organcmnist-swin-base-finetuned

This model is a fine-tuned version of microsoft/swin-large-patch4-window7-224-in22k on the medmnist-v2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2582
• Accuracy: 0.9317
• Precision: 0.9295
• Recall: 0.9177
• F1: 0.9229

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: 0.005
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.7563	0.9988	203	0.1859	0.9365	0.9432	0.9127	0.9201
0.6145	1.9975	406	0.1260	0.9640	0.9630	0.9608	0.9600
0.6476	2.9963	609	0.0926	0.9774	0.9715	0.9754	0.9723
0.5719	4.0	813	0.0912	0.9770	0.9749	0.9746	0.9740
0.5374	4.9988	1016	0.1281	0.9695	0.9730	0.9690	0.9699
0.5615	5.9975	1219	0.1088	0.9791	0.9839	0.9819	0.9825
0.4959	6.9963	1422	0.1134	0.9741	0.9812	0.9742	0.9768
0.425	8.0	1626	0.1016	0.9808	0.9816	0.9820	0.9815
0.3151	8.9988	1829	0.1368	0.9804	0.9843	0.9832	0.9834
0.3347	9.9877	2030	0.1156	0.9837	0.9853	0.9864	0.9856

Framework versions

• PEFT 0.11.1
• Transformers 4.40.2
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1