metadata

license: apache-2.0
base_model: google/vit-base-patch16-384
tags:
  - image-classification
  - generated_from_trainer
datasets:
  - imagefolder
metrics:
  - accuracy
model-index:
  - name: google-vit-base-patch16-384-batch_16_epoch_4_classes_24
    results:
      - task:
          name: Image Classification
          type: image-classification
        dataset:
          name: bengali_food_images
          type: imagefolder
          config: default
          split: train
          args: default
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.9899425287356322

google-vit-base-patch16-384-batch_16_epoch_4_classes_24

This model is a fine-tuned version of google/vit-base-patch16-384 on the bengali_food_images dataset. It achieves the following results on the evaluation set:

Loss: 0.0635
Accuracy: 0.9899

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: 0.0002
train_batch_size: 16
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 4
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.2947	0.07	100	0.2491	0.9353
0.1703	0.14	200	0.2377	0.9339
0.0797	0.21	300	0.1413	0.9641
0.1035	0.28	400	0.1057	0.9641
0.0532	0.35	500	0.1711	0.9483
0.1004	0.42	600	0.1746	0.9526
0.0962	0.49	700	0.1598	0.9555
0.1579	0.56	800	0.1741	0.9440
0.0532	0.63	900	0.0974	0.9670
0.1594	0.7	1000	0.2842	0.9181
0.0488	0.77	1100	0.2928	0.9224
0.1122	0.84	1200	0.3095	0.9138
0.1252	0.91	1300	0.1411	0.9569
0.0517	0.97	1400	0.1378	0.9684
0.047	1.04	1500	0.2595	0.9483
0.0478	1.11	1600	0.1425	0.9583
0.0107	1.18	1700	0.1135	0.9684
0.0021	1.25	1800	0.1428	0.9598
0.036	1.32	1900	0.1851	0.9583
0.0733	1.39	2000	0.1801	0.9583
0.0549	1.46	2100	0.1917	0.9598
0.0442	1.53	2200	0.1538	0.9655
0.0196	1.6	2300	0.1411	0.9698
0.0809	1.67	2400	0.1862	0.9540
0.0004	1.74	2500	0.1325	0.9698
0.0404	1.81	2600	0.1246	0.9713
0.0691	1.88	2700	0.1961	0.9598
0.0088	1.95	2800	0.1841	0.9684
0.0029	2.02	2900	0.1057	0.9813
0.0005	2.09	3000	0.1131	0.9741
0.0001	2.16	3100	0.0892	0.9813
0.0002	2.23	3200	0.0757	0.9828
0.0186	2.3	3300	0.0794	0.9784
0.0127	2.37	3400	0.1100	0.9770
0.0048	2.44	3500	0.1386	0.9799
0.0048	2.51	3600	0.0635	0.9899
0.001	2.58	3700	0.0997	0.9799
0.0005	2.65	3800	0.1119	0.9756
0.0006	2.72	3900	0.1292	0.9713
0.0003	2.79	4000	0.1186	0.9770
0.0137	2.86	4100	0.0969	0.9770
0.0001	2.92	4200	0.0738	0.9842
0.0001	2.99	4300	0.1236	0.9828
0.0001	3.06	4400	0.0932	0.9856
0.0001	3.13	4500	0.0992	0.9799
0.0001	3.2	4600	0.0960	0.9828
0.0001	3.27	4700	0.1123	0.9799
0.0001	3.34	4800	0.1107	0.9813
0.0029	3.41	4900	0.1041	0.9842
0.0001	3.48	5000	0.1074	0.9828
0.0001	3.55	5100	0.1111	0.9799
0.0001	3.62	5200	0.1088	0.9784
0.0001	3.69	5300	0.0936	0.9813
0.0001	3.76	5400	0.0915	0.9799
0.0001	3.83	5500	0.0897	0.9799
0.0001	3.9	5600	0.0875	0.9770
0.0	3.97	5700	0.0856	0.9784

Framework versions

Transformers 4.39.3
Pytorch 2.1.2
Datasets 2.18.0
Tokenizers 0.15.2