Why WRate is high but ERate is low?

[yzli9]

I have trained 25 epoch, the WRate reached 0.778 but ERate is only 0.010 ?
Below is my configuration:
lr: 0.9
type: l2
batch_size: 8
workers: 4
use_flag: True
context_loss: True
word_state_loss: True

[liuzhuang1024]

I have trained 25 epoch, the WRate reached 0.778 but ERate is only 0.010 ? Below is my configuration: lr: 0.9 type: l2 batch_size: 8 workers: 4 use_flag: True context_loss: True word_state_loss: True

Thank you for your attention. You can follow your advice and conduct the following steps:

1. Disable the majority of features and revert to the CAN or pure WAP methods.
2. Retrain the model and assess the metrics for any improvements.
3. If the metrics remain low, you need thoroughly review our training configuration to identify any potential errors, ensuring data quality and addressing any other issues that may be affecting performance.