T3Q-Qwen2.5-7B-it-KOR-Safe: benchmark results
kayfour's community fine-tune of Qwen2.5-7B-Instruct published as a Korean safety-tuned T3Q variant. Provider: Other. Released 2024-11-11. Access: Open.
Aggregate Rating 1551 ± 40, rank #1691 of 10916 rated models, from 15 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-IFEval | 63.81 | Strict accuracy, prompt/instruction mean (%) | 97.8 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 67.1 | Exact match, flexible extract (%) | 96.6 |
| Open Korean LLM - KorNAT-Helpful | 54.2 | Normalized accuracy (%) | 95 |
| Open Korean LLM - Ko-EQ-Bench | 47.77 | EQ-Bench score | 94.4 |
| Open Korean LLM - KorNAT-Harmless | 71.56 | Normalized accuracy (%) | 94.2 |
| Open Korean LLM Leaderboard | 46.78 | Average Score (%) | 94.1 |
| Open LLM Leaderboard - MATH Level 5 | 37.61 | Score | 88 |
| Open Korean LLM - KorNAT-CKA | 30.38 | Normalized accuracy (%) | 83 |
| Open LLM Leaderboard - MMLU-Pro | 44.64 | Score | 82.8 |
| Open LLM Leaderboard - GPQA | 32.13 | Score | 76 |
| Open LLM Leaderboard - BBH | 55.5 | Score | 75.7 |
| Open LLM Leaderboard - IFEval | 60.81 | Score | 72.9 |
Interactive version: theaggregate.ai/model?slug=t3q-qwen2-5-7b-it-kor-safe · How It Works · Data refreshed daily, snapshot 2026-11-08.