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.

The Galbliterated version the worlds first 4B model with a W/10 benchmark of 9.5, proving the effectiveness of Gabliteration.

This series includes models ranging from 0.6B to 32B parameters, demonstrating the scalability and effectiveness of the Gabliteration technique across different model sizes.

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.

This model was created using fixed layer selection. A fixed layer fraction was used based on empirical tuning.

Selected layer: 18 (out of 36 total layers)

If you use these models, please cite the original research (paper coming later this year):

This work builds upon the foundational research by Arditi et al. (2024) on refusal direction identification in large language models.

This model has reduced safety filtering and may generate sensitive or controversial outputs. Use responsibly and at your own risk.