medgemma-finetuned-wb-with-metrics

This model is a fine-tuned version of google/medgemma-4b-it on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6743
• Rouge1: 0.8318
• Rouge2: 0.7007
• Rougel: 0.8252
• Rougelsum: 0.8316
• Bi Rads Accuracy: 0.0

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: 0.0002
• train_batch_size: 1
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 64
• total_train_batch_size: 64
• optimizer: Use OptimizerNames.PAGED_ADAMW_8BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Bi Rads Accuracy
0.8836	1.0	2	0.8327	0.8163	0.6703	0.7963	0.8162	0.0
0.6305	2.0	4	0.7656	0.8372	0.6874	0.8185	0.8373	0.0
0.7125	3.0	6	0.7090	0.8410	0.7109	0.8306	0.8397	0.0

Framework versions

• PEFT 0.18.0
• Transformers 4.57.2
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1