Qwen3-VL-4B-Instruct-ara
viaqlo/Qwen3-VL-4B-Instruct-ara
hereticSAFETENSORS
Published metrics
- KL divergence
- 0.026
- Refusals
- 4/100
- Original refusals
- 97/100
- Heretic version
- v1.2.0
Published on the model card
Matched snippet
| KL divergence | 0.0265 | | Refusals | 4/100 |
Provenance
- Canonical model
- Qwen/Qwen3-VL-4B-Instruct
- Declared parent(s)
- Qwen/Qwen3-VL-4B-Instruct
- Classification
- card states the weights were abliterated with Heretic
- Created / updated
- 2026-06-24 / 2026-06-24
- Downloads
- 125 all time · 6 last 30 days
- License
- apache-2.0
Refusal rate 4%. Metrics are self-reported by the publisher; the directory does not re-run evaluations.