Training configurations for the text correction module on different data sets
| Model | BART | FastCorrect | ||
|---|---|---|---|---|
| Data set | AISHELL-1 | TEDLIUM-2 | LibriSpeech | AISHELL-1 |
| Epochs | 10 | 10 | 10 | 25 |
| Batch size | 128 | 64 | 64 | – |
| Max tokens | – | – | – | 9,000 |
| Gradient accumulation | 2 | 4 | 4 | 4 |
| Learning rate | 1e-6 | 1e-6 | 1e-6 | 5e-4 |
| Warm up Ratio | 0.3 | 0.3 | 0.3 | – |
| Model | FastCorrect | |||
|---|---|---|---|---|
| Data set | AISHELL-1 | TEDLIUM-2 | LibriSpeech | AISHELL-1 |
| Epochs | 10 | 10 | 10 | 25 |
| Batch size | 128 | 64 | 64 | – |
| Max tokens | – | – | – | 9,000 |
| Gradient accumulation | 2 | 4 | 4 | 4 |
| Learning rate | 1e-6 | 1e-6 | 1e-6 | 5e-4 |
| Warm up Ratio | 0.3 | 0.3 | 0.3 | – |
Sharing content requires targeting cookies to be enabled. Please update your cookie preferences to use this feature.