beit-base-patch16-224-OT

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: 0.4801
• Accuracy: 0.8226

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	0.89	4	1.7603	0.1452
No log	2.0	9	1.6852	0.1452
1.7571	2.89	13	1.5655	0.1452
1.7571	4.0	18	1.3816	0.1452
1.5255	4.89	22	1.2599	0.3226
1.5255	6.0	27	1.1534	0.4839
1.2245	6.89	31	1.0641	0.4839
1.2245	8.0	36	1.0372	0.4355
1.0438	8.89	40	0.9988	0.4355
1.0438	10.0	45	0.9260	0.5161
1.0438	10.89	49	0.9085	0.7097
0.9727	12.0	54	0.8433	0.7258
0.9727	12.89	58	0.7529	0.7742
0.8469	14.0	63	0.7187	0.7581
0.8469	14.89	67	0.6806	0.7258
0.6908	16.0	72	0.6576	0.7581
0.6908	16.89	76	0.5742	0.7903
0.6064	18.0	81	0.6447	0.7581
0.6064	18.89	85	0.5602	0.7742
0.5303	20.0	90	0.4943	0.7903
0.5303	20.89	94	0.5304	0.7903
0.5303	22.0	99	0.4801	0.8226
0.4903	22.89	103	0.4849	0.8226
0.4903	24.0	108	0.5710	0.7742
0.4261	24.89	112	0.4803	0.7903
0.4261	26.0	117	0.5671	0.7258
0.4122	26.89	121	0.4585	0.8065
0.4122	28.0	126	0.5910	0.7097
0.3739	28.89	130	0.5821	0.7581
0.3739	30.0	135	0.5329	0.7742
0.3739	30.89	139	0.4423	0.8226
0.3896	32.0	144	0.4716	0.7581
0.3896	32.89	148	0.4786	0.7903
0.3472	34.0	153	0.4538	0.7903
0.3472	34.89	157	0.4553	0.7903
0.3349	35.56	160	0.4528	0.7903

Framework versions

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