pep_summarization

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

• Loss: 0.1242
• Rouge1: 75.3806
• Rouge2: 74.6735
• Rougel: 75.5866
• Rougelsum: 75.5446
• Gen Len: 85.3188

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: 1e-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: 5.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	69	0.0957	72.6601	71.6824	72.6858	72.4668	95.4493
No log	2.0	138	0.1345	75.0063	74.0782	75.0597	74.8943	92.0145
No log	3.0	207	0.1412	75.3012	74.5492	75.4246	75.324	85.4638
No log	4.0	276	0.1089	74.8426	74.0317	74.8939	74.8128	85.0435
No log	5.0	345	0.1242	75.3806	74.6735	75.5866	75.5446	85.3188

Framework versions

• Transformers 4.38.0.dev0
• Pytorch 2.1.2+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0