swin-tiny-patch4-window7-224-finetuned-eurosat

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: 1.0291
• Accuracy: 0.5714

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: 90

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	1	0.9799	0.4286
No log	2.0	2	0.9703	0.4286
No log	3.0	3	0.9703	0.4286
No log	4.0	4	0.9699	0.4286
No log	5.0	5	0.9699	0.4286
No log	6.0	6	0.9881	0.4286
No log	7.0	7	0.9881	0.4286
No log	8.0	8	1.0213	0.4286
No log	9.0	9	1.0213	0.4286
0.426	10.0	10	1.0291	0.5714
0.426	11.0	11	1.0291	0.5714
0.426	12.0	12	0.9996	0.5714
0.426	13.0	13	0.9996	0.5714
0.426	14.0	14	0.8998	0.5714
0.426	15.0	15	0.8998	0.5714
0.426	16.0	16	0.8356	0.5714
0.426	17.0	17	0.8356	0.5714
0.426	18.0	18	0.8575	0.5714
0.426	19.0	19	0.8575	0.5714
0.324	20.0	20	0.9310	0.4286
0.324	21.0	21	0.9310	0.4286
0.324	22.0	22	1.0029	0.4286
0.324	23.0	23	1.0029	0.4286
0.324	24.0	24	1.0582	0.4286
0.324	25.0	25	1.0582	0.4286
0.324	26.0	26	1.0812	0.4286
0.324	27.0	27	1.0812	0.4286
0.324	28.0	28	1.0345	0.4286
0.324	29.0	29	1.0345	0.4286
0.2536	30.0	30	0.9996	0.4286
0.2536	31.0	31	0.9996	0.4286
0.2536	32.0	32	0.9401	0.5714
0.2536	33.0	33	0.9401	0.5714
0.2536	34.0	34	0.8978	0.5714
0.2536	35.0	35	0.8978	0.5714
0.2536	36.0	36	0.9056	0.5714
0.2536	37.0	37	0.9056	0.5714
0.2536	38.0	38	0.9364	0.5714
0.2536	39.0	39	0.9364	0.5714
0.2176	40.0	40	1.0523	0.5714
0.2176	41.0	41	1.0523	0.5714
0.2176	42.0	42	1.1687	0.4286
0.2176	43.0	43	1.1687	0.4286
0.2176	44.0	44	1.1968	0.4286
0.2176	45.0	45	1.1968	0.4286
0.2176	46.0	46	1.1604	0.4286
0.2176	47.0	47	1.1604	0.4286
0.2176	48.0	48	1.0505	0.4286
0.2176	49.0	49	1.0505	0.4286
0.1597	50.0	50	0.9059	0.5714
0.1597	51.0	51	0.9059	0.5714
0.1597	52.0	52	0.8606	0.5714
0.1597	53.0	53	0.8606	0.5714
0.1597	54.0	54	0.8946	0.5714
0.1597	55.0	55	0.8946	0.5714
0.1597	56.0	56	0.9643	0.5714
0.1597	57.0	57	0.9643	0.5714
0.1597	58.0	58	1.0598	0.5714
0.1597	59.0	59	1.0598	0.5714
0.1231	60.0	60	1.1833	0.5714
0.1231	61.0	61	1.1833	0.5714
0.1231	62.0	62	1.2730	0.5714
0.1231	63.0	63	1.2730	0.5714
0.1231	64.0	64	1.3132	0.4286
0.1231	65.0	65	1.3132	0.4286
0.1231	66.0	66	1.3025	0.4286
0.1231	67.0	67	1.3025	0.4286
0.1231	68.0	68	1.2702	0.4286
0.1231	69.0	69	1.2702	0.4286
0.1364	70.0	70	1.2411	0.4286
0.1364	71.0	71	1.2411	0.4286
0.1364	72.0	72	1.2222	0.4286
0.1364	73.0	73	1.2222	0.4286
0.1364	74.0	74	1.2257	0.4286
0.1364	75.0	75	1.2257	0.4286
0.1364	76.0	76	1.2552	0.4286
0.1364	77.0	77	1.2552	0.4286
0.1364	78.0	78	1.2701	0.5714
0.1364	79.0	79	1.2701	0.5714
0.0937	80.0	80	1.2753	0.5714
0.0937	81.0	81	1.2753	0.5714
0.0937	82.0	82	1.2797	0.5714
0.0937	83.0	83	1.2797	0.5714
0.0937	84.0	84	1.2840	0.5714
0.0937	85.0	85	1.2840	0.5714
0.0937	86.0	86	1.2895	0.5714
0.0937	87.0	87	1.2895	0.5714
0.0937	88.0	88	1.2917	0.5714
0.0937	89.0	89	1.2917	0.5714
0.1082	90.0	90	1.2931	0.5714

Framework versions

• Transformers 4.30.1
• Pytorch 2.0.1+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3