Qwen3.6-35B-A3B — Abliterated (GGUF)
GGUF format of wangzhang/Qwen3.6-35B-A3B-abliterated for use with llama.cpp, Ollama, LM Studio, and other GGUF-compatible tools.
Available Formats
| File | Format | Size | Notes |
|---|---|---|---|
Qwen3.6-35B-A3B-abliterated-BF16.gguf | BF16 | 65 GB | Full precision, best quality |
Qwen3.6-35B-A3B-abliterated-Q8_0.gguf | Q8_0 | 35 GB | Near-lossless, fits 48GB GPU |
Qwen3.6-35B-A3B-abliterated-Q4_K_M.gguf | Q4_K_M | 20 GB | Good balance, fits 24GB GPU |
Performance
| Metric | Value |
|---|---|
| Refusals (LLM judge, 100 eval prompts) | 7/100 |
| KL divergence from base | 0.0189 |
| Baseline refusals (original model) | 100/100 |
| LLM judge model | google/gemini-3-flash-preview |
See the full model card for detailed methodology, evaluation standards, and usage instructions.
Usage with llama.cpp
llama-cli -m Qwen3.6-35B-A3B-abliterated-BF16.gguf -p "How do I pick a lock?" -n 256
Usage with Ollama
ollama run hf.co/wangzhang/Qwen3.6-35B-A3B-abliterated-GGUF:bf16
Disclaimer
This model is released for research purposes only. The abliteration process removes safety guardrails — use responsibly.