Josiefied-Qwen2.5-7B-Instruct-abliterated-v2: benchmark results
Goekdeniz-Guelmez's abliterated and re-fine-tuned Qwen2.5 7B Instruct from the JOSIEFIED series, built for refusal-free responses. Provider: Goekdeniz-Guelmez. Released 2024-09-20. Access: Open.
Aggregate Rating 1512 ± 17, rank #747 of 1656 rated models, from 144 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Portuguese LLM - FaQuAD NLI | 85.38 | Macro F1 (%) | 98.6 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 68.99 | Exact match, flexible extract (%) | 97.6 |
| Open Korean LLM - Ko-IFEval | 61.9 | Strict accuracy, prompt/instruction mean (%) | 97.1 |
| Open Korean LLM Leaderboard | 48.09 | Average Score (%) | 95.7 |
| Open Ko-LLM Leaderboard | 48.09 | Average (%) | 95.6 |
| Open LLM Leaderboard - IFEval | 78.41 | Score | 95 |
| Open LLM Leaderboard - MATH Level 5 | 47.21 | Score | 95 |
| Open Korean LLM - KorNAT-Helpful | 54.08 | Normalized accuracy (%) | 94.8 |
| Open Portuguese LLM - ASSIN2 RTE | 94.04 | Macro F1 (%) | 94.2 |
| Open Korean LLM - Ko-EQ-Bench | 39.72 | EQ-Bench score | 90.9 |
| Open Korean LLM - KorNAT-CKA | 31.01 | Normalized accuracy (%) | 85.8 |
| Open Portuguese LLM - ENEM | 72.57 | Accuracy (%) | 83.5 |
Interactive version: theaggregate.ai/model?slug=josiefied-qwen2-5-7b-instruct-abliterated-v2 · How It Works · Data refreshed daily, snapshot 2026-10-20.