mle-policy-multiwoz21

This is a MLE model trained on MultiWOZ 2.1

Refer to ConvLab-3 for model description and usage.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 32
• seed: 0
• optimizer: Adam
• num_epochs: 24
• use checkpoint which performed best on validation set

Framework versions

• Transformers 4.18.0
• Pytorch 1.10.2+cu111