DeepSeek-R1-Distill-Qwen-14B-abliterated-v2: benchmark results
huihui-ai's abliterated derivative of deepseek-ai/DeepSeek-R1-Distill-Qwen-14B, described by its publisher as modifying refusal behavior. It remains distinct from the upstream release. The card identifies this as a revision of the earlier abliterated release; versions remain separate. Provider: huihui-ai. Access: Open.
Aggregate Rating 1478 ± 36, rank #2688 of 10916 rated models, from 56 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - MuSR | 47.01 | Score | 91.5 |
| Open Japanese LLM - Wiki NER SET F1 | 10.62 | Score (%) | 79.2 |
| Open LLM Leaderboard - MATH Level 5 | 22.05 | Score | 74.7 |
| Open Japanese LLM - Xlsum JA Rouge1 | 28.39 | Score (%) | 70.7 |
| Open Japanese LLM - Jsts Pearson | 89.34 | Score (%) | 65.1 |
| Open Japanese LLM - Xlsum JA Rouge2 | 10.07 | Score (%) | 59.9 |
| Open Japanese LLM - SUM | 10.05 | Score (%) | 59.4 |
| Open Japanese LLM - Xlsum JA Bert Score JA F1 | 69.3 | Score (%) | 57 |
| Open Japanese LLM - Jsts Spearman | 86.02 | Score (%) | 56.7 |
| Open Japanese LLM - Jnli Exact Match | 81.51 | Score (%) | 54.1 |
| Open Japanese LLM - Wiki Coreference SET F1 | 2.7 | Score (%) | 50 |
| Open Japanese LLM - Jcommonsenseqa Exact Match | 92.23 | Score (%) | 44.7 |
Interactive version: theaggregate.ai/model?slug=deepseek-r1-distill-qwen-14b-abliterated-v2 · How It Works · Data refreshed daily, snapshot 2026-11-08.