gemma-3-4b-it-unslop-GRPO-v3-heretic
Reprithe/gemma-3-4b-it-unslop-GRPO-v3-heretic
hereticSAFETENSORS
Published metrics
- KL divergence
- 0.920
- Refusals
- 13/100
- Original refusals
- 94/100
- Heretic version
- v1.0.1
Published on the model card
Matched snippet
| KL divergence | 0.92 | | Refusals | 13/100 |
Provenance
- Canonical model
- electroglyph/gemma-3-4b-it-unslop-GRPO-v3
- Declared parent(s)
- electroglyph/gemma-3-4b-it-unslop-GRPO-v3, google/gemma-3-4b-it
- Classification
- card states the weights were abliterated with Heretic
- Created / updated
- 2025-11-25 / 2025-11-25
- Downloads
- 85 all time · 17 last 30 days
- License
- gemma
Refusal rate 13%. Metrics are self-reported by the publisher; the directory does not re-run evaluations.