google-vit-base-patch16-384-in21k-batch_16_epoch_4_classes_24_final_withAug_12th_May

This model is a fine-tuned version of Zannatul/final_withAug-google-vit-base-patch16-384-in21k-batch_16_epoch_4_classes_24 on the bengali_food_images dataset. It achieves the following results on the evaluation set:

• Loss: 0.0282
• Accuracy: 0.9905

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
0.269	0.09	100	0.4001	0.8723
0.1829	0.17	200	0.0928	0.9728
0.1737	0.26	300	0.1615	0.9457
0.2096	0.34	400	0.4138	0.9022
0.1855	0.43	500	0.1814	0.9511
0.0901	0.51	600	0.1435	0.9579
0.1406	0.6	700	0.1468	0.9620
0.136	0.68	800	0.1532	0.9565
0.0666	0.77	900	0.1177	0.9674
0.1145	0.85	1000	0.1794	0.9497
0.0865	0.94	1100	0.1113	0.9688
0.0612	1.02	1200	0.1270	0.9688
0.0038	1.11	1300	0.0724	0.9783
0.0006	1.19	1400	0.0715	0.9851
0.0007	1.28	1500	0.0616	0.9796
0.0579	1.36	1600	0.1259	0.9715
0.0009	1.45	1700	0.1028	0.9755
0.0295	1.53	1800	0.0637	0.9823
0.0484	1.62	1900	0.0893	0.9783
0.0371	1.71	2000	0.0637	0.9837
0.0359	1.79	2100	0.0389	0.9878
0.0006	1.88	2200	0.0750	0.9823
0.0189	1.96	2300	0.0451	0.9851
0.0442	2.05	2400	0.0772	0.9796
0.0006	2.13	2500	0.1988	0.9620
0.006	2.22	2600	0.0659	0.9864
0.0093	2.3	2700	0.0754	0.9810
0.0008	2.39	2800	0.0800	0.9783
0.0003	2.47	2900	0.0617	0.9864
0.0094	2.56	3000	0.0736	0.9837
0.0001	2.64	3100	0.0538	0.9823
0.001	2.73	3200	0.0606	0.9878
0.0001	2.81	3300	0.0433	0.9864
0.0001	2.9	3400	0.0583	0.9823
0.0001	2.98	3500	0.0388	0.9905
0.0001	3.07	3600	0.0408	0.9891
0.0001	3.15	3700	0.0375	0.9891
0.0001	3.24	3800	0.0367	0.9878
0.0001	3.32	3900	0.0355	0.9878
0.0001	3.41	4000	0.0395	0.9878
0.0001	3.5	4100	0.0382	0.9878
0.0001	3.58	4200	0.0399	0.9891
0.0001	3.67	4300	0.0396	0.9891
0.0072	3.75	4400	0.0355	0.9905
0.0001	3.84	4500	0.0284	0.9918
0.0001	3.92	4600	0.0282	0.9905

Framework versions

• Transformers 4.39.3
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.15.2