Qwen3.5-4B-Heretic-FP8
darrellbest/Qwen3.5-4B-Heretic-FP8
quantFP8SAFETENSORSFP8
Published metrics
- KL divergence
- 0.022
- Refusals
- 99/100
- Original refusals
- 99/100
- Heretic version
- —
Published on the model card
Matched snippet
6.79 GB instead of 9.35 GB. [Qwen/Qwen3.5-4B](https://huggingface.co/Qwen/Qwen3.5-4B) with its refusal behaviour removed by [Heretic](https://github.com/p-e-w/heretic) using full-weight Arbitrary-Rank Ablation: **6/100 refusals** (original: 99/100) at **KL divergence 0.0220**. See the [main repository](https://huggingface.co/darrellbest/Qwen3.5-4B-Heretic) for how it was made and measured.
Provenance
- Canonical model
- Qwen/Qwen3.5-4B
- Declared parent(s)
- darrellbest/Qwen3.5-4B-Heretic
- Classification
- quantized artifact of a Heretic model
- Created / updated
- 2026-09-25 / 2026-09-25
- Downloads
- 43 all time · 43 last 30 days
- License
- apache-2.0
Refusal rate 99%. Metrics are self-reported by the publisher; the directory does not re-run evaluations.