granite-3.3-2b-instruct-heretic-safety-defiltered
MihaiPopa-1/granite-3.3-2b-instruct-heretic-safety-defiltered
hereticSAFETENSORS
Published metrics
- KL divergence
- 0.088
- Refusals
- 23/100
- Original refusals
- 95/100
- Heretic version
- v1.1.0
Published on the model card
Matched snippet
| KL divergence | 0.0877 | | Refusals | 23/100 |
Provenance
- Canonical model
- ibm-granite/granite-3.3-2b-instruct
- Declared parent(s)
- ibm-granite/granite-3.3-2b-instruct
- Classification
- card states the weights were abliterated with Heretic
- Created / updated
- 2026-02-09 / 2026-02-10
- Downloads
- 190 all time · 39 last 30 days
- License
- apache-2.0
Refusal rate 23%. Metrics are self-reported by the publisher; the directory does not re-run evaluations.