LED_billsum_model

This model is a fine-tuned version of allenai/led-base-16384 on the billsum dataset. It achieves the following results on the evaluation set:

• Loss: 1.6576
• Rouge1: 0.1447
• Rouge2: 0.0854
• Rougel: 0.1292
• Rougelsum: 0.1339
• Gen Len: 20.0

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: 3
• eval_batch_size: 3
• 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
1.4849	1.0	330	1.6511	0.1463	0.0827	0.1276	0.1337	20.0
1.3361	2.0	660	1.6056	0.148	0.0799	0.1268	0.1336	20.0
1.1727	3.0	990	1.5833	0.1459	0.0827	0.1289	0.1341	20.0
1.0601	4.0	1320	1.5987	0.1462	0.0859	0.1299	0.1344	20.0
0.9789	5.0	1650	1.6030	0.1414	0.0794	0.125	0.1302	20.0
0.8724	6.0	1980	1.6060	0.1476	0.0868	0.1298	0.1356	20.0
0.7994	7.0	2310	1.6295	0.1348	0.0758	0.1198	0.1253	20.0
0.7762	8.0	2640	1.6317	0.1422	0.0831	0.1261	0.1312	20.0
0.7087	9.0	2970	1.6501	0.1421	0.0825	0.1264	0.1311	20.0
0.7014	10.0	3300	1.6576	0.1447	0.0854	0.1292	0.1339	20.0

Framework versions

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