beit-base-patch16-224-OT-alt

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

• Loss: 0.4274
• Accuracy: 0.9032

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.91	5	1.7093	0.1774
1.7744	2.0	11	1.6178	0.1774
1.7744	2.91	16	1.4730	0.1774
1.5823	4.0	22	1.2754	0.1774
1.5823	4.91	27	1.1455	0.5645
1.27	6.0	33	1.0147	0.6290
1.27	6.91	38	0.9790	0.5484
1.079	8.0	44	1.0474	0.4516
1.079	8.91	49	0.8796	0.7581
1.005	10.0	55	0.7759	0.7742
0.8479	10.91	60	0.7421	0.8226
0.8479	12.0	66	0.6760	0.8548
0.7695	12.91	71	0.5933	0.8387
0.7695	14.0	77	0.6372	0.7742
0.6591	14.91	82	0.5653	0.8387
0.6591	16.0	88	0.4950	0.8710
0.5675	16.91	93	0.5040	0.8226
0.5675	18.0	99	0.4274	0.9032
0.5134	18.91	104	0.4617	0.8548
0.4418	20.0	110	0.4245	0.8871
0.4418	20.91	115	0.4922	0.8387
0.402	22.0	121	0.5112	0.8226
0.402	22.91	126	0.4696	0.8548
0.4039	24.0	132	0.4014	0.8710
0.4039	24.91	137	0.5006	0.8226
0.4216	26.0	143	0.5351	0.8548
0.4216	26.91	148	0.5203	0.8548
0.3593	28.0	154	0.4082	0.8548
0.3593	28.91	159	0.4017	0.8710
0.3638	30.0	165	0.4068	0.8871
0.3509	30.91	170	0.3991	0.8871
0.3509	32.0	176	0.3965	0.8710
0.3426	32.91	181	0.3921	0.8710
0.3426	34.0	187	0.3998	0.8710
0.3253	34.91	192	0.4102	0.8871
0.3253	36.0	198	0.4081	0.8871
0.3085	36.36	200	0.4083	0.8871

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu118
• Datasets 2.16.1
• Tokenizers 0.15.0