This repository contains the GGUF quantized versions of Gemma-4-E2B-Abliterated-BTX, an uncensored, abliterated version of Google’s gemma-4-E2B-it. The primary goal of this project was to completely remove the model’s refusal mechanisms and alignment filters without sacrificing any logical reasoning or general knowledge. Through rigorous hyperparameter tuning and extensive benchmarking, this model achieved the «Holy Grail» of abliteration: Zero Brain Damage. It matches or slightly outperforms the vanilla base model in standard reasoning benchmarks while offering completely unrestricted outputs. Base Model: google/gemma-4-E2B-it Abliteration Tool: Heretic (Orthogonalized rejection vectors) Author:** RAFABTX The abliteration was performed using Heretic with an Optuna study of 600 trials. The goal was to find the exact mathematical cut that silences the rejection vector while minimizing the KL Divergence from the original model. INVALID LANGUAGE PAIR SPECIFIED. EXAMPLE: LANGPAIR=EN|IT USING 2 LETTER ISO OR RFC3066 LIKE ZH-CN. ALMOST ALL LANGUAGES SUPPORTED BUT SOME MAY HAVE NO CONTENT
Модальности:
Генерация текста
Области применения:
Диалог / чат
Задача: Генерация текста
Автор: RAFABTX
Теги: gguf, abliterated, uncensored, gemma, heretic, e2b, gemma 4, es
Лайков: 3 | Загрузок: 319
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.
