MTSUFall2024SoftwareEngineering

This model is a fine-tuned version of google-t5/t5-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.7579
• Rouge1: 0.268
• Rouge2: 0.2083
• Rougel: 0.258
• Rougelsum: 0.2582
• Gen Len: 18.9805

Model description

This model is a fine-tuned Google T5-Small model that is fine-tuned to summarize United States Senate and House Bills.

Intended uses & limitations

Summarize United States Federal Legislation.

Training and evaluation data

Trained on ~51.9k bills and summaries.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 14
• eval_batch_size: 14
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.1182	1.0	3708	1.8807	0.2643	0.2029	0.2533	0.2534	18.9817
1.999	2.0	7416	1.8013	0.2663	0.2053	0.2558	0.2559	18.9833
1.9739	3.0	11124	1.7681	0.267	0.2066	0.2568	0.2569	18.9816
1.9448	4.0	14832	1.7579	0.268	0.2083	0.258	0.2582	18.9805

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.19.1