vit-base-beans

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the beans dataset. It achieves the following results on the evaluation set:

• Loss: 0.0079
• Accuracy: 1.0

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 1337
• distributed_type: multi-GPU
• num_devices: 2
• total_train_batch_size: 16
• total_eval_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5.0

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss
0.2859	1.0	130	0.9624	0.2189
0.1316	2.0	260	0.9699	0.1334
0.1438	3.0	390	0.9699	0.0981
0.0833	4.0	520	0.9925	0.0656
0.1107	5.0	650	0.9774	0.0817
0.098	11.0	715	0.9925	0.0570
0.0935	12.0	780	1.0	0.0418
0.0907	13.0	845	0.9699	0.1093
0.0947	14.0	910	1.0	0.0347
0.1259	15.0	975	0.9850	0.0710
0.0325	16.0	1040	0.9774	0.0587
0.1397	17.0	1105	0.9925	0.0495
0.0456	18.0	1170	0.9774	0.0519
0.0439	19.0	1235	1.0	0.0216
0.0484	20.0	1300	0.9925	0.0316
0.0276	21.0	1365	1.0	0.0192
0.0348	22.0	1430	1.0	0.0177
0.0326	23.0	1495	1.0	0.0175
0.1014	24.0	1560	0.9925	0.0235
0.0395	25.0	1625	0.9850	0.0451
0.0265	26.0	1690	0.9925	0.0297
0.0569	27.0	1755	0.9925	0.0263
0.0666	28.0	1820	0.9850	0.0245
0.0285	29.0	1885	0.9774	0.0418
0.0892	30.0	1950	0.9925	0.0204
0.0371	31.0	2015	0.9850	0.0339
0.0105	32.0	2080	1.0	0.0143
0.0563	33.0	2145	1.0	0.0140
0.0573	34.0	2210	1.0	0.0102
0.0409	35.0	2275	1.0	0.0096
0.0523	36.0	2340	0.9925	0.0149
0.0131	37.0	2405	0.9925	0.0197
0.0329	38.0	2470	1.0	0.0109
0.0577	39.0	2535	1.0	0.0096
0.0085	40.0	2600	0.9925	0.0147
0.0618	41.0	2665	1.0	0.0094
0.0847	42.0	2730	0.9925	0.0197
0.0291	43.0	2795	1.0	0.0089
0.0568	44.0	2860	1.0	0.0087
0.0077	45.0	2925	1.0	0.0104
0.008	46.0	2990	1.0	0.0138
0.0272	47.0	3055	1.0	0.0081
0.008	48.0	3120	1.0	0.0084
0.0112	49.0	3185	1.0	0.0082
0.013	50.0	3250	1.0	0.0079

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1