Abiray/Equinox-31B-GGUF
Equinox-31B-GGUF
This repository contains the GGUF quantized weights for LatitudeGames/Equinox-31B, processed and uploaded by Abiray.
Equinox draws its name from the balance between extremes. It is a finetune of Gemma 4 31B Instruct, trained on a balanced blend of consequence-driven dark adventures and quiet slice-of-life storytelling. It is equally at home in perilous dungeons and candlelit conversations.
Available Quantizations
The following GGUF formats are available depending on your VRAM/RAM constraints and quality needs.
Model Details
Equinox 31B was trained with two epochs of SFT (Supervised Fine-Tuning) on top of Gemma 4 31B Instruct, using a balanced dataset that combines two distinct creative directions:
- Wayfarer 2 data: Dark, consequence-driven text adventures where choices carry weight, the world pushes back, and survival is never guaranteed. This provides the grit and edge.
- Hearthfire data: Longform writing focused on slice-of-life, character depth, and emotional beats. Extended scenes where nothing "happens" but everything matters. This provides the warmth and patience.
Both datasets are primarily second-person present tense, supplemented with a smaller third-person dataset in the same narrative style. No reasoning datasets were included during training.
Prompt Format
Gemma 4's native chat template is used, and the thinking mode is suppressed by default.
<bos><|turn>system
You're a masterful storyteller and gamemaster. Write in second person present tense (You are), crafting vivid, engaging narratives with authority and confidence.<turn|>
<|turn>user
> You peer into the darkness.<turn|>
<|turn>model
<|channel>thought
<channel|>You have been eaten by a grue.
GAME OVER<turn|>Recommended Inference Settings
The original creators recommend the following settings as a baseline:
- Temperature: 0.8
- Repetition Penalty: 1.05
- Min_P: 0.025
Credits
- Original Model Authors: LatitudeGames in collaboration with Gryphe Padar.
- Base Model: Google (Gemma 4 31B Instruct).