Information Extraction on FDA
84.5AccuracyApriel-1.6
Evaluation Results
| Method | Links | |
|---|---|---|
| Apriel-1.6Speedup @32k=1.0x, Speedup @16k=1.0x2026.04 | 84.5 | |
| Qwen-3.5 27BSpeedup @32k=0.55x, Speedup @16k=0.5x2026.04 | 84.4 | |
| Nemotron-Nano 12B v2Speedup @32k=5.85x, Speedup @16k=4.3x2026.04 | 82.5 | |
| Reg|Lklhd–18Speedup @32k=4.76x, Speedup @16k=2.2x2026.04 | 81.7 | |
| S1: Distil. Idealized|All–6Speedup @32k=6.13x, Speedup @16k=2.5x2026.04 | 80.9 | |
| Reg|Lklhd–26Speedup @32k=2.85x, Speedup @16k=1.5x2026.04 | 79.9 | |
| OLMo-Hybrid-Think 7BSpeedup @32k=2.51x, Speedup @16k=2.1x2026.04 | 78.5 | |
| all-attentionSpeedup @32k=1.0x, Speedup @16k=1.0x2026.04 | 78.3 | |
| Idealized|All–18Speedup @32k=1.99x, Speedup @16k=1.1x2026.04 | 78 | |
| Idealized|Lklhd–6Speedup @32k=6.2x, Speedup @16k=2.4x2026.04 | 76.9 | |
| Idealized|All–6Speedup @32k=6.13x, Speedup @16k=2.5x2026.04 | 75.6 | |
| Nemotron-3-Nano 30BSpeedup @32k=4.09x, Speedup @16k=2.8x2026.04 | 74.7 | |
| Apriel-H1 15BSpeedup @32k=1.97x, Speedup @16k=1.9x2026.04 | 73.2 | |
| Reg|Lklhd–13Speedup @32k=6.9x, Speedup @16k=2.7x2026.04 | 68.6 | |
| Reg|Lklhd–10Speedup @32k=10.69x, Speedup @16k=4.2x2026.04 | 65.2 | |
| Falcon-H1R 7BSpeedup @32k=4.61x, Speedup @16k=3.4x2026.04 | 63.6 | |
| SATFORMERInput length=2k tokens, Evaluation protocol=Arora et al. [2024]2026.05 | 21.96 | |
| MUDDFORMERInput length=2k tokens, Evaluation protocol=Arora et al. [2024]2026.05 | 20.32 | |
| RESFORMERInput length=2k tokens, Evaluation protocol=Arora et al. [2024]2026.05 | 19.41 | |
| TRANSFORMERInput length=2k tokens, Evaluation protocol=Arora et al. [2024]2026.05 | 19.14 | |
| DENSEFORMERInput length=2k tokens, Evaluation protocol=Arora et al. [2024]2026.05 | 16.42 | |
| HYPERCONNECTIONSInput length=2k tokens, Evaluation protocol=Arora et al. [2024]2026.05 | 15.78 |