Model
QuartzNet PyTorch checkpoint trained on LibriSpeech (test-other 10.41% WER)
Use the NGC CLI to download:
Copied!
1 Version
21.03.0_ampSelected09/22/2022 5:24 PM UTC72.72 MBAccuracy: 0260 EpochsBatch Size: 1152GPU: A100 Copied!
21.03.0_ampSelected
09/22/2022 5:24 PM UTC72.72 MBAccuracy: 0260 EpochsBatch Size: 1152GPU: A100
Copied!
architecture
| Key | Value |
|---|---|
| type | quartznet10x5_speedp-online-1.15_speca: SpecAugment (2x15, 2x55); Speed perturbation U(0.85, 1.15) |
performance
| Key | Value |
|---|---|
| test_other_wer | 10.41 |
| dev_other_wer | 10.67 |
| dev_clean_wer | 3.51 |
| test_clean_wer | 3.82 |
training
| Key | Value |
|---|---|
| batch_size | 1152 |
| training_precision | AMP |
| dataset | LibriSpeech |
| epochs | 260 |