wmc_v2_vit_base_wm811k_cls_contra_learning_0916

This model is a fine-tuned version of google/vit-base-patch16-224 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0661
• Accuracy: 0.9768
• Precision: 0.9627
• Recall: 0.9551
• F1: 0.9585

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: 2e-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
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
1.1711	0.1697	100	0.6405	0.7559	0.6494	0.5790	0.5526
0.7143	0.3394	200	0.3116	0.8971	0.8478	0.7631	0.7576
0.584	0.5091	300	0.2060	0.9489	0.9137	0.8836	0.8940
0.4654	0.6788	400	0.1431	0.9603	0.9190	0.9289	0.9230
0.4465	0.8485	500	0.1176	0.9679	0.9458	0.9295	0.9373
0.3368	1.0182	600	0.1395	0.9550	0.9338	0.9244	0.9248
0.3741	1.1880	700	0.1541	0.9528	0.9287	0.9328	0.9269
0.3191	1.3577	800	0.1039	0.9697	0.9510	0.9453	0.9470
0.3354	1.5274	900	0.0952	0.9709	0.9530	0.9539	0.9529
0.3122	1.6971	1000	0.0799	0.9761	0.9456	0.9665	0.9556
0.295	1.8668	1100	0.0770	0.9758	0.9615	0.9534	0.9567
0.2993	2.0365	1200	0.0650	0.9794	0.9655	0.9597	0.9624
0.227	2.2062	1300	0.0717	0.9763	0.9598	0.9573	0.9584
0.2508	2.3759	1400	0.0653	0.9785	0.9605	0.9621	0.9613
0.3053	2.5456	1500	0.0629	0.9797	0.9623	0.9617	0.9620
0.2183	2.7153	1600	0.0676	0.9767	0.9597	0.9553	0.9572
0.219	2.8850	1700	0.0661	0.9768	0.9627	0.9551	0.9585

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1