ft-t5-with-dill-sum

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

• Loss: 2.3109
• Rouge1: 0.1886
• Rouge2: 0.104
• Rougel: 0.166
• Rougelsum: 0.1659
• 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: 5e-05
• train_batch_size: 8
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.5462	1.0	31	2.4185	0.187	0.1023	0.1637	0.1639	19.0
2.5478	2.0	62	2.4166	0.187	0.1018	0.1637	0.1639	19.0
2.5729	3.0	93	2.4114	0.1868	0.1015	0.1637	0.1638	19.0
2.5806	4.0	124	2.4072	0.1855	0.1006	0.1626	0.1627	19.0
2.5231	5.0	155	2.4025	0.1877	0.1042	0.165	0.165	19.0
2.5245	6.0	186	2.3948	0.1869	0.1024	0.1642	0.1642	19.0
2.5273	7.0	217	2.3860	0.1886	0.1032	0.1652	0.1653	19.0
2.4941	8.0	248	2.3765	0.188	0.1033	0.1649	0.165	19.0
2.4612	9.0	279	2.3698	0.19	0.1057	0.1671	0.1671	19.0
2.463	10.0	310	2.3578	0.1882	0.1039	0.1662	0.1663	19.0
2.4539	11.0	341	2.3491	0.1898	0.1057	0.1667	0.1667	19.0
2.441	12.0	372	2.3392	0.1901	0.1055	0.1669	0.1668	19.0
2.4389	13.0	403	2.3292	0.1893	0.1053	0.1666	0.1665	19.0
2.3945	14.0	434	2.3203	0.1903	0.1051	0.1676	0.1675	19.0
2.4148	15.0	465	2.3109	0.1886	0.104	0.166	0.1659	19.0

Framework versions

• Transformers 4.41.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1