Named Entity Recognition on CrossNER Held out AI Literature Music Politics Science
60.35Average F1Agents-K1 (4B)
Evaluation results
| Method | Links | |
|---|---|---|
| Agents-K1 (4B)Model size=4B, Training=GRPO on IEPile, Fine-tuning=False2026.06 | 60.35 | |
| Qwen3-32BModel size=32B, Training=Base checkpoint, Fine-tuning=False2026.06 | 60.06 | |
| Qwen3-8BModel size=8B, Training=Base checkpoint, Fine-tuning=False2026.06 | 57.73 | |
| Qwen3-4BModel size=4B, Training=Base checkpoint, Fine-tuning=False2026.06 | 57.31 |