jialicheng/unlearn_cifar100_swin-base_random_label_8_87

87

This model is a fine-tuned version of microsoft/swin-base-patch4-window7-224 on the cifar100 dataset. It achieves the following results on the evaluation set:

• Loss: 0.4980
• Accuracy: 0.876
• Dt Accuracy: 0.876
• Df Accuracy: 0.0243
• Unlearn Overall Accuracy: 0.9654
• Unlearn Time: None

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: 0.0002
• train_batch_size: 128
• eval_batch_size: 256
• seed: 87
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Overall Accuracy	Unlearn Overall Accuracy	Time
No log	1.0	391	0.5345	0.8778	0.2314	0.2314	None
1.3864	2.0	782	0.5238	0.8808	0.2270	0.2270	None
1.1798	3.0	1173	0.5057	0.8642	0.2524	0.2524	None
1.0607	4.0	1564	0.4875	0.859	0.2604	0.2604	None
1.0607	5.0	1955	0.4741	0.8195	0.3173	0.3173	None
0.9712	6.0	2346	0.5035	0.709	0.4554	0.4554	None
0.8619	7.0	2737	0.5391	0.622	0.5479	0.5479	None
0.787	8.0	3128	0.5409	0.4895	0.6676	0.6676	None
0.7117	9.0	3519	0.5071	0.4577	0.6957	0.6957	None
0.7117	10.0	3910	0.5160	0.353	0.7726	0.7726	None
0.6285	11.0	4301	0.5175	0.287	0.8165	0.8165	None
0.5809	12.0	4692	0.5343	0.1998	0.8673	0.8673	None
0.532	13.0	5083	0.5431	0.1505	0.8954	0.8954	None
0.532	14.0	5474	0.5263	0.1328	0.9064	0.9064	None
0.4848	15.0	5865	0.5163	0.0757	0.9375	0.9375	None
0.4401	16.0	6256	0.5162	0.0777	0.9371	0.9371	None
0.4038	17.0	6647	0.5145	0.0488	0.9503	0.9503	None
0.367	18.0	7038	0.5007	0.0423	0.9563	0.9563	None
0.367	19.0	7429	0.4956	0.0267	0.9644	0.9644	None
0.341	20.0	7820	0.4980	0.0243	0.9654	0.9654	None

Framework versions

• Transformers 4.39.3
• Pytorch 2.2.2+cu118
• Datasets 2.18.0
• Tokenizers 0.15.2