beit-base-patch16-224-DMAE-5e-5

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: 0.6236
• Accuracy: 0.8043

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	0.86	3	1.3816	0.4565
No log	2.0	7	1.3504	0.4565
1.3654	2.86	10	1.3075	0.4565
1.3654	4.0	14	1.2444	0.4565
1.3654	4.86	17	1.2089	0.4565
1.2386	6.0	21	1.1711	0.4565
1.2386	6.86	24	1.1492	0.4565
1.2386	8.0	28	1.1050	0.4783
1.132	8.86	31	1.0928	0.6522
1.132	10.0	35	1.0230	0.6739
1.132	10.86	38	0.9864	0.6522
1.015	12.0	42	1.0060	0.6739
1.015	12.86	45	0.8790	0.7174
1.015	14.0	49	0.8533	0.6739
0.9217	14.86	52	0.8203	0.6957
0.9217	16.0	56	0.8152	0.6739
0.9217	16.86	59	0.8598	0.6522
0.7989	18.0	63	0.7979	0.6957
0.7989	18.86	66	0.7555	0.6739
0.7309	20.0	70	0.7019	0.7609
0.7309	20.86	73	0.7254	0.7174
0.7309	22.0	77	0.7268	0.6739
0.6267	22.86	80	0.7096	0.6739
0.6267	24.0	84	0.7568	0.6739
0.6267	24.86	87	0.6565	0.7391
0.564	26.0	91	0.7035	0.7391
0.564	26.86	94	0.7081	0.7391
0.564	28.0	98	0.6236	0.8043
0.5625	28.86	101	0.6091	0.8043
0.5625	30.0	105	0.6523	0.7609
0.5625	30.86	108	0.6235	0.7826
0.5185	32.0	112	0.5974	0.7826
0.5185	32.86	115	0.5944	0.7826
0.5185	34.0	119	0.6005	0.7826
0.5005	34.29	120	0.6005	0.7826

Framework versions

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