vit-KAIYI

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.4642
• F1: 0.6998
• Roc Auc: 0.8236

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
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc
0.4021	1.0	506	0.3330	0.5653	0.7446
0.244	2.0	1012	0.2762	0.6360	0.7862
0.1793	3.0	1518	0.2503	0.6857	0.8154
0.1382	4.0	2024	0.2399	0.6991	0.8232
0.1089	5.0	2530	0.2336	0.7010	0.8244
0.0869	6.0	3036	0.2307	0.7092	0.8291
0.0698	7.0	3542	0.2318	0.7134	0.8316
0.0563	8.0	4048	0.2402	0.7076	0.8282
0.0459	9.0	4554	0.2402	0.7040	0.8261
0.0371	10.0	5060	0.2514	0.7053	0.8269
0.0306	11.0	5566	0.2582	0.7065	0.8275
0.0249	12.0	6072	0.2672	0.7050	0.8267
0.0206	13.0	6578	0.2735	0.7060	0.8273
0.0171	14.0	7084	0.2822	0.7039	0.8261
0.0142	15.0	7590	0.2934	0.7085	0.8288
0.0122	16.0	8096	0.2980	0.7066	0.8276
0.0104	17.0	8602	0.3126	0.6994	0.8234
0.0087	18.0	9108	0.3213	0.7040	0.8261
0.0074	19.0	9614	0.3245	0.7042	0.8262
0.0063	20.0	10120	0.3412	0.7038	0.8260
0.0053	21.0	10626	0.3525	0.7031	0.8256
0.0044	22.0	11132	0.3584	0.7038	0.8260
0.0039	23.0	11638	0.3710	0.7026	0.8253
0.0033	24.0	12144	0.3787	0.7020	0.8249
0.0029	25.0	12650	0.3867	0.7023	0.8251
0.0025	26.0	13156	0.3946	0.7015	0.8246
0.0023	27.0	13662	0.4052	0.7004	0.8240
0.002	28.0	14168	0.4115	0.6996	0.8235
0.0018	29.0	14674	0.4196	0.7005	0.8240
0.0016	30.0	15180	0.4360	0.7013	0.8245
0.0014	31.0	15686	0.4297	0.7008	0.8242
0.0013	32.0	16192	0.4444	0.6980	0.8226
0.0012	33.0	16698	0.4420	0.7010	0.8243
0.0011	34.0	17204	0.4468	0.6998	0.8237
0.001	35.0	17710	0.4529	0.7005	0.8240
0.001	36.0	18216	0.4566	0.7000	0.8237
0.0009	37.0	18722	0.4599	0.6997	0.8236
0.0009	38.0	19228	0.4621	0.6999	0.8237
0.0009	39.0	19734	0.4649	0.6993	0.8233
0.0008	40.0	20240	0.4642	0.6998	0.8236

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.3