Natural Language Understanding on KoBEST
93.3BoolQ ScoreEXAONE-2.4B + SCRIPT
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| EXAONE-2.4B + SCRIPTBackbone=EXAONE-2.4B, Enhancement=SCRIPT2026.04 | 93.3 | 93.3 | 82.46 | 86 | 94.96 | — | |
| EXAONE-2.4BBackbone=EXAONE-2.4B2026.04 | 92.59 | 93.3 | 82.14 | 85.6 | 94.21 | — | |
| KoGPT3-1.2B + SCRIPTBackbone=KoGPT3-1.2B, Enhancement=SCRIPT2026.04 | 77.63 | 82.8 | 74.65 | 79.3 | 96.48 | — | |
| KoGPT3-1.2B + SCRIPTModel Type=Monolingual, Integrated Method=SCRIPT2026.04 | 77.63 | 82.8 | 74.65 | 79.3 | 96.48 | 82.17 | |
| KoGPT3-1.2BBackbone=KoGPT3-1.2B2026.04 | 77.32 | 82.8 | 72.78 | 78.9 | 96.31 | — | |
| KoGPT3-1.2BModel Type=Monolingual2026.04 | 77.32 | 82.8 | 72.78 | 78.9 | 96.31 | 81.62 | |
| mGPT-1.3BModel Type=Multilingual2026.04 | 71.19 | 69.7 | 68.38 | 76.1 | 89.4 | 74.95 | |
| mGPT-1.3B + SCRIPTModel Type=Multilingual, Integrated Method=SCRIPT2026.04 | 70.72 | 70.6 | 69.17 | 76.7 | 90.42 | 75.52 | |
| KoGPT2base + SCRIPTBackbone=KoGPT2base, Enhancement=SCRIPT2026.04 | 68.28 | 70.9 | 68.18 | 72.4 | 89.47 | — | |
| KoGPT2baseBackbone=KoGPT2base2026.04 | 67.22 | 68.9 | 67.07 | 69.1 | 88.5 | — | |
| BERTbase + SCRIPTBackbone=BERTbase, Enhancement=SCRIPT2026.04 | 62.32 | 61.3 | 74.3 | 64.4 | 83.38 | — | |
| KOMBObaseBackbone=KOMBObase2026.04 | 61.4 | 61 | 68.91 | 63.8 | 79.07 | — | |
| BERTbaseBackbone=BERTbase2026.04 | 60.75 | 60.9 | 73.14 | 63.2 | 83.12 | — |