vit-base-patch16-224

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

• Loss: 0.6740
• Accuracy: 0.79
• Precision: 0.7955
• Recall: 0.79
• F1 Score: 0.7923

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: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• 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	Precision	Recall	F1 Score
No log	1.0	4	0.5895	0.725	0.5256	0.725	0.6094
No log	2.0	8	0.5737	0.725	0.5256	0.725	0.6094
No log	3.0	12	0.5746	0.7333	0.6978	0.7333	0.6589
No log	4.0	16	0.5449	0.7292	0.7126	0.7292	0.6263
No log	5.0	20	0.5943	0.7208	0.7362	0.7208	0.7270
No log	6.0	24	0.5124	0.75	0.7360	0.75	0.6895
No log	7.0	28	0.6057	0.6625	0.7301	0.6625	0.6797
No log	8.0	32	0.5059	0.7583	0.7376	0.7583	0.7214
No log	9.0	36	0.5734	0.7125	0.7474	0.7125	0.7237
No log	10.0	40	0.5069	0.7458	0.7182	0.7458	0.7116
No log	11.0	44	0.5135	0.775	0.7659	0.775	0.7689
No log	12.0	48	0.4943	0.775	0.7601	0.775	0.7610
0.5275	13.0	52	0.5654	0.7458	0.7790	0.7458	0.7557
0.5275	14.0	56	0.5257	0.7625	0.7636	0.7625	0.7631
0.5275	15.0	60	0.5107	0.7875	0.7813	0.7875	0.7836
0.5275	16.0	64	0.5514	0.7333	0.7655	0.7333	0.7434
0.5275	17.0	68	0.5004	0.7833	0.7698	0.7833	0.7699
0.5275	18.0	72	0.5999	0.7125	0.7738	0.7125	0.7269
0.5275	19.0	76	0.4975	0.7667	0.7554	0.7667	0.7589
0.5275	20.0	80	0.5120	0.7917	0.7981	0.7917	0.7944
0.5275	21.0	84	0.5203	0.7833	0.7876	0.7833	0.7853
0.5275	22.0	88	0.5304	0.8042	0.8051	0.8042	0.8046
0.5275	23.0	92	0.5475	0.825	0.825	0.825	0.8250
0.5275	24.0	96	0.5757	0.7458	0.7661	0.7458	0.7531
0.2422	25.0	100	0.5669	0.7875	0.7829	0.7875	0.7848
0.2422	26.0	104	0.5489	0.7958	0.7931	0.7958	0.7943
0.2422	27.0	108	0.5372	0.8	0.7982	0.8	0.7990
0.2422	28.0	112	0.5500	0.8208	0.8160	0.8208	0.8176
0.2422	29.0	116	0.5682	0.8042	0.8033	0.8042	0.8037
0.2422	30.0	120	0.5899	0.8083	0.8050	0.8083	0.8064
0.2422	31.0	124	0.6217	0.8	0.8063	0.8	0.8026
0.2422	32.0	128	0.6063	0.8125	0.8053	0.8125	0.8068
0.2422	33.0	132	0.5843	0.8042	0.8033	0.8042	0.8037
0.2422	34.0	136	0.6020	0.8125	0.8073	0.8125	0.8091
0.2422	35.0	140	0.6180	0.8042	0.8092	0.8042	0.8063
0.2422	36.0	144	0.6287	0.8208	0.8171	0.8208	0.8186
0.2422	37.0	148	0.6231	0.825	0.8234	0.825	0.8242
0.0631	38.0	152	0.6260	0.8292	0.8300	0.8292	0.8296
0.0631	39.0	156	0.6278	0.8333	0.8294	0.8333	0.8308
0.0631	40.0	160	0.6325	0.8208	0.8200	0.8208	0.8204
0.0631	41.0	164	0.6370	0.8083	0.8013	0.8083	0.8032
0.0631	42.0	168	0.6371	0.8125	0.8100	0.8125	0.8111
0.0631	43.0	172	0.6404	0.8042	0.8016	0.8042	0.8027
0.0631	44.0	176	0.6640	0.8292	0.8227	0.8292	0.8229
0.0631	45.0	180	0.6636	0.8208	0.8185	0.8208	0.8195
0.0631	46.0	184	0.6826	0.8083	0.8122	0.8083	0.8100
0.0631	47.0	188	0.6756	0.8208	0.8185	0.8208	0.8195
0.0631	48.0	192	0.6695	0.8292	0.8246	0.8292	0.8261
0.0631	49.0	196	0.6669	0.825	0.8198	0.825	0.8213
0.0264	50.0	200	0.6658	0.825	0.8198	0.825	0.8213

Framework versions

• Transformers 4.33.3
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.13.3