hushem_1x_deit_tiny_adamax_lr0001_fold5

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

• Loss: 0.8924
• Accuracy: 0.6585

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: 0.0001
• 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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.67	1	1.9330	0.2439
No log	2.0	3	1.4362	0.3659
No log	2.67	4	1.3806	0.3902
No log	4.0	6	1.3304	0.4634
No log	4.67	7	1.3017	0.4390
No log	6.0	9	1.1836	0.4878
1.2323	6.67	10	1.1688	0.5610
1.2323	8.0	12	1.1361	0.5366
1.2323	8.67	13	1.1291	0.5366
1.2323	10.0	15	1.0782	0.6098
1.2323	10.67	16	1.0358	0.6585
1.2323	12.0	18	1.0020	0.6098
1.2323	12.67	19	1.0059	0.6098
0.3527	14.0	21	0.9293	0.6098
0.3527	14.67	22	0.9162	0.6341
0.3527	16.0	24	0.9233	0.6098
0.3527	16.67	25	0.9213	0.6098
0.3527	18.0	27	0.9193	0.6098
0.3527	18.67	28	0.9345	0.6098
0.04	20.0	30	0.8872	0.6585
0.04	20.67	31	0.8549	0.6829
0.04	22.0	33	0.8221	0.6829
0.04	22.67	34	0.8117	0.7073
0.04	24.0	36	0.8041	0.7561
0.04	24.67	37	0.8128	0.7561
0.04	26.0	39	0.8413	0.6829
0.0062	26.67	40	0.8565	0.6585
0.0062	28.0	42	0.8789	0.6585
0.0062	28.67	43	0.8864	0.6585
0.0062	30.0	45	0.8920	0.6585
0.0062	30.67	46	0.8925	0.6585
0.0062	32.0	48	0.8929	0.6585
0.0062	32.67	49	0.8927	0.6585
0.0031	33.33	50	0.8924	0.6585

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1