resnet-152-finetuned_resnet152-adam-optimizer5e-4-autotags

This model is a fine-tuned version of microsoft/resnet-152 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.2399
• Accuracy: 0.9305

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: 0.0005
• 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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.4009	0.99	65	2.1414	0.3971
0.9201	1.99	130	0.8123	0.7210
0.7575	2.99	195	0.5730	0.8124
0.4792	3.99	260	0.4166	0.8648
0.4253	4.99	325	0.3811	0.8810
0.3331	5.99	390	0.4290	0.8705
0.2347	6.99	455	0.4600	0.8952
0.1732	7.99	520	0.3018	0.8924
0.1777	8.99	585	0.4851	0.8914
0.1298	9.99	650	0.2941	0.92
0.1164	10.99	715	0.3915	0.9095
0.1284	11.99	780	0.3701	0.9152
0.0986	12.99	845	0.3416	0.9171
0.0944	13.99	910	0.3145	0.9210
0.0929	14.99	975	0.2677	0.9229
0.1014	15.99	1040	0.2745	0.9295
0.0971	16.99	1105	0.2932	0.9267
0.0691	17.99	1170	0.2174	0.9333
0.0557	18.99	1235	0.2233	0.9324
0.06	19.99	1300	0.2399	0.9305

Framework versions

• Transformers 4.25.1
• Pytorch 1.13.1+cu117
• Datasets 2.11.0
• Tokenizers 0.13.2