beit-base-patch16-224-DMAE

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: 0.6703
• Accuracy: 0.7609

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: 0.00015
• 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	1.3117	0.4565
No log	2.0	7	1.2539	0.4565
1.2875	2.86	10	1.2355	0.4565
1.2875	4.0	14	1.1827	0.4783
1.2875	4.86	17	1.0758	0.6522
1.1369	6.0	21	1.0942	0.5
1.1369	6.86	24	1.1312	0.5217
1.1369	8.0	28	1.0906	0.5217
1.1009	8.86	31	0.8704	0.6957
1.1009	10.0	35	1.0023	0.5870
1.1009	10.86	38	1.0288	0.5870
0.9152	12.0	42	0.7874	0.7174
0.9152	12.86	45	0.7166	0.7174
0.9152	14.0	49	0.7269	0.6957
0.8444	14.86	52	0.8481	0.6957
0.8444	16.0	56	0.7589	0.6304
0.8444	16.86	59	0.7590	0.6304
0.8085	18.0	63	0.8320	0.6304
0.8085	18.86	66	0.7469	0.7391
0.6941	20.0	70	0.8337	0.6304
0.6941	20.86	73	0.7928	0.7174
0.6941	22.0	77	0.8765	0.6522
0.5822	22.86	80	0.7139	0.7174
0.5822	24.0	84	0.7477	0.6957
0.5822	24.86	87	0.6987	0.7174
0.5174	26.0	91	0.6815	0.7391
0.5174	26.86	94	0.7332	0.7174
0.5174	28.0	98	0.6582	0.7391
0.48	28.86	101	0.7273	0.7391
0.48	30.0	105	0.7595	0.6957
0.48	30.86	108	0.7136	0.7391
0.4159	32.0	112	0.6703	0.7609
0.4159	32.86	115	0.6736	0.7609
0.4159	34.0	119	0.6866	0.7609
0.3472	34.29	120	0.6873	0.7609

Framework versions

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