BART_large_CNN_GNAD

This model is a fine-tuned version of Einmalumdiewelt/BART_large_CNN_GNAD on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.9761
• Rouge1: 27.0918
• Rouge2: 7.9818
• Rougel: 17.7781
• Rougelsum: 22.6727
• Gen Len: 96.0567

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10.0

Training results

Framework versions

• Transformers 4.22.0.dev0
• Pytorch 1.12.0+cu113
• Datasets 2.4.0
• Tokenizers 0.12.1