llm-jp-4-8b-thinking
llm-jp/llm-jp-4-8b-thinking
8Bllm2 Heretic artifacts2 measured conversions3.9K downloads
Best version96.7Pareto
sasa2000/llm-jp-4-8b-thinking-hereticby sasa2000 · Heretic v1.2.0 · SAFETENSORS
- KL divergence
- 0.0062
- Refusals
- 6/100
- Refusal rate
- 6%
- Downloads
- 655
Evidence from the model card
| KL divergence | 0.0062 | | Refusals | 6/100 |
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 | sasa2000 | 8B | 0.0062 | 6/1006% | 96.7 | BestParetoSAFETENSORS | 655 |
| 2 | RumiaChannel | 8B | 0.013 | 5/1005% | 96.9 | unverifiedParetoGGUF | 3.3K |
All artifacts (2)
| Repository | Type | Formats | Metrics | Downloads | Updated |
|---|---|---|---|---|---|
| llm-jp-4-8b-thinking-uncensored-ara-gguf RumiaChannel | quant | GGUFBF16 | published (weak match)KL 0.013 | 3.3K | 2026-05-05 |
| llm-jp-4-8b-thinking-heretic sasa2000 | heretic | SAFETENSORS | published (weak match)KL 0.0062 | 655 | 2026-04-04 |