bart-base-lora-summarization-medical

This model is a fine-tuned version of facebook/bart-base on the 'mystic-leung/medical_cord19' dataset. It achieves the following results on the evaluation set:

• Loss: 2.4119
• Rouge1: 0.4304
• Rouge2: 0.2352
• Rougel: 0.3663
• Rougelsum: 0.3660
• Gen Len: 18.1767

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: 42
• 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: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
2.5079	1.0	6250	2.121959	0.4263	0.2290	0.3597	0.3594	18.3300
2.4566	2.0	12500	2.084411	0.4267	0.2312	0.3622	0.3618	18.2773
2.4242	3.0	18750	2.061557	0.4311	0.2358	0.3660	0.3656	18.1307
2.4058	4.0	25000	2.053182	0.4316	0.2367	0.3660	0.3659	18.1753
2.4119	5.0	31250	2.052128	0.4304	0.2352	0.3663	0.3660	18.1767

Framework versions

• PEFT 0.10.0
• Transformers 4.40.1
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1