swinv2-tiny-patch4-window8-256-dmae-va-U5-42C

This model is a fine-tuned version of microsoft/swinv2-tiny-patch4-window8-256 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7073
• Accuracy: 0.7667

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: 1e-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: 42

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.9032	7	1.3926	0.35
1.4087	1.9355	15	1.3365	0.4167
1.3807	2.9677	23	1.2813	0.4167
1.35	4.0	31	1.2407	0.4
1.35	4.9032	38	1.2116	0.4833
1.2933	5.9355	46	1.1653	0.4833
1.2426	6.9677	54	1.1151	0.5167
1.1771	8.0	62	1.0441	0.6
1.1771	8.9032	69	0.9990	0.5667
1.0983	9.9355	77	0.9456	0.6333
1.0338	10.9677	85	0.9160	0.6833
0.9665	12.0	93	0.8940	0.6833
0.9133	12.9032	100	0.8753	0.6
0.9133	13.9355	108	0.8518	0.6667
0.8521	14.9677	116	0.8515	0.65
0.8461	16.0	124	0.8407	0.65
0.808	16.9032	131	0.8218	0.65
0.808	17.9355	139	0.8170	0.6833
0.7779	18.9677	147	0.7972	0.7167
0.758	20.0	155	0.7817	0.7333
0.7416	20.9032	162	0.7678	0.7167
0.7344	21.9355	170	0.7650	0.7167
0.7344	22.9677	178	0.7428	0.7333
0.7091	24.0	186	0.7280	0.75
0.6876	24.9032	193	0.7235	0.75
0.6887	25.9355	201	0.7278	0.75
0.6887	26.9677	209	0.7264	0.75
0.6897	28.0	217	0.7228	0.75
0.6637	28.9032	224	0.7163	0.75
0.6924	29.9355	232	0.7073	0.7667
0.6234	30.9677	240	0.7057	0.7667
0.6234	32.0	248	0.7090	0.7667
0.6652	32.9032	255	0.7052	0.7667
0.6343	33.9355	263	0.7009	0.7667
0.6327	34.9677	271	0.7017	0.7667
0.6327	36.0	279	0.7023	0.7667
0.6339	36.9032	286	0.7027	0.7667
0.6275	37.9355	294	0.7031	0.7667

Framework versions

• Transformers 4.40.0
• Pytorch 2.2.1+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1