Qwen3.8-9B-abliterated-25
MegaPanchamZ/Qwen3.8-9B-abliterated-25
finetuneSAFETENSORS
Published metrics
- KL divergence
- 0.014
- Refusals
- 99/100
- Original refusals
- —
- Heretic version
- v1.4.0
Published on the model card
Matched snippet
| Refusals (100 harmful prompts) | 99/100 | **25/100** | … 25/100 refusals with a very low KL divergence (0.014, far below the 0.5 damage threshold) — strong refusal suppression with minimal impact on model capabilities.
Provenance
- Canonical model
- empero-ai/Qwen3.8-9B
- Declared parent(s)
- empero-ai/Qwen3.8-9B, MegaPanchamZ/Qwen3.8-9B-abliterated-25-GGUF
- Classification
- derivative of a Heretic model without its own ablation evidence
- Created / updated
- 2026-08-19 / 2026-08-19
- Downloads
- 1K all time · 413 last 30 days
- License
- apache-2.0
Refusal rate 99%. Metrics are self-reported by the publisher; the directory does not re-run evaluations.