gpt2-medium-nlg-multiwoz21

This model is a fine-tuned version of GPT2-medium on MultiWOZ21.

Refer to ConvLab-3 for model description and usage.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-5
• train_batch_size: 64
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• optimizer: AdamW
• lr_scheduler_type: linear
• num_epochs: 20

Framework versions

• Transformers 4.23.1
• Pytorch 1.10.1+cu111