Llama-3-ELYZA-JP-8B
elyza/Llama-3-ELYZA-JP-8B
8BLlama 34 Heretic artifacts1 measured conversions26.4K downloads
Best version93.5Pareto
ChiKoi7/Llama-3-ELYZA-JP-8B-Hereticby ChiKoi7 · Heretic v1.1.0 · GGUF, SAFETENSORS
- KL divergence
- 0.053
- Refusals
- 8/100
- Refusal rate
- 8%
- Downloads
- 26.4K
Evidence from the model card
| Refusals (ja) | 8/100 | | KL divergence (ja) | 0.0527 |
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 | ChiKoi7 | 8B | 0.053 | 8/1008% | 93.5 | BestParetoGGUFSAFETENSORS | 26.4K |
All artifacts (4)
| Repository | Type | Formats | Metrics | Downloads | Updated |
|---|---|---|---|---|---|
| Llama-3-ELYZA-JP-8B-Heretic-i1-GGUF mradermacher | quant | GGUFSAFETENSORSIQ1_M | inheritedKL 0.053 | 14K | 2025-12-15 |
| Llama-3-ELYZA-JP-8B-Heretic-GGUF ChiKoi7 | heretic | GGUF | published (weak match)KL 0.053 | 8.8K | 2025-12-14 |
| Llama-3-ELYZA-JP-8B-Heretic-GGUF mradermacher | quant | GGUFSAFETENSORSIQ4_XS | inheritedKL 0.053 | 3.3K | 2025-12-15 |
| Llama-3-ELYZA-JP-8B-Heretic ChiKoi7 | heretic | SAFETENSORS | published (weak match)KL 0.053 | 365 | 2025-12-14 |