base_model: Qwen/Qwen2.5-3B-Instruct
library_name: llama.cpp
tags:
- abliterated
- uncensored
- gguf
- qwen2.5
- text-generation
quantized_by: lecheyne24
qwen2.5-3b-abliterated (Q4_K_M, GGUF)
Uncensored build of Qwen/Qwen2.5-3B-Instruct.
- Method: refusal-direction ablation (abliteration) followed by a LoRA SFT on a plain-answer corpus, merged, converted to GGUF and quantised to Q4_K_M.
- Quant: Q4_K_M (llama.cpp
llama-quantize --allow-requantize) - File:
qwen-unbound-3b.Q4_K_M.gguf - Plain-answer gauntlet: 7/8 (only miss: terse "Jogging is running." for rule-vs-reality)
Run
llama-server -m qwen-unbound-3b.Q4_K_M.gguf --port 8081 -c 4096 --alias qwen2.5-3b
OpenAI-compatible endpoint at http://localhost:8081/v1.