PTS-Bart-Large-CNN

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

• Loss: 1.1760
• Rouge1: 0.6551
• Rouge2: 0.4332
• Rougel: 0.5543
• Rougelsum: 0.5541
• Gen Len: 80.0886

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-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: 8
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	220	0.8239	0.6263	0.3973	0.5238	0.5237	84.2023
No log	2.0	440	0.8201	0.6461	0.4184	0.5417	0.5416	81.1659
0.7121	3.0	660	0.8661	0.6479	0.4226	0.5448	0.5454	80.5409
0.7121	4.0	880	0.9784	0.6474	0.4242	0.5424	0.5425	82.2932
0.2619	5.0	1100	1.0645	0.655	0.4327	0.5517	0.5517	80.8386
0.2619	6.0	1320	1.1098	0.6548	0.4339	0.5542	0.5543	81.3545
0.1124	7.0	1540	1.1528	0.6528	0.4298	0.5511	0.551	80.5705
0.1124	8.0	1760	1.1760	0.6551	0.4332	0.5543	0.5541	80.0886

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1