Science Question Answering on ARC Korean (test)
92AccuracyTrinity Large
Evaluation Results
| Method | Links | |
|---|---|---|
| Trinity LargeArchitecture=MoE, Evaluation Mode=Zero-shot2026.02 | 92 | |
| Qwen3Size=8B, Evaluation Mode=Zero-shot2026.02 | 87 | |
| Qwen3Size=4B, Evaluation Mode=Zero-shot2026.02 | 81 | |
| DatologySize=8B, Evaluation Mode=Zero-shot2026.02 | 69 | |
| TrillionLabsSize=7B, Evaluation Mode=Zero-shot2026.02 | 68 | |
| DatologySize=3B, Evaluation Mode=Zero-shot2026.02 | 62 | |
| Granite-4.0Size=Micro, Evaluation Mode=Zero-shot2026.02 | 61 | |
| LFM2.5Size=1.2B, Evaluation Mode=Zero-shot2026.02 | 61 | |
| Llama-3.1Size=8B, Evaluation Mode=Zero-shot2026.02 | 59 | |
| SmolLM3Size=3B, Evaluation Mode=Zero-shot2026.02 | 57 | |
| Llama-3.2Size=3B, Evaluation Mode=Zero-shot2026.02 | 47 | |
| Llama-3.2Size=1B, Evaluation Mode=Zero-shot2026.02 | 25 |