3_loa

This model is a fine-tuned version of google/flan-t5-large on the billsum dataset. It achieves the following results on the evaluation set:

• Loss: 1.4825
• Rouge1: 0.201
• Rouge2: 0.1132
• Rougel: 0.1753
• Rougelsum: 0.1755
• Gen Len: 19.0

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: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.1079	1.0	989	1.6673	0.2028	0.1092	0.1748	0.1751	19.0
1.8481	2.0	1978	1.6150	0.1979	0.1061	0.1715	0.1717	19.0
1.7889	3.0	2967	1.5833	0.1994	0.11	0.1727	0.1727	19.0
1.7319	4.0	3956	1.5584	0.1978	0.1084	0.1718	0.1718	19.0
1.7279	5.0	4945	1.5440	0.2016	0.1106	0.1755	0.1756	19.0
1.7386	6.0	5934	1.5326	0.1991	0.1086	0.1734	0.1736	19.0
1.6972	7.0	6923	1.5251	0.2013	0.1122	0.1759	0.176	19.0
1.6732	8.0	7912	1.5145	0.2024	0.1123	0.1766	0.1766	19.0
1.6597	9.0	8901	1.5079	0.2019	0.1125	0.1751	0.1753	19.0
1.6151	10.0	9890	1.5045	0.201	0.1123	0.1758	0.1761	19.0
1.6381	11.0	10879	1.4997	0.2009	0.1116	0.1755	0.1756	19.0
1.6148	12.0	11868	1.4974	0.2018	0.1133	0.1763	0.1765	19.0
1.6196	13.0	12857	1.4940	0.2014	0.1129	0.1756	0.1756	19.0
1.6137	14.0	13846	1.4914	0.2025	0.1136	0.1766	0.1768	19.0
1.6313	15.0	14835	1.4873	0.2032	0.114	0.1769	0.1771	19.0
1.6098	16.0	15824	1.4847	0.2012	0.1133	0.175	0.1754	19.0
1.6061	17.0	16813	1.4845	0.2019	0.1138	0.1752	0.1755	19.0
1.5918	18.0	17802	1.4833	0.2011	0.1129	0.1747	0.175	19.0
1.5842	19.0	18791	1.4824	0.2013	0.1133	0.1753	0.1755	19.0
1.5964	20.0	19780	1.4825	0.201	0.1132	0.1753	0.1755	19.0

Framework versions

• Transformers 4.31.0
• Pytorch 1.13.1.post200
• Datasets 2.10.0
• Tokenizers 0.13.2