Benchmark Summary
ONNX q8
| Suite | F1 | Examples/s |
|---|---|---|
| Irish core | 0.9733 | 29.4666 |
| Edge | 0.9500 | 28.0216 |
| Finance | 1.0000 | 30.7139 |
| Finance boundary | 1.0000 | 29.9458 |
| User PPSN | - | - |
| GA weak PPSN | 1.0000 | 11.5878 |
| Multilingual PPSN | 0.9371 | 38.4199 |
| Hardening exact | 1.0000 | 22.2860 |
| QA feedback exact | 0.8750 | 64.8216 |
| UAT replay exact | 0.9032 | 63.7971 |
Full checkpoint
| Suite | F1 | Examples/s |
|---|---|---|
| Irish core | 0.9733 | 43.9580 |
| Edge | - | - |
| Multilingual PPSN | - | - |
| Hardening exact | - | - |
QA feedback exact, strict boundary (IoU 1.0) |
0.6875 | - |