hushem_1x_deit_tiny_adamax_lr0001_fold1

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: 1.3365
• Accuracy: 0.5778

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	1.6054	0.2444
No log	2.0	3	1.3436	0.3556
No log	2.67	4	1.3392	0.2889
No log	4.0	6	1.3661	0.2444
No log	4.67	7	1.3117	0.3333
No log	6.0	9	1.4031	0.2889
1.1803	6.67	10	1.2845	0.4222
1.1803	8.0	12	1.3559	0.3333
1.1803	8.67	13	1.3178	0.4
1.1803	10.0	15	1.1302	0.5778
1.1803	10.67	16	1.2145	0.5556
1.1803	12.0	18	1.3484	0.4
1.1803	12.67	19	1.1709	0.5333
0.3935	14.0	21	1.1495	0.5556
0.3935	14.67	22	1.2656	0.4889
0.3935	16.0	24	1.1929	0.5333
0.3935	16.67	25	1.1205	0.5556
0.3935	18.0	27	1.1729	0.5333
0.3935	18.67	28	1.2656	0.5111
0.0911	20.0	30	1.3172	0.5556
0.0911	20.67	31	1.2343	0.5556
0.0911	22.0	33	1.1439	0.6
0.0911	22.67	34	1.1167	0.6222
0.0911	24.0	36	1.1537	0.6
0.0911	24.67	37	1.2658	0.5778
0.0911	26.0	39	1.3705	0.5556
0.0269	26.67	40	1.3468	0.5778
0.0269	28.0	42	1.2914	0.6
0.0269	28.67	43	1.2807	0.6
0.0269	30.0	45	1.2833	0.6
0.0269	30.67	46	1.3004	0.5778
0.0269	32.0	48	1.3271	0.5778
0.0269	32.67	49	1.3342	0.5778
0.0102	33.33	50	1.3365	0.5778

Framework versions

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