Swin-DA2-final-AMD-Wet

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: 1.0322
• Accuracy: 0.8068

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: 80

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.6093	0.97	15	1.6090	0.2045
1.6042	2.0	31	1.6055	0.2045
1.586	2.97	46	1.5730	0.1932
1.4855	4.0	62	1.4935	0.2614
1.3738	4.97	77	1.3273	0.5
1.1557	6.0	93	1.0828	0.625
1.0441	6.97	108	0.8971	0.6023
0.9358	8.0	124	0.7635	0.6818
0.7707	8.97	139	0.8718	0.6477
0.7161	10.0	155	0.6903	0.7727
0.6799	10.97	170	0.8108	0.7273
0.6402	12.0	186	0.7147	0.7273
0.5131	12.97	201	0.7521	0.75
0.5262	14.0	217	0.7967	0.7273
0.4757	14.97	232	0.7084	0.7614
0.4758	16.0	248	0.7529	0.7727
0.4184	16.97	263	0.7769	0.7727
0.398	18.0	279	0.8496	0.7386
0.3591	18.97	294	0.8204	0.7273
0.3536	20.0	310	0.8589	0.7614
0.2589	20.97	325	0.9754	0.7045
0.3218	22.0	341	1.0231	0.7159
0.3151	22.97	356	0.9173	0.7386
0.2708	24.0	372	0.9598	0.7273
0.2802	24.97	387	0.9050	0.7386
0.3114	26.0	403	0.8725	0.7727
0.2794	26.97	418	0.9579	0.7386
0.26	28.0	434	1.0064	0.7273
0.2961	28.97	449	1.1056	0.75
0.297	30.0	465	0.8761	0.7727
0.2044	30.97	480	1.0461	0.7614
0.1884	32.0	496	0.9889	0.75
0.2156	32.97	511	1.0186	0.7727
0.194	34.0	527	1.0900	0.7727
0.2085	34.97	542	1.0762	0.75
0.1909	36.0	558	1.0325	0.7841
0.1551	36.97	573	1.1497	0.7045
0.2106	38.0	589	1.0304	0.7727
0.1771	38.97	604	1.0794	0.7841
0.1567	40.0	620	1.0634	0.7955
0.1856	40.97	635	1.0716	0.7614
0.185	42.0	651	1.0322	0.8068
0.1239	42.97	666	1.1516	0.7614
0.1617	44.0	682	1.0322	0.7841
0.1221	44.97	697	1.0553	0.8068
0.1433	46.0	713	1.0597	0.7727
0.216	46.97	728	1.1586	0.75
0.1807	48.0	744	1.0873	0.7727
0.185	48.97	759	1.2490	0.7727
0.1554	50.0	775	1.2223	0.7614
0.1359	50.97	790	1.2345	0.75
0.0929	52.0	806	1.1833	0.7614
0.1379	52.97	821	1.2581	0.7386
0.145	54.0	837	1.3023	0.75
0.134	54.97	852	1.2469	0.75
0.1974	56.0	868	1.2671	0.7386
0.122	56.97	883	1.2676	0.7273
0.1487	58.0	899	1.2846	0.7273
0.1282	58.97	914	1.1780	0.75
0.0989	60.0	930	1.2320	0.75
0.0997	60.97	945	1.2792	0.7386
0.1058	62.0	961	1.2126	0.7614
0.1105	62.97	976	1.2561	0.7386
0.0957	64.0	992	1.1702	0.7614
0.1326	64.97	1007	1.1839	0.75
0.0838	66.0	1023	1.2728	0.7386
0.1163	66.97	1038	1.2736	0.75
0.0926	68.0	1054	1.2641	0.75
0.102	68.97	1069	1.3310	0.75
0.0996	70.0	1085	1.3120	0.7273
0.081	70.97	1100	1.3358	0.7273
0.1305	72.0	1116	1.3440	0.7273
0.1131	72.97	1131	1.3126	0.7273
0.0883	74.0	1147	1.2848	0.7386
0.0873	74.97	1162	1.2802	0.7386
0.0991	76.0	1178	1.2711	0.75
0.0881	76.97	1193	1.2746	0.75
0.0895	77.42	1200	1.2752	0.75

Framework versions

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