Multilingual GRPO on Thai
42.06Best-k AccuracyOne Step+KMM
Evaluation Results
| Method | Links | |
|---|---|---|
| One Step+KMMEvaluation setting=GRPO, Random seeds=52026.05 | 42.06 | |
| TV+KMMEvaluation setting=GRPO, Random seeds=52026.05 | 40.27 | |
| TVEvaluation setting=GRPO, Random seeds=52026.05 | 38.62 | |
| One StepEvaluation setting=GRPO, Random seeds=52026.05 | 37.8 | |
| RandomEvaluation setting=GRPO, Random seeds=52026.05 | 37.59 |