Question Answering on TriviaQA (accuracy %)
28AccuracySmolLM2 360M
Evaluation results
| Method | Links | |
|---|---|---|
| SmolLM2 360MTraining tokens=4T, Evaluation Protocol=0-shot2026.04 | 28 | |
| Qwen2.5 0.5BTraining tokens=18T, Evaluation Protocol=0-shot2026.04 | 27.8 | |
| Granite 350MTraining tokens=10–12T, Evaluation Protocol=0-shot2026.04 | 25.2 | |
| Nautile-370MTraining tokens=∼0.8T, Evaluation Protocol=0-shot2026.04 | 23.8 | |
| LFM2.5 350MTraining tokens=28T, Evaluation Protocol=0-shot2026.04 | 22.9 |