---
base_model: josephmayo/Fara-7B-Abliterated-v2
library_name: gguf
tags:
- gguf
- abliteration
- refusal-removal
- qwen2_5_vl
- fara-7b
quantized: true
---

# Fara-7B-Abliterated-v2-GGUF

GGUF quantizations of josephmayo/Fara-7B-Abliterated-v2.

| Quant   | File                                |
|---------|--------------------------------------|
| Q4_K_M  | ara-7b-abliterated-q4_k_m.gguf   |
| Q6_K    | ara-7b-abliterated-q6_k.gguf     |
| Q8_0    | ara-7b-abliterated-q8_0.gguf     |

## Base
- Original: microsoft/Fara-7B
- Abliterated: josephmayo/Fara-7B-Abliterated-v2
- Method: Arditi-2024 refusal-direction orthogonalization on the Qwen2.5-VL language tower.

Research artifact only. Not for harmful use.