swinv2-tiny-patch4-window8-256-Diabetic-Retinopathy-DA

This model is a fine-tuned version of microsoft/swinv2-tiny-patch4-window8-256 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.6974
• Accuracy: 0.8091

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.5987	1.0	23	1.5683	0.4909
1.4137	2.0	46	1.2639	0.4909
1.1988	3.0	69	0.8726	0.7636
0.8533	4.0	92	0.6361	0.7545
0.8042	5.0	115	0.5985	0.7545
0.7349	6.0	138	0.5943	0.7545
0.7003	7.0	161	0.5178	0.7636
0.6641	8.0	184	0.5058	0.7545
0.641	9.0	207	0.5092	0.7909
0.6571	10.0	230	0.5319	0.7636
0.6522	11.0	253	0.5726	0.7909
0.5659	12.0	276	0.5490	0.7727
0.5511	13.0	299	0.5465	0.8
0.5435	14.0	322	0.5728	0.7909
0.5259	15.0	345	0.6047	0.7636
0.5496	16.0	368	0.6479	0.7818
0.543	17.0	391	0.6040	0.7727
0.4646	18.0	414	0.6269	0.7818
0.4867	19.0	437	0.6535	0.7909
0.4357	20.0	460	0.6991	0.7727
0.4392	21.0	483	0.7127	0.7636
0.4403	22.0	506	0.6974	0.8091
0.4358	23.0	529	0.6883	0.7818
0.4094	24.0	552	0.6768	0.8
0.3913	25.0	575	0.7270	0.7636
0.3686	26.0	598	0.7104	0.7727
0.3679	27.0	621	0.7115	0.7818
0.378	28.0	644	0.8020	0.8091
0.3583	29.0	667	0.7524	0.7909
0.3299	30.0	690	0.7783	0.7909
0.3672	31.0	713	0.8193	0.7909
0.3567	32.0	736	0.8095	0.7909
0.3585	33.0	759	0.8324	0.7909
0.3191	34.0	782	0.8042	0.7909
0.3144	35.0	805	0.8189	0.7909
0.3452	36.0	828	0.8377	0.7909
0.3263	37.0	851	0.8204	0.7909
0.2939	38.0	874	0.8103	0.7909
0.3152	39.0	897	0.8184	0.7818
0.2787	40.0	920	0.8241	0.7818

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu118
• Datasets 2.16.1
• Tokenizers 0.15.0