jialicheng/unlearn_cifar100_swin-base_random_label_6_42

42

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: 0.4605
• Accuracy: 0.887
• Dt Accuracy: 0.887
• Df Accuracy: 0.0247
• Unlearn Overall Accuracy: 0.9709
• 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: 42
• 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	0.5295	0.8873	0.2167	0.2167	None
1.3324	2.0	782	0.4877	0.891	0.2113	0.2113	None
1.1366	3.0	1173	0.5020	0.877	0.2331	0.2331	None
1.0197	4.0	1564	0.4836	0.851	0.2724	0.2724	None
1.0197	5.0	1955	0.4607	0.8253	0.3092	0.3092	None
0.9076	6.0	2346	0.4997	0.7397	0.4201	0.4201	None
0.8133	7.0	2737	0.4802	0.627	0.5457	0.5457	None
0.7391	8.0	3128	0.4875	0.5163	0.6491	0.6491	None
0.6743	9.0	3519	0.5113	0.4307	0.7164	0.7164	None
0.6743	10.0	3910	0.5024	0.3383	0.7846	0.7846	None
0.6002	11.0	4301	0.5063	0.258	0.8351	0.8351	None
0.551	12.0	4692	0.4918	0.214	0.8641	0.8641	None
0.5027	13.0	5083	0.4968	0.1667	0.8918	0.8918	None
0.5027	14.0	5474	0.4826	0.1327	0.9110	0.9110	None
0.454	15.0	5865	0.4752	0.0963	0.9312	0.9312	None
0.4188	16.0	6256	0.4738	0.0657	0.9479	0.9479	None
0.3824	17.0	6647	0.4686	0.0437	0.9599	0.9599	None
0.3579	18.0	7038	0.4766	0.028	0.9672	0.9672	None
0.3579	19.0	7429	0.4646	0.0267	0.9694	0.9694	None
0.3325	20.0	7820	0.4605	0.0247	0.9709	0.9709	None

Framework versions

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