vit-base-patch16-224-R1-40

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

• Loss: 1.7212
• Accuracy: 0.7541

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: 5.5e-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: 0.05
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.3233	0.99	38	1.2355	0.5574
0.8643	1.99	76	0.9297	0.5902
0.4464	2.98	114	1.1190	0.6393
0.3092	4.0	153	0.9861	0.7049
0.1628	4.99	191	1.1221	0.6721
0.121	5.99	229	1.1710	0.6885
0.1138	6.98	267	1.1993	0.7213
0.1124	8.0	306	1.2636	0.6885
0.0748	8.99	344	1.3881	0.7049
0.0877	9.99	382	1.2892	0.7213
0.0642	10.98	420	1.3759	0.7049
0.0675	12.0	459	1.4283	0.7213
0.0694	12.99	497	1.3616	0.7213
0.0689	13.99	535	1.3864	0.7213
0.0378	14.98	573	1.4322	0.7213
0.0472	16.0	612	1.6004	0.7213
0.044	16.99	650	1.5810	0.7049
0.0386	17.99	688	1.6404	0.6885
0.0341	18.98	726	1.5698	0.7377
0.0328	20.0	765	1.6720	0.6885
0.0444	20.99	803	1.6269	0.7213
0.0342	21.99	841	1.6345	0.7377
0.0324	22.98	879	1.7916	0.7049
0.023	24.0	918	1.8753	0.6885
0.048	24.99	956	1.7679	0.7377
0.0202	25.99	994	1.7212	0.7541
0.0336	26.98	1032	1.7305	0.7377
0.0163	28.0	1071	1.7576	0.7049
0.0186	28.99	1109	1.7540	0.7377
0.0189	29.99	1147	1.6594	0.7541
0.039	30.98	1185	1.7423	0.7213
0.0194	32.0	1224	1.7148	0.7377
0.0205	32.99	1262	1.6965	0.7377
0.0186	33.99	1300	1.7553	0.7541
0.0177	34.98	1338	1.7476	0.7377
0.0132	36.0	1377	1.7506	0.7541
0.0068	36.99	1415	1.6917	0.7377
0.0121	37.99	1453	1.7276	0.7541
0.0129	38.98	1491	1.7218	0.7541
0.0067	39.74	1520	1.7220	0.7541

Framework versions

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