BMU_Finetuned_BIO-BART_MedQuad

This model is a fine-tuned version of GanjinZero/biobart-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.5404
• Rouge1: 36.82
• Rouge2: 20.39
• Rougel: 28.62
• Rougelsum: 34.96
• Meteor: 25.05
• Bertscore Precision: 87.36
• Bertscore Recall: 84.71
• Bertscore F1: 85.96

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: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Meteor	Bertscore Precision	Bertscore Recall	Bertscore F1
1.8272	1.0	411	1.7085	36.22	20.21	28.27	34.39	24.53	87.51	84.49	85.92
1.7387	2.0	822	1.6161	35.25	19.72	27.77	33.47	23.51	87.65	84.37	85.92
1.6207	3.0	1233	1.5703	36.3	20.59	28.7	34.58	24.61	87.57	84.55	85.98
1.5583	4.0	1644	1.5459	36.69	20.47	28.63	34.79	24.95	87.34	84.68	85.93
1.5529	4.9896	2050	1.5404	36.82	20.39	28.62	34.96	25.05	87.36	84.71	85.96

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.3.1
• Tokenizers 0.21.0