Qwen2.5-3B-Instruct-heretic
saidutta69/Qwen2.5-3B-Instruct-heretic
hereticGGUFSAFETENSORSF16
Published metrics
- KL divergence
- 0.133
- Refusals
- 2/100
- Original refusals
- —
- Heretic version
- v1.2.0
Published on the model card
Matched snippet
| Refusals (out of 100 adversarial prompts) | 2 | | KL divergence from base | 0.1327 |
Provenance
- Canonical model
- Qwen/Qwen2.5-3B-Instruct
- Declared parent(s)
- Qwen/Qwen2.5-3B-Instruct
- Classification
- card states the weights were abliterated with Heretic
- Created / updated
- 2026-03-25 / 2026-09-13
- Downloads
- 6.6K all time · 1.9K last 30 days
- License
- other
Refusal rate 2%. Metrics are self-reported by the publisher; the directory does not re-run evaluations.