SouthMemphis/vit-military-aircraft

vit-base-beans

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

• Loss: 0.3643
• Accuracy: 0.9027

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: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
3.5924	0.0620	100	3.5675	0.1927
3.0189	0.1239	200	3.0313	0.3047
2.5541	0.1859	300	2.5575	0.3956
2.114	0.2478	400	2.2332	0.4571
1.9624	0.3098	500	1.9455	0.5596
1.6749	0.3717	600	1.7370	0.5787
1.5852	0.4337	700	1.4947	0.6439
1.1875	0.4957	800	1.4151	0.6468
1.5114	0.5576	900	1.2709	0.6820
1.3122	0.6196	1000	1.1940	0.6939
1.0721	0.6815	1100	1.0757	0.7261
0.8249	0.7435	1200	0.9666	0.7576
0.7944	0.8055	1300	0.9101	0.7708
0.8032	0.8674	1400	0.9011	0.7691
0.7479	0.9294	1500	0.7409	0.8067
0.5997	0.9913	1600	0.7326	0.8110
0.5005	1.0533	1700	0.6769	0.8211
0.4107	1.1152	1800	0.6375	0.8374
0.4596	1.1772	1900	0.6302	0.8304
0.2544	1.2392	2000	0.5805	0.8400
0.2983	1.3011	2100	0.5480	0.8501
0.3214	1.3631	2200	0.5053	0.8683
0.2384	1.4250	2300	0.4929	0.8713
0.2397	1.4870	2400	0.4664	0.8742
0.3448	1.5489	2500	0.4690	0.8755
0.3129	1.6109	2600	0.4351	0.8843
0.1027	1.6729	2700	0.4311	0.8846
0.2086	1.7348	2800	0.4088	0.8897
0.1683	1.7968	2900	0.4133	0.8919
0.2767	1.8587	3000	0.3851	0.8964
0.1582	1.9207	3100	0.3703	0.9018
0.1421	1.9827	3200	0.3643	0.9027

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.2.0
• Tokenizers 0.19.1