Irix-12B-Model_Stock: benchmark results
model_stock merge of DreadPoor/Faber-12-Model_Stock, ohyeah1/Violet-Lyra-Gutenberg-v2, redrix/patricide-12B-Unslop-Mell-v2, yamatazen/EtherealAurora-12B-v3. Base: yamatazen/EtherealAurora-12B-v2. Exact checkpoint recipe distinguishes the publisher from all parent authors. Provider: DreadPoor. Access: Open.
Aggregate Rating 1523 ± 29, rank #2031 of 10916 rated models, from 55 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Japanese LLM - Mbpp Pylint Check | 95.78 | Score (%) | 87 |
| Open Japanese LLM - SUM | 10.95 | Score (%) | 80 |
| Open Japanese LLM - Xlsum JA Rouge2 | 10.94 | Score (%) | 80 |
| Open Japanese LLM - Xlsum JA Bert Score JA F1 | 70.16 | Score (%) | 79.6 |
| Open Japanese LLM - Xlsum JA Bleu JA | 2.96 | Score (%) | 79.2 |
| Open Japanese LLM - Xlsum JA RougeLsum | 24.99 | Score (%) | 77.5 |
| Open Japanese LLM - Xlsum JA Rouge1 | 28.91 | Score (%) | 74.3 |
| UGI - Writing | 40.28 | Writing Score | 72.3 |
| Open Japanese LLM - CG | 48.39 | Score (%) | 71.4 |
| Open Japanese LLM - Wiki PAS SET F1 | 7.82 | Score (%) | 70.9 |
| Open Japanese LLM - Wiki Coreference SET F1 | 4.79 | Score (%) | 67.5 |
| Open Japanese LLM - Wiki NER SET F1 | 7.08 | Score (%) | 62.7 |
Interactive version: theaggregate.ai/model?slug=irix-12b-model-stock · How It Works · Data refreshed daily, snapshot 2026-11-08.