swin-tiny-patch4-window7-224

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

• Loss: 0.5534
• Accuracy: 0.7433
• Precision: 0.7306
• Recall: 0.7433
• F1 Score: 0.7344

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: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• 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	Precision	Recall	F1 Score
No log	1.0	4	0.7306	0.4	0.6521	0.4	0.3821
No log	2.0	8	0.5815	0.7333	0.8050	0.7333	0.6286
No log	3.0	12	0.5700	0.725	0.5256	0.725	0.6094
No log	4.0	16	0.5635	0.725	0.5256	0.725	0.6094
No log	5.0	20	0.5509	0.7292	0.8028	0.7292	0.6191
No log	6.0	24	0.5356	0.7417	0.7438	0.7417	0.6589
No log	7.0	28	0.5353	0.75	0.7360	0.75	0.6895
No log	8.0	32	0.5299	0.7375	0.7090	0.7375	0.6668
No log	9.0	36	0.5335	0.7667	0.7509	0.7667	0.7310
No log	10.0	40	0.5344	0.7417	0.7315	0.7417	0.6644
No log	11.0	44	0.5297	0.7458	0.7279	0.7458	0.6821
No log	12.0	48	0.5202	0.75	0.7360	0.75	0.6895
0.5942	13.0	52	0.5325	0.7542	0.7411	0.7542	0.7452
0.5942	14.0	56	0.5139	0.7583	0.7505	0.7583	0.7039
0.5942	15.0	60	0.5528	0.7417	0.7347	0.7417	0.7377
0.5942	16.0	64	0.5070	0.7625	0.7437	0.7625	0.7277
0.5942	17.0	68	0.5193	0.775	0.7594	0.775	0.7592
0.5942	18.0	72	0.5090	0.7583	0.7448	0.7583	0.7487
0.5942	19.0	76	0.5189	0.7792	0.7847	0.7792	0.7816
0.5942	20.0	80	0.5214	0.775	0.7795	0.775	0.7770
0.5942	21.0	84	0.5188	0.775	0.7710	0.775	0.7728
0.5942	22.0	88	0.5029	0.7667	0.7526	0.7667	0.7557
0.5942	23.0	92	0.5061	0.7833	0.7734	0.7833	0.7761
0.5942	24.0	96	0.5350	0.7667	0.7713	0.7667	0.7687
0.4829	25.0	100	0.5149	0.7542	0.7330	0.7542	0.7337
0.4829	26.0	104	0.5283	0.7583	0.7737	0.7583	0.7641
0.4829	27.0	108	0.5109	0.7792	0.7647	0.7792	0.7646
0.4829	28.0	112	0.5258	0.775	0.7729	0.775	0.7739
0.4829	29.0	116	0.5207	0.7625	0.745	0.7625	0.7468
0.4829	30.0	120	0.5306	0.75	0.7357	0.75	0.7400
0.4829	31.0	124	0.5455	0.75	0.7375	0.75	0.7417
0.4829	32.0	128	0.5653	0.7458	0.7380	0.7458	0.7412
0.4829	33.0	132	0.5565	0.7417	0.7212	0.7417	0.7256
0.4829	34.0	136	0.5468	0.7708	0.7658	0.7708	0.7679
0.4829	35.0	140	0.5268	0.7833	0.7723	0.7833	0.7747
0.4829	36.0	144	0.5260	0.775	0.7710	0.775	0.7728
0.4829	37.0	148	0.5281	0.775	0.7659	0.775	0.7689
0.3846	38.0	152	0.5385	0.7708	0.7742	0.7708	0.7724
0.3846	39.0	156	0.5253	0.7708	0.7623	0.7708	0.7653
0.3846	40.0	160	0.5319	0.7708	0.7719	0.7708	0.7714
0.3846	41.0	164	0.5311	0.775	0.7631	0.775	0.7660
0.3846	42.0	168	0.5325	0.7792	0.7683	0.7792	0.7711
0.3846	43.0	172	0.5254	0.7667	0.7606	0.7667	0.7631
0.3846	44.0	176	0.5232	0.7708	0.7623	0.7708	0.7653
0.3846	45.0	180	0.5291	0.7708	0.7640	0.7708	0.7667
0.3846	46.0	184	0.5356	0.7708	0.7607	0.7708	0.7639
0.3846	47.0	188	0.5400	0.7708	0.7607	0.7708	0.7639
0.3846	48.0	192	0.5409	0.7667	0.7540	0.7667	0.7573
0.3846	49.0	196	0.5403	0.7667	0.7540	0.7667	0.7573
0.3353	50.0	200	0.5397	0.7708	0.7592	0.7708	0.7624

Framework versions

• Transformers 4.33.3
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3