beit-base-patch16-224-OT-2

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: 0.5047
• Accuracy: 0.8387

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: 3.5e-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	1.8532	0.0806
1.7494	2.0	11	1.7818	0.0806
1.7494	2.91	16	1.6613	0.0806
1.6235	4.0	22	1.4651	0.0806
1.6235	4.91	27	1.3293	0.0806
1.3836	6.0	33	1.2034	0.5161
1.3836	6.91	38	1.1748	0.3710
1.2192	8.0	44	1.0815	0.4677
1.2192	8.91	49	1.0238	0.5
1.093	10.0	55	1.0225	0.4516
0.9938	10.91	60	0.9650	0.6452
0.9938	12.0	66	0.9314	0.6935
0.9235	12.91	71	0.9490	0.6452
0.9235	14.0	77	0.8234	0.7258
0.8258	14.91	82	0.8159	0.7258
0.8258	16.0	88	0.7514	0.7419
0.716	16.91	93	0.7469	0.7419
0.716	18.0	99	0.6734	0.7903
0.6026	18.91	104	0.6926	0.7581
0.5725	20.0	110	0.7952	0.7258
0.5725	20.91	115	0.6284	0.7742
0.554	22.0	121	0.6317	0.7742
0.554	22.91	126	0.6361	0.7419
0.5162	24.0	132	0.5501	0.8226
0.5162	24.91	137	0.6278	0.7581
0.4768	26.0	143	0.5868	0.7903
0.4768	26.91	148	0.5047	0.8387
0.4488	28.0	154	0.5264	0.7903
0.4488	28.91	159	0.4942	0.8387
0.4281	30.0	165	0.5127	0.8387
0.4126	30.91	170	0.5027	0.8387
0.4126	32.0	176	0.5387	0.7742
0.4326	32.91	181	0.5251	0.7903
0.4326	34.0	187	0.5091	0.8065
0.3765	34.91	192	0.5142	0.8065
0.3765	36.0	198	0.5142	0.7903
0.3913	36.36	200	0.5144	0.7903

Framework versions

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