BEiT-DMAE-DA

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: 0.4816
• Accuracy: 0.9130

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
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.3475	0.96	11	1.3279	0.3261
1.1875	2.0	23	1.1320	0.3478
0.9998	2.96	34	0.9957	0.5435
0.8836	4.0	46	0.8436	0.5870
0.7593	4.96	57	0.7904	0.6304
0.6939	6.0	69	0.6656	0.8261
0.4924	6.96	80	0.6724	0.6739
0.4444	8.0	92	0.5951	0.7826
0.337	8.96	103	0.5222	0.8261
0.3213	10.0	115	0.6814	0.8043
0.2689	10.96	126	0.5913	0.7826
0.2538	12.0	138	0.6228	0.7826
0.2032	12.96	149	0.6992	0.7609
0.2152	14.0	161	0.7730	0.7609
0.1713	14.96	172	0.7762	0.7609
0.2042	16.0	184	0.7652	0.7174
0.1668	16.96	195	0.5512	0.8478
0.1743	18.0	207	0.7311	0.7826
0.1226	18.96	218	0.7115	0.8043
0.1537	20.0	230	0.6800	0.7609
0.1311	20.96	241	0.5864	0.8478
0.1335	22.0	253	0.6346	0.8261
0.0981	22.96	264	0.6541	0.8043
0.1248	24.0	276	0.7017	0.8261
0.1183	24.96	287	0.6964	0.8261
0.0946	26.0	299	0.6450	0.8261
0.0957	26.96	310	0.7057	0.8043
0.1692	28.0	322	0.6635	0.8043
0.0967	28.96	333	0.5040	0.8696
0.094	30.0	345	0.5588	0.8913
0.0843	30.96	356	0.5398	0.8696
0.0851	32.0	368	0.5806	0.8478
0.0955	32.96	379	0.4816	0.9130
0.1157	34.0	391	0.5289	0.8696
0.072	34.96	402	0.5657	0.8913
0.091	36.0	414	0.5566	0.8478
0.0891	36.96	425	0.5729	0.8478
0.0732	38.0	437	0.5915	0.8261
0.0647	38.26	440	0.5902	0.8261

Framework versions

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