pegasus-samsum

This model is a fine-tuned version of google/pegasus-large on the samsum dataset. It achieves the following results on the evaluation set:

• Loss: 1.4091
• Rouge1: 0.4659
• Rouge2: 0.2345
• Rougel: 0.3946
• Rougelsum: 0.3951
• Gen Len: 17.7467

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
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
1.8025	0.27	500	1.4403	0.4466	0.2101	0.3832	0.3841	21.64
1.5936	0.54	1000	1.3766	0.4786	0.2374	0.4017	0.4013	21.24
1.5926	0.81	1500	1.3910	0.5118	0.2643	0.4282	0.4286	20.2267
1.5067	1.09	2000	1.4028	0.4982	0.261	0.4155	0.4157	20.4267
1.5712	1.36	2500	1.4236	0.4712	0.234	0.3964	0.3969	17.0
1.6177	1.63	3000	1.4151	0.4768	0.2382	0.4019	0.4022	16.28
1.6289	1.9	3500	1.4112	0.4744	0.2346	0.402	0.4033	17.0267
1.6326	2.17	4000	1.4096	0.4682	0.234	0.3985	0.3994	17.1333
1.5929	2.44	4500	1.4093	0.4637	0.2342	0.3939	0.3942	17.16
1.4351	2.72	5000	1.4090	0.4684	0.2346	0.3953	0.3955	17.8133
1.6445	2.99	5500	1.4091	0.4659	0.2345	0.3946	0.3951	17.7467

Framework versions

• Transformers 4.33.1
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3