pegasuscnn-dailymail_billsum_model

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

• Loss: 1.6747
• Rouge1: 0.4804
• Rouge2: 0.2362
• Rougel: 0.3218
• Rougelsum: 0.3218
• Gen Len: 123.3669

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: 5
• eval_batch_size: 5
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.6227	1.0	198	1.9091	0.4289	0.1938	0.2945	0.2947	120.1855
1.9714	2.0	396	1.8147	0.4517	0.2093	0.3059	0.3061	120.7742
1.903	3.0	594	1.7646	0.4607	0.2207	0.3098	0.3102	121.121
1.7973	4.0	792	1.7362	0.4719	0.2264	0.3179	0.3178	122.3185
1.7868	5.0	990	1.7137	0.4779	0.2314	0.3191	0.3192	123.2379
1.7457	6.0	1188	1.6958	0.4748	0.2296	0.3171	0.317	123.2056
1.6687	7.0	1386	1.6873	0.4795	0.2352	0.3216	0.3216	123.2702
1.6751	8.0	1584	1.6806	0.4835	0.2384	0.3248	0.3245	122.8266
1.6564	9.0	1782	1.6758	0.4814	0.2359	0.3217	0.3216	123.2984
1.6333	10.0	1980	1.6747	0.4804	0.2362	0.3218	0.3218	123.3669

Framework versions

• Transformers 4.34.1
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1