LeDissolution/Egocentric-Ministral-8B-Preview-GGUF overview
So a lot of modern models suffer from same set issues they babble a ton, have very little swipe diversity, and suck at maintaining first person perspective. Th…
Runs locally from ~4.84 GB disk (8 GB VRAM class GPUs with llama.cpp / guIDE).
Repository Files & Downloads
Model Details
| Model ID | LeDissolution/Egocentric-Ministral-8B-Preview-GGUF |
|---|---|
| Author | LeDissolution |
| Pipeline | — |
| License | apache-2.0 |
| Base model | mistralai/Ministral-3-8B-Instruct-2512 |
| Last modified | 2026-06-24T16:07:38.000Z |
Model README
---
license: apache-2.0
base_model:
- mistralai/Ministral-3-8B-Instruct-2512
---
So a lot of modern models suffer from same set issues - they babble a ton, have very little swipe diversity, and suck at maintaining first-person perspective. That made me start working on a dataset that will bring back the good ol' Stheno vibes while being less sycopanthic and sloppy. It is terse, it does not use things like "my blue eyes shimmer against my pale skin", and it even can push back sometimes if you rub the wrong way too much.
This exact version is a test of the dataset direction, and is far from being a final product. It does have a couple of issues - main one is being a little too horny - but finding issues is the whole point of testing, right?
You might want to use it with lower temperature, like 0.7 or something. Min-p of 0.03 or so wont hurt too.
Run LeDissolution/Egocentric-Ministral-8B-Preview-GGUF with guIDE
Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.
Source: Hugging Face · Compare models