smids_10x_deit_tiny_sgd_0001_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: 0.3767
• Accuracy: 0.85

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
• 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
1.081	1.0	750	1.0837	0.4183
0.9664	2.0	1500	0.9960	0.4867
0.9109	3.0	2250	0.9172	0.5567
0.8062	4.0	3000	0.8454	0.6017
0.7721	5.0	3750	0.7755	0.6483
0.6595	6.0	4500	0.7101	0.6883
0.5526	7.0	5250	0.6552	0.715
0.6103	8.0	6000	0.6092	0.7533
0.6584	9.0	6750	0.5725	0.7633
0.5611	10.0	7500	0.5452	0.7983
0.5536	11.0	8250	0.5225	0.8067
0.522	12.0	9000	0.5053	0.8117
0.5314	13.0	9750	0.4913	0.8167
0.4936	14.0	10500	0.4799	0.8267
0.5195	15.0	11250	0.4692	0.8283
0.4075	16.0	12000	0.4606	0.8283
0.4566	17.0	12750	0.4529	0.8317
0.4172	18.0	13500	0.4445	0.8367
0.4556	19.0	14250	0.4390	0.83
0.4667	20.0	15000	0.4334	0.8333
0.3932	21.0	15750	0.4273	0.8333
0.4625	22.0	16500	0.4229	0.8367
0.418	23.0	17250	0.4180	0.8383
0.3957	24.0	18000	0.4143	0.8367
0.4114	25.0	18750	0.4106	0.8367
0.4039	26.0	19500	0.4070	0.8367
0.3652	27.0	20250	0.4039	0.84
0.3862	28.0	21000	0.4011	0.84
0.4364	29.0	21750	0.3984	0.84
0.3781	30.0	22500	0.3966	0.8417
0.3636	31.0	23250	0.3941	0.8383
0.3588	32.0	24000	0.3920	0.84
0.4007	33.0	24750	0.3903	0.8417
0.3328	34.0	25500	0.3888	0.84
0.3699	35.0	26250	0.3865	0.8417
0.3686	36.0	27000	0.3852	0.8433
0.315	37.0	27750	0.3840	0.8467
0.3799	38.0	28500	0.3828	0.8467
0.3659	39.0	29250	0.3817	0.8467
0.3715	40.0	30000	0.3806	0.8467
0.3582	41.0	30750	0.3798	0.8467
0.4093	42.0	31500	0.3792	0.8483
0.3651	43.0	32250	0.3786	0.8483
0.3713	44.0	33000	0.3782	0.85
0.3675	45.0	33750	0.3776	0.85
0.3336	46.0	34500	0.3773	0.85
0.4464	47.0	35250	0.3769	0.85
0.3703	48.0	36000	0.3768	0.85
0.366	49.0	36750	0.3767	0.85
0.311	50.0	37500	0.3767	0.85

Framework versions

• Transformers 4.32.1
• Pytorch 2.1.0+cu121
• Datasets 2.12.0
• Tokenizers 0.13.2