bart-large-cnn-billsum

This model is a fine-tuned version of facebook/bart-large-cnn on the billsum dataset. It achieves the following results on the evaluation set:

• Loss: 1.7658
• Rouge1: 0.5014
• Rouge2: 0.2463
• Rougel: 0.3189
• Rougelsum: 0.3752
• Gen Len: 125.5645

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: 2e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	248	1.8112	0.4809	0.2299	0.3067	0.3716	113.1371
No log	2.0	496	1.7501	0.5089	0.2484	0.325	0.3844	123.9435
1.7258	3.0	744	1.7386	0.5008	0.2412	0.3163	0.3732	127.2056
1.7258	4.0	992	1.7658	0.5014	0.2463	0.3189	0.3752	125.5645

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.1+cu116
• Datasets 2.8.0
• Tokenizers 0.13.2