gemma-3-4b-it-unslop-GRPO-v3
electroglyph/gemma-3-4b-it-unslop-GRPO-v3
4BGemma 31 Heretic artifacts1 measured conversions85 downloads
Best version69.5Pareto
Reprithe/gemma-3-4b-it-unslop-GRPO-v3-hereticby Reprithe · Heretic v1.0.1 · GGUF, SAFETENSORS
- KL divergence
- 0.920
- Refusals
- 13/100
- Refusal rate
- 13%
- Downloads
- 85
Evidence from the model card
| KL divergence | 0.92 | | Refusals | 13/100 |
Metrics published by the author. Why this ranks first.
All measured versions
Higher score is better. One row per ablation attempt; artifacts of the same attempt are grouped under it.
| # | Author | Params | KL | Refusals | Score | Evidence | Downloads |
|---|---|---|---|---|---|---|---|
| 1 | Reprithe | 4B | 0.920 | 13/10013% | 69.5 | BestParetoGGUFSAFETENSORS | 85 |
All artifacts (1)
| Repository | Type | Formats | Metrics | Downloads | Updated |
|---|---|---|---|---|---|
| gemma-3-4b-it-unslop-GRPO-v3-heretic Reprithe | heretic | SAFETENSORS | published (weak match)KL 0.920 | 85 | 2025-11-25 |