swin-tiny-patch4-window7-224-finetuned-200k

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.4347
• Accuracy: 0.7961

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: 128
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 512
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.634	0.99	36	0.6243	0.6262
0.5551	1.99	72	0.5186	0.7250
0.5183	2.98	108	0.4826	0.7673
0.4854	4.0	145	0.5640	0.7261
0.4645	4.99	181	0.4598	0.7817
0.4655	5.99	217	0.4787	0.7786
0.4582	6.98	253	0.4483	0.7899
0.4415	8.0	290	0.4709	0.7765
0.4546	8.99	326	0.4717	0.7817
0.4566	9.99	362	0.4538	0.7951
0.4675	10.98	398	0.4491	0.7817
0.4449	12.0	435	0.4992	0.7652
0.4349	12.99	471	0.4627	0.7817
0.4253	13.99	507	0.4492	0.7858
0.4278	14.98	543	0.4442	0.7951
0.4567	16.0	580	0.4362	0.7899
0.4205	16.99	616	0.4550	0.7889
0.4233	17.99	652	0.4336	0.7909
0.4014	18.98	688	0.4565	0.7889
0.4176	20.0	725	0.4323	0.7940
0.411	20.99	761	0.4348	0.7951
0.4128	21.99	797	0.4378	0.7971
0.4045	22.98	833	0.4317	0.7951
0.4001	24.0	870	0.4452	0.7868
0.4061	24.99	906	0.4286	0.7920
0.4033	25.99	942	0.4306	0.7951
0.3953	26.98	978	0.4320	0.7920
0.3924	28.0	1015	0.4338	0.7940
0.4056	28.99	1051	0.4329	0.7930
0.4032	29.79	1080	0.4347	0.7961

Framework versions

• Transformers 4.33.3
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3