SF-RHS-DA

This model is a fine-tuned version of MBZUAI/swiftformer-xs on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.5233
• Accuracy: 0.8224

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: 3e-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
0.693	0.99	35	0.6927	0.6822
0.6919	2.0	71	0.6910	0.6916
0.6872	2.99	106	0.6821	0.6822
0.6613	4.0	142	0.6552	0.6542
0.6196	4.99	177	0.6403	0.6168
0.5695	6.0	213	0.6128	0.6449
0.5436	6.99	248	0.6765	0.5701
0.4836	8.0	284	0.6075	0.6542
0.4902	8.99	319	0.5788	0.6355
0.4759	10.0	355	0.5284	0.7196
0.4746	10.99	390	0.5532	0.6822
0.4067	12.0	426	0.5356	0.7383
0.4138	12.99	461	0.5042	0.7477
0.3752	14.0	497	0.5063	0.7383
0.4158	14.99	532	0.4952	0.7570
0.3646	16.0	568	0.5440	0.7383
0.3644	16.99	603	0.5146	0.7757
0.3411	18.0	639	0.5208	0.7757
0.3052	18.99	674	0.5785	0.7383
0.3398	20.0	710	0.5366	0.7383
0.3103	20.99	745	0.5751	0.7290
0.3168	22.0	781	0.5194	0.7664
0.2927	22.99	816	0.5008	0.7944
0.2874	24.0	852	0.5216	0.7944
0.3021	24.99	887	0.5695	0.7570
0.2978	26.0	923	0.5643	0.7570
0.2743	26.99	958	0.5767	0.7570
0.2753	28.0	994	0.5125	0.7664
0.2773	28.99	1029	0.5246	0.7664
0.2775	30.0	1065	0.5473	0.7850
0.268	30.99	1100	0.5286	0.7664
0.2586	32.0	1136	0.5233	0.7850
0.2458	32.99	1171	0.5451	0.7757
0.2524	34.0	1207	0.5268	0.7850
0.2438	34.99	1242	0.5228	0.7757
0.2429	36.0	1278	0.5391	0.7664
0.2689	36.99	1313	0.5237	0.7850
0.2362	38.0	1349	0.5561	0.7664
0.2656	38.99	1384	0.5233	0.8224
0.264	39.44	1400	0.5112	0.8037

Framework versions

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