rsna-intracranial-hemorrhage-detection

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.2164
• Accuracy: 0.6152

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: 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
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.5655	1.0	238	1.5235	0.4039
1.3848	2.0	477	1.3622	0.4692
1.2812	3.0	716	1.2811	0.5150
1.2039	4.0	955	1.1795	0.5556
1.1641	5.0	1193	1.1627	0.5534
1.1961	6.0	1432	1.1393	0.5705
1.1382	7.0	1671	1.0921	0.5804
0.9653	8.0	1910	1.0790	0.5876
0.9346	9.0	2148	1.0727	0.5931
0.9083	10.0	2387	1.0605	0.5994
0.8936	11.0	2626	1.0147	0.6146
0.8504	12.0	2865	1.0849	0.5818
0.8544	13.0	3103	1.0349	0.6052
0.7884	14.0	3342	1.0435	0.6074
0.7974	15.0	3581	1.0082	0.6127
0.7921	16.0	3820	1.0438	0.6017
0.709	17.0	4058	1.0484	0.6094
0.6646	18.0	4297	1.0554	0.6221
0.6832	19.0	4536	1.0455	0.6124
0.7076	20.0	4775	1.0905	0.6
0.7442	21.0	5013	1.1094	0.6008
0.6332	22.0	5252	1.0777	0.6063
0.6417	23.0	5491	1.0765	0.6141
0.6267	24.0	5730	1.1057	0.6091
0.6082	25.0	5968	1.0962	0.6171
0.6191	26.0	6207	1.1178	0.6039
0.5654	27.0	6446	1.1386	0.5948
0.5776	28.0	6685	1.1121	0.6105
0.5531	29.0	6923	1.1497	0.6030
0.6275	30.0	7162	1.1796	0.6028
0.5373	31.0	7401	1.1306	0.6132
0.4775	32.0	7640	1.1523	0.6058
0.5469	33.0	7878	1.1634	0.6127
0.4934	34.0	8117	1.1853	0.616
0.5233	35.0	8356	1.2018	0.6055
0.4896	36.0	8595	1.1585	0.6108
0.5122	37.0	8833	1.1874	0.6146
0.4726	38.0	9072	1.1608	0.6193
0.4372	39.0	9311	1.2403	0.6132
0.498	40.0	9550	1.1752	0.6201
0.4813	41.0	9788	1.2005	0.6166
0.4762	42.0	10027	1.2285	0.6022
0.4852	43.0	10266	1.2192	0.6119
0.4332	44.0	10505	1.2391	0.6218
0.3998	45.0	10743	1.1779	0.6196
0.4467	46.0	10982	1.2048	0.6284
0.4332	47.0	11221	1.2302	0.6188
0.4529	48.0	11460	1.2220	0.6188
0.4281	49.0	11698	1.2013	0.624
0.4199	49.84	11900	1.2164	0.6152

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3