vuiseng9/swin-base-food101-int8-structured44.5-20eph

jpqd-swin-b-20eph-r1.00-s2e5-mock-main-merge-pr2

This model is a fine-tuned version of microsoft/swin-base-patch4-window7-224 on the food101 dataset. It achieves the following results on the evaluation set:

• Loss: 0.2715
• Accuracy: 0.9179

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
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
5.1452	0.42	500	5.4928	0.6440
0.9839	0.84	1000	0.7956	0.8580
0.8533	1.27	1500	0.4392	0.8911
0.6123	1.69	2000	0.3768	0.8983
12.3076	2.11	2500	12.0798	0.8953
49.301	2.54	3000	48.6292	0.8343
75.6345	2.96	3500	75.7027	0.6777
94.2556	3.38	4000	93.5852	0.5604
103.3226	3.8	4500	103.1255	0.5702
107.3423	4.23	5000	107.9250	0.5359
108.9013	4.65	5500	108.5225	0.5882
2.045	5.07	6000	1.1149	0.8154
1.3377	5.49	6500	0.6747	0.8665
0.7565	5.92	7000	0.5814	0.8765
0.7493	6.34	7500	0.5460	0.8840
0.7693	6.76	8000	0.5109	0.8851
0.6082	7.19	8500	0.4893	0.8895
0.7575	7.61	9000	0.4521	0.8943
0.7943	8.03	9500	0.4465	0.8941
0.5521	8.45	10000	0.4119	0.8967
0.6536	8.88	10500	0.4071	0.9010
0.5164	9.3	11000	0.3945	0.9010
0.6687	9.72	11500	0.3884	0.9030
0.4374	10.14	12000	0.3764	0.9040
0.7326	10.57	12500	0.3678	0.9060
0.6148	10.99	13000	0.3602	0.9057
0.6068	11.41	13500	0.3566	0.9075
0.6105	11.83	14000	0.3456	0.9074
0.5277	12.26	14500	0.3383	0.9107
0.5255	12.68	15000	0.3328	0.9097
0.4536	13.1	15500	0.3268	0.9108
0.5337	13.52	16000	0.3256	0.9107
0.5299	13.95	16500	0.3161	0.9124
0.3037	14.37	17000	0.3162	0.9123
0.4171	14.79	17500	0.3078	0.9124
0.5375	15.22	18000	0.3002	0.9116
0.2722	15.64	18500	0.2953	0.9134
0.3684	16.06	19000	0.2960	0.9137
0.4369	16.48	19500	0.2918	0.9150
0.3346	16.91	20000	0.2856	0.9171
0.3645	17.33	20500	0.2856	0.9162
0.4475	17.75	21000	0.2833	0.9157
0.2553	18.17	21500	0.2788	0.9167
0.5098	18.6	22000	0.2766	0.9164
0.4149	19.02	22500	0.2732	0.9177
0.3737	19.44	23000	0.2734	0.9181
0.325	19.86	23500	0.2715	0.9176

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu117
• Datasets 2.10.1
• Tokenizers 0.13.2