bart-base-summarization-medical-51

This model is a fine-tuned version of facebook/bart-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.1339
• Rouge1: 0.4189
• Rouge2: 0.2237
• Rougel: 0.356
• Rougelsum: 0.3558
• Gen Len: 18.191

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: 3e-05
• train_batch_size: 4
• eval_batch_size: 1
• seed: 51
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 6
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.7175	1.0	1250	2.2013	0.4107	0.219	0.35	0.3496	17.772
2.5872	2.0	2500	2.1674	0.4158	0.2228	0.3536	0.3531	18.005
2.5979	3.0	3750	2.1502	0.415	0.2215	0.351	0.3509	18.111
2.5579	4.0	5000	2.1428	0.4178	0.226	0.3548	0.3549	18.143
2.5332	5.0	6250	2.1350	0.4172	0.2243	0.3546	0.3542	18.149
2.5331	6.0	7500	2.1339	0.4189	0.2237	0.356	0.3558	18.191

Framework versions

• PEFT 0.12.0
• Transformers 4.42.4
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1