Automatic Speech Recognition on ContextASR-Speech-En
7.26Plain WERFP16
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FP16Backbone=Q3-1.7B, Calibration=LS-clean2026.05 | 7.26 | 13.57 | |
| TARQBackbone=Q3-1.7B, Calibration=LS-clean2026.05 | 8.77 | 17.56 | |
| GPTQBackbone=Q3-1.7B, Calibration=LS-clean2026.05 | 8.89 | 17.67 | |
| OmniQuantBackbone=Q3-1.7B, Calibration=LS-clean2026.05 | 10.24 | 18.59 | |
| RTNBackbone=Q3-1.7B, Calibration=LS-clean2026.05 | 10.62 | 19.69 | |
| AWQBackbone=Q3-1.7B, Calibration=LS-clean2026.05 | 11.33 | 17.67 | |
| TARQBackbone=W-base, Calibration=LS-clean2026.05 | 45.32 | 68.7 | |
| FP16Backbone=W-base, Calibration=LS-clean2026.05 | 45.74 | 67.54 | |
| GPTQBackbone=W-base, Calibration=LS-clean2026.05 | 46.42 | 69.6 | |
| AWQBackbone=W-base, Calibration=LS-clean2026.05 | 49.02 | 72.36 | |
| RTNBackbone=W-base, Calibration=LS-clean2026.05 | 50.48 | 72.41 | |
| OmniQuantBackbone=W-base, Calibration=LS-clean2026.05 | 51.23 | 72.92 |