yujiepan/internal.swin-base-food101-int8-structured30.56

swin-food101-jpqd-1to2r1-epo7-finetuned-student

This model is a fine-tuned version of skylord/swin-finetuned-food101 on the food101 dataset. It achieves the following results on the evaluation set:

• Loss: 0.1947
• Accuracy: 0.9213

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: 16
• eval_batch_size: 128
• 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
• num_epochs: 7.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.2342	0.42	500	0.1993	0.9099
0.2891	0.84	1000	0.1912	0.9137
67.4995	1.27	1500	66.4760	0.8035
109.8398	1.69	2000	109.5154	0.4499
0.6337	2.11	2500	0.4865	0.8826
0.6605	2.54	3000	0.3551	0.9013
0.4013	2.96	3500	0.3176	0.9044
0.3949	3.38	4000	0.2839	0.9079
0.4632	3.8	4500	0.2652	0.9118
0.3717	4.23	5000	0.2459	0.9147
0.3308	4.65	5500	0.2439	0.9159
0.4232	5.07	6000	0.2259	0.9169
0.3426	5.49	6500	0.2147	0.9199
0.331	5.92	7000	0.2086	0.9189
0.3032	6.34	7500	0.2036	0.9201
0.3393	6.76	8000	0.1978	0.9204

Framework versions

• Transformers 4.26.0
• Pytorch 1.13.1+cu116
• Datasets 2.8.0
• Tokenizers 0.13.2