hushem_1x_deit_tiny_adamax_lr001_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: 1.1657
• 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: 0.001
• 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	4.7722	0.2439
No log	2.0	3	1.4567	0.2439
No log	2.67	4	1.8233	0.2683
No log	4.0	6	1.3918	0.2439
No log	4.67	7	1.4247	0.2195
No log	6.0	9	1.3988	0.2439
1.9646	6.67	10	1.3700	0.3415
1.9646	8.0	12	1.3164	0.3902
1.9646	8.67	13	1.2953	0.3902
1.9646	10.0	15	1.0825	0.5366
1.9646	10.67	16	0.9280	0.7561
1.9646	12.0	18	0.9474	0.5610
1.9646	12.67	19	0.9791	0.5122
1.1934	14.0	21	1.3039	0.3902
1.1934	14.67	22	1.3242	0.3902
1.1934	16.0	24	0.8880	0.6341
1.1934	16.67	25	0.8367	0.6341
1.1934	18.0	27	0.8476	0.6098
1.1934	18.67	28	0.9406	0.5854
0.8297	20.0	30	1.1819	0.4878
0.8297	20.67	31	0.9194	0.5610
0.8297	22.0	33	0.7486	0.6829
0.8297	22.67	34	1.1493	0.6341
0.8297	24.0	36	1.2217	0.5854
0.8297	24.67	37	0.7746	0.6829
0.8297	26.0	39	0.8320	0.6585
0.5433	26.67	40	1.2210	0.5610
0.5433	28.0	42	1.3782	0.5366
0.5433	28.67	43	1.1529	0.6098
0.5433	30.0	45	1.0361	0.6585
0.5433	30.67	46	1.1089	0.6585
0.5433	32.0	48	1.1802	0.6098
0.5433	32.67	49	1.1774	0.6585
0.2758	33.33	50	1.1657	0.6585

Framework versions

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