vit-base-16-thesis-demo-HAM10000

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the ahishamm/HAM_db_enhanced_balanced_reduced_50_20_20_50 dataset. It achieves the following results on the evaluation set:

• Loss: 0.5296
• Accuracy: 0.8344
• Recall: 0.8344
• F1: 0.8344
• Precision: 0.8344

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	Recall	F1	Precision
1.4855	0.12	50	1.3519	0.5093	0.5093	0.5093	0.5093
1.044	0.23	100	1.0515	0.6268	0.6268	0.6268	0.6268
1.0774	0.35	150	1.2104	0.5681	0.5681	0.5681	0.5681
0.9508	0.46	200	1.0624	0.6061	0.6061	0.6061	0.6061
0.9522	0.58	250	0.9338	0.6449	0.6449	0.6449	0.6449
0.774	0.69	300	0.8939	0.6676	0.6676	0.6676	0.6676
0.7675	0.81	350	0.7742	0.7183	0.7183	0.7183	0.7183
0.7167	0.92	400	0.7695	0.7216	0.7216	0.7216	0.7216
0.5204	1.04	450	0.8005	0.7303	0.7303	0.7303	0.7303
0.456	1.15	500	0.8523	0.6903	0.6903	0.6903	0.6903
0.5421	1.27	550	0.6753	0.7543	0.7543	0.7543	0.7543
0.4446	1.38	600	0.6042	0.7810	0.7810	0.7810	0.7810
0.455	1.5	650	0.6913	0.7410	0.7410	0.7410	0.7410
0.4175	1.61	700	0.6142	0.7810	0.7810	0.7810	0.7810
0.3626	1.73	750	0.5831	0.8004	0.8004	0.8004	0.8004
0.4816	1.84	800	0.5586	0.7891	0.7891	0.7891	0.7891
0.3257	1.96	850	0.5759	0.7991	0.7991	0.7991	0.7991
0.3111	2.07	900	0.6100	0.7931	0.7931	0.7931	0.7931
0.2052	2.19	950	0.5674	0.8111	0.8111	0.8111	0.8111
0.2273	2.3	1000	0.5975	0.8017	0.8017	0.8017	0.8017
0.3007	2.42	1050	0.5714	0.8204	0.8204	0.8204	0.8204
0.2812	2.53	1100	0.6081	0.8004	0.8004	0.8004	0.8004
0.2661	2.65	1150	0.5653	0.8224	0.8224	0.8224	0.8224
0.1796	2.76	1200	0.5447	0.8338	0.8338	0.8338	0.8338
0.1882	2.88	1250	0.5357	0.8284	0.8284	0.8284	0.8284
0.1596	3.0	1300	0.5296	0.8344	0.8344	0.8344	0.8344
0.075	3.11	1350	0.5876	0.8198	0.8198	0.8198	0.8198
0.1128	3.23	1400	0.5612	0.8338	0.8338	0.8338	0.8338
0.0677	3.34	1450	0.5911	0.8331	0.8331	0.8331	0.8331
0.0794	3.46	1500	0.5971	0.8304	0.8304	0.8304	0.8304
0.0367	3.57	1550	0.5634	0.8378	0.8378	0.8378	0.8378
0.0279	3.69	1600	0.5674	0.8391	0.8391	0.8391	0.8391
0.0216	3.8	1650	0.5777	0.8358	0.8358	0.8358	0.8358
0.0161	3.92	1700	0.5608	0.8438	0.8438	0.8438	0.8438

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0