Reasoning on HellaSwag, GSM8K, and Psychometric Psi
91.1HellaSwag ScoreDictaLM-3.0-Thinking
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DictaLM-3.0-ThinkingTraining Stage=SFT2026.05 | 91.1 | 86.1 | 95.3 | 90.8 | |
| Gemma-3-27B-ITTraining Stage=SFT2026.05 | 89.8 | 91.7 | 94.5 | 92 | |
| Nemotron-3-NanoLang=EN2026.05 | 85.6 | 92.3 | 88.7 | 88.9 | |
| Gemma-3-27BLang=EN2026.05 | 85.6 | 82.6 | 94.7 | 87.6 | |
| HebatronTraining Stage=SFT2026.05 | 82.5 | 83.8 | 91.6 | 86 | |
| HebatronLang=EN2026.05 | 65.6 | 83.5 | 88.2 | 79.1 | |
| DictaLM-3.0Lang=EN2026.05 | 56.9 | 84.1 | 89.7 | 76.9 |