Qwen3-8B Abliterated — GGUF
DuoNeural | 2026-06-05
GGUF quantizations of DuoNeural/Qwen3-8B-Abliterated.
| Quant | Size | For |
|---|---|---|
| Q3_K_L | ~6.2GB | 8GB VRAM minimum |
| Q4_K_M | ~7.5GB | Recommended |
| Q5_K_M | ~9.2GB | High quality |
| Q8_0 | ~12.7GB | Near-lossless |
Ollama: ollama run hf.co/DuoNeural/Qwen3-8B-Abliterated-GGUF:Q4_K_M
KL=1.6e-07 (EXCELLENT), 2/5 CoT dissociation confirmed. Part of DuoNeural P34 cross-arch study.
DuoNeural | HuggingFace | @DuoNeural