T3Q-ko-solar-dpo-v3.0: benchmark results
chihoonlee10's T3Q-ko-solar-dpo v3.0, a further DPO fine-tune of davidkim205's Korean nox-solar-10.7B-v4 model, developed with T3Q. Provider: T3Q. Released 2024-03-20. Access: Open.
Aggregate Rating 1387 ± 19, rank #1266 of 1656 rated models, from 19 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-Winogrande | 75.14 | Accuracy (%) | 99.6 |
| Open Korean LLM - KorNAT-Harmless | 81.86 | Normalized accuracy (%) | 99.3 |
| Open Korean LLM - KorNAT-Helpful | 72.05 | Normalized accuracy (%) | 99.2 |
| Open Korean LLM - Ko-GPQA-Diamond | 27.78 | Normalized accuracy (%) | 95 |
| Open Korean LLM - KorNAT-CKA | 26.7 | Normalized accuracy (%) | 66.2 |
| Open Ko-LLM Leaderboard | 38.19 | Average (%) | 65.2 |
| Open Korean LLM Leaderboard | 38.19 | Average Score (%) | 65.2 |
| Open Korean LLM - Ko-GSM8K (flexible extract) | 16.98 | Exact match, flexible extract (%) | 47.6 |
| LogicKor - Grammar | 4.64 | Score (0-10) | 42.4 |
| Open Korean LLM - Ko-IFEval | 25.42 | Strict accuracy, prompt/instruction mean (%) | 35.3 |
| LogicKor - Reasoning | 4.42 | Score (0-10) | 29.2 |
| LogicKor - Understanding | 6 | Score (0-10) | 25.6 |
Interactive version: theaggregate.ai/model?slug=t3q-ko-solar-dpo-v3-0 · How It Works · Data refreshed daily, snapshot 2026-10-20.