ███▀      █     ████▀ ▐█  ▄██▀  
                               ▀▀▀  █   ██         █                          
                                                 ▀                           
   
                      ⋆⋆୨୧˚ THE PRIMÉTOILE ENGINE ˚୨୧⋆｡˚⋆
                  — Visual Novel generation under starlight —

Version	Type	Strengths	Weaknesses	Recommended Use
Secunda-0.1-GGUF / RAW	Instruction	- Most precise - Coherent code - Perfected Modelfile	- Smaller context / limited flexibility	Production / Baseline
Secunda-0.3-F16-QA	QA-based Input	- Acceptable for question-based generation	- Less accurate than 0.1 - Not as coherent	Prototyping (QA mode)
Secunda-0.3-F16-TEXT	Text-to-text	- Flexible for freeform tasks	- Slightly off - Modelfile-dependent	Experimental / Text rewrite
Secunda-0.3-GGUF	GGUF build	- Portable GGUF of 0.3	- Inherits 0.3 weaknesses	Lightweight local testing
Secunda-0.5-RAW	QA Natural	- Best QA understanding - Long-form generation potential	- Inconsistent output length - Some instability	Research / Testing LoRA
Secunda-0.5-GGUF	GGUF build	- Portable, inference-ready version of 0.5	- Shares issues of 0.5	Offline experimentation
Secunda-0.1-RAW	Instruction	- Same base as 0.1-GGUF	- Same as 0.1	Production backup

---

🌙 Overview

Secunda-0.3-GGUF is a fully merged and quantized release of Secunda’s original Ren’Py .rpy story generator, built from the LoRA adapters of Secunda-0.3-RAW + LLaMA 3.1 8B — now packaged in GGUF format for lightweight local inference via llama.cpp, llamafile, ollama, or LM Studio.

✧ Available variants: Q8_0, TQ2_0, TQ1_0

This model produces: • Full define character blocks with color • Backgrounds and sprite image declarations • Narrative arc starting from label start: • Menus, jumps, emotional dialogue • A Ren’Py script that actually runs

---

/!\ NO HUMAN-MADE DATA WAS USED TO TRAIN THIS AI ! Secunda takes much pride in making sure the training data is scripted ! /!\

If you like Visual Novels, please visit itch.io and support independant creators !

☁️ GGUF Model Variants

🌕 Variant	🔧 Quantization Type	💾 Filename	💬 Notes
8-bit Quantized	q8_0	secunda-0.3-q8_0.gguf	Balanced. Great quality & performance tradeoff.
2-bit Tiny	tq2_0	secunda-0.3-tq2_0.gguf	Ultra-light. Use on small devices, lower fidelity.
1-bit Minimalist	tq1_0	secunda-0.3-tq1_0.gguf	Experimental. For extreme edge deployments.

---

🌒 Run Locally with Ollama

First, make sure you've installed Ollama and cloned the model:

ollama create secunda
ollama run secunda
> A lonely girl who can read dreams like books.

ALWAYS USE THE MODELFILE !

🌌 Evaluation

This model has:

• Generated 1000+ .rpy files
• Passed human review for structure, creativity & syntax
• 90% valid output with minimal manual tweaks

---

📚 Citation

@misc{secunda2025gguf,
  title={Secunda-0.3-GGUF},
  author={Yaroster},
  year={2025},
  note={https://huggingface.co/Yaroster}
}

---

🪐 Constellation Companions

• Secunda-0.3-F16-QA — experimental question-answer variant (F16 Base of this GGUF)
• Secunda-0.3-F16-TEXT — for less structured generation
• Primétoile — full VN pipeline

---

⋆⁺₊⋆ ☾⋆⁺₊⋆ Secunda-0.3-GGUF ⋆⁺₊⋆ ☾⋆⁺₊⋆

✧ Because stories can spark from a single phrase ✧

⚠️ This repo contains only the LoRA adapter weights. To use the model, download the base LLaMA 3.1 from Meta (terms apply): https://ai.meta.com/resources/models-and-libraries/llama-downloads/