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: 0.3209
• Accuracy: 0.8902

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: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	8	2.7448	0.0314
2.7716	2.0	16	2.5834	0.1765
2.5974	3.0	24	2.3608	0.3020
2.3426	4.0	32	2.1157	0.3333
1.9747	5.0	40	1.7539	0.4627
1.9747	6.0	48	1.3641	0.6078
1.5182	7.0	56	1.0755	0.6471
1.198	8.0	64	0.8743	0.7216
1.0206	9.0	72	0.7666	0.7294
0.8731	10.0	80	0.7035	0.7490
0.8731	11.0	88	0.6122	0.7608
0.7938	12.0	96	0.6508	0.7490
0.7286	13.0	104	0.5081	0.7961
0.659	14.0	112	0.5536	0.7961
0.6232	15.0	120	0.5079	0.8
0.6232	16.0	128	0.4483	0.8314
0.6028	17.0	136	0.4096	0.8157
0.5333	18.0	144	0.3710	0.8510
0.5053	19.0	152	0.4810	0.8039
0.4717	20.0	160	0.4121	0.8235
0.4717	21.0	168	0.4021	0.8392
0.4728	22.0	176	0.3780	0.8588
0.4347	23.0	184	0.3374	0.8745
0.4545	24.0	192	0.4056	0.8431
0.3954	25.0	200	0.4088	0.8745
0.3954	26.0	208	0.4169	0.8392
0.4145	27.0	216	0.3262	0.8706
0.3895	28.0	224	0.4235	0.8706
0.4185	29.0	232	0.3482	0.8706
0.3686	30.0	240	0.3088	0.8824
0.3686	31.0	248	0.3230	0.8902
0.3617	32.0	256	0.3473	0.8824
0.3136	33.0	264	0.3793	0.8627
0.3482	34.0	272	0.3477	0.8588
0.3519	35.0	280	0.3692	0.8667
0.3519	36.0	288	0.3611	0.8627
0.3311	37.0	296	0.3233	0.8745
0.3222	38.0	304	0.3416	0.8627
0.3013	39.0	312	0.3198	0.8824
0.2871	40.0	320	0.3308	0.8667
0.2871	41.0	328	0.3246	0.8667
0.3154	42.0	336	0.3943	0.8667
0.2735	43.0	344	0.3186	0.8784
0.2911	44.0	352	0.3132	0.8824
0.266	45.0	360	0.3204	0.8980
0.266	46.0	368	0.3097	0.8784
0.2686	47.0	376	0.3075	0.8902
0.2818	48.0	384	0.3192	0.8902
0.2492	49.0	392	0.3434	0.8745
0.276	50.0	400	0.3237	0.8824
0.276	51.0	408	0.3450	0.8745
0.245	52.0	416	0.3284	0.8706
0.2292	53.0	424	0.3263	0.8902
0.2252	54.0	432	0.3216	0.8745
0.2483	55.0	440	0.3359	0.8863
0.2483	56.0	448	0.3314	0.8902
0.2549	57.0	456	0.3932	0.8745
0.2247	58.0	464	0.3189	0.8745
0.2344	59.0	472	0.3251	0.8745
0.2315	60.0	480	0.3289	0.8824
0.2315	61.0	488	0.3058	0.8745
0.2109	62.0	496	0.2999	0.8863
0.2325	63.0	504	0.3078	0.8980
0.2126	64.0	512	0.3531	0.8784
0.1975	65.0	520	0.3394	0.8902
0.1975	66.0	528	0.3113	0.8902
0.1998	67.0	536	0.3365	0.8941
0.2208	68.0	544	0.2854	0.9020
0.2126	69.0	552	0.3170	0.8941
0.2352	70.0	560	0.3155	0.8824
0.2352	71.0	568	0.3327	0.8824
0.1724	72.0	576	0.3503	0.8902
0.2038	73.0	584	0.3309	0.8824
0.1919	74.0	592	0.3299	0.8902
0.2199	75.0	600	0.3347	0.8863
0.2199	76.0	608	0.3471	0.8824
0.2075	77.0	616	0.3437	0.8863
0.2206	78.0	624	0.3161	0.8824
0.1655	79.0	632	0.3227	0.8784
0.1765	80.0	640	0.3302	0.8784
0.1765	81.0	648	0.3153	0.8745
0.1832	82.0	656	0.3010	0.8745
0.185	83.0	664	0.3266	0.8941
0.1627	84.0	672	0.3192	0.8941
0.176	85.0	680	0.3125	0.8863
0.176	86.0	688	0.3241	0.8745
0.1723	87.0	696	0.3124	0.8784
0.1477	88.0	704	0.3109	0.8745
0.1703	89.0	712	0.3196	0.8824
0.1919	90.0	720	0.3186	0.8980
0.1919	91.0	728	0.3178	0.8902
0.1465	92.0	736	0.3241	0.8824
0.155	93.0	744	0.3281	0.8784
0.1829	94.0	752	0.3263	0.8824
0.167	95.0	760	0.3282	0.8824
0.167	96.0	768	0.3290	0.8824
0.166	97.0	776	0.3253	0.8902
0.1756	98.0	784	0.3231	0.8863
0.157	99.0	792	0.3215	0.8902
0.1492	100.0	800	0.3209	0.8902

Framework versions

• Transformers 4.33.3
• Pytorch 1.11.0+cu113
• Datasets 2.14.5
• Tokenizers 0.13.3