Hanhpt23/SwinLarge

microsoft/swin-large-patch4-window12-384-in22k

This model is a fine-tuned version of microsoft/swin-large-patch4-window12-384-in22k on the NIH-Xray dataset. It achieves the following results on the evaluation set:

• Loss: 3.7711
• Accuracy: 0.4938

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: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• 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
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.8318	0.9984	315	1.7651	0.5437
1.6067	2.0	631	1.6393	0.5455
1.406	2.9984	946	1.6472	0.5490
1.3983	4.0	1262	1.7344	0.5455
0.7272	4.9984	1577	2.1283	0.5258
0.3975	6.0	1893	2.5229	0.5134
0.2648	6.9984	2208	3.0333	0.5080
0.1232	8.0	2524	3.4626	0.5241
0.0873	8.9984	2839	3.6219	0.5027
0.0554	9.9842	3150	3.7711	0.4938

Framework versions

• Transformers 4.41.1
• Pytorch 2.3.0
• Datasets 2.19.1
• Tokenizers 0.19.1