Multitask Thai Language Evaluation on Strict Thai Evaluation Suite
7.89MTB Chat ScoreTyphoon-S-8B
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Typhoon-S-8Bbase_model=ThaiLLM-8B, recipe=SFT+OPD (full-logits)2026.01 | 7.89 | 96.6 | 76.45 | 67.06 | 37 | 71.2 | |
| Qwen3-8Btype=Instruction-tuned baseline2026.01 | 7.08 | 95.4 | 80.47 | 63.66 | 23 | 66.66 |