swinv2-large-patch4-window12-192-22k-finetuned-eurosat-50

This model is a fine-tuned version of microsoft/swinv2-large-patch4-window12-192-22k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.6967
• Accuracy: 0.7220

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-06
• 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: 0.005
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.97	9	1.6984	0.3729
No log	1.95	18	1.5150	0.4881
1.6944	2.92	27	1.3304	0.5390
1.6944	4.0	37	1.1761	0.6
1.3633	4.97	46	1.0588	0.6373
1.3633	5.95	55	0.9952	0.6475
1.1208	6.92	64	0.9326	0.6610
1.1208	8.0	74	0.8785	0.6712
0.9891	8.97	83	0.8478	0.6746
0.9891	9.95	92	0.8144	0.6847
0.9011	10.92	101	0.7774	0.7017
0.9011	12.0	111	0.7567	0.6983
0.8143	12.97	120	0.7525	0.6949
0.8143	13.95	129	0.7309	0.7051
0.8143	14.92	138	0.7141	0.7119
0.7926	16.0	148	0.7095	0.7186
0.7926	16.97	157	0.7057	0.7220
0.7439	17.95	166	0.6988	0.7220
0.7439	18.92	175	0.6967	0.7220
0.7533	19.46	180	0.6967	0.7220

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu118
• Datasets 2.13.1
• Tokenizers 0.13.3