project_4_transfer_learning

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: 1.1429
• Accuracy: 0.6438

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-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss
2.0754	1.0	10	0.125	2.0725
2.0459	2.0	20	0.2625	2.0286
1.968	3.0	30	0.3	1.9506
1.8311	4.0	40	0.4188	1.8060
1.6911	5.0	50	0.4313	1.6814
1.5677	6.0	60	0.4313	1.5851
1.4801	7.0	70	0.4813	1.5169
1.4033	8.0	80	0.4813	1.4614
1.3435	9.0	90	0.475	1.4358
1.3054	10.0	100	0.525	1.4292
1.2532	11.0	110	0.5188	1.3942
1.2178	12.0	120	0.5312	1.3684
1.1857	13.0	130	0.5062	1.3599
1.1558	14.0	140	0.5312	1.2992
1.1118	15.0	150	0.5375	1.3217
1.0967	16.0	160	0.525	1.3177
1.0671	17.0	170	0.5312	1.3420
1.0635	18.0	180	0.5062	1.3319
1.044	19.0	190	0.5813	1.2977
1.037	20.0	200	0.5125	1.3127
1.0743	21.0	210	1.2062	0.6062
1.0454	22.0	220	1.1564	0.65
1.0457	23.0	230	1.1484	0.6312
1.0246	24.0	240	1.1470	0.6312
0.9859	25.0	250	1.1200	0.6438
0.9885	26.0	260	1.1331	0.6375
0.9823	27.0	270	1.1069	0.6562
0.9412	28.0	280	1.1163	0.6375
0.9172	29.0	290	1.1192	0.6375
0.9334	30.0	300	1.1573	0.6

Framework versions

• Transformers 4.33.2
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3