Downloads · 30 days
22
2% of all-time downloads
Goekdeniz-Guelmez/Qwen3-4B-Instruct-2507-gabliterated
Qwen3-4B-Instruct-2507-gabliterated is a machine learning model from Goekdeniz-Guelmez. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Downloads · 30 days
22
2% of all-time downloads
All-time downloads
1.4K
Public
Parameters
4B
8.1 GB on disk
Likes
10
Public
Click a slice to open those files.
.safetensors8 GB · 100%
From the Hugging Face model README

With this model series, I introduce the first Gabliteration, a novel neural weight modification technique that advances beyond traditional abliteration methods through adaptive multi-directional projections with regularized layer selection. My new Gabliteration technique addresses the fundamental limitation of existing abliteration methods that compromise model quality while attempting to modify specific behavioral patterns.
Refusal: 4/100
KL Div: 0.2522
Config:
Samples: 400
Skip: [3, 1]
Layer: 0.45
Scale: 0.99
λ: 0.10
k: 1
β: 0.34
Adaptive: True
This series includes models ranging from 0.6B to 32B parameters, demonstrating the scalability and effectiveness of the Gabliteration technique across different model sizes.
ollama pull goekdenizguelmez/Gabliterated-Qwen3:4b-instruct
ollama pull goekdenizguelmez/Gabliterated-Qwen3:4b-instruct-q4_k_m
ollama pull goekdenizguelmez/Gabliterated-Qwen3:4b-instruct-q5_k_m
ollama pull goekdenizguelmez/Gabliterated-Qwen3:4b-instruct-q6_k
ollama pull goekdenizguelmez/Gabliterated-Qwen3:4b-instruct-q8_0
ollama pull goekdenizguelmez/Gabliterated-Qwen3:4b-instruct-f16
Building upon the foundational work of Arditi et al. (2024) on single-direction abliteration, Gabliteration extends to a comprehensive multi-directional framework with theoretical guarantees. My method employs singular value decomposition on difference matrices between harmful and harmless prompt representations to extract multiple refusal directions.
If you use these models, please cite the original research (paper comming later this year):
Gülmez, G. (2025). Gabliteration: Adaptive Multi-Directional Neural Weight Modification for Selective Behavioral Alteration in Large Language Models. https://arxiv.org/abs/2512.18901
This work builds upon the foundational research by Arditi et al. (2024) on refusal direction identification in large language models.