EElayoutlmv3_jordyvl_rvl_cdip_100_examples_per_class_2023-07-06_ent
This model is a fine-tuned version of microsoft/layoutlmv3-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.3493
- Accuracy: 0.645
- Exit 0 Accuracy: 0.1125
- Exit 1 Accuracy: 0.155
- Exit 2 Accuracy: 0.3775
- Exit 3 Accuracy: 0.5225
- Exit 4 Accuracy: 0.5875
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: 2e-05
- train_batch_size: 12
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 24
- total_train_batch_size: 288
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 60
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy | Exit 0 Accuracy | Exit 1 Accuracy | Exit 2 Accuracy | Exit 3 Accuracy | Exit 4 Accuracy |
---|---|---|---|---|---|---|---|---|---|
No log | 0.72 | 2 | 2.7604 | 0.1075 | 0.09 | 0.0675 | 0.1075 | 0.0625 | 0.0625 |
No log | 1.72 | 4 | 2.7329 | 0.1125 | 0.0725 | 0.065 | 0.13 | 0.0625 | 0.0625 |
No log | 2.72 | 6 | 2.6989 | 0.1325 | 0.08 | 0.06 | 0.1375 | 0.0625 | 0.0625 |
No log | 3.72 | 8 | 2.6608 | 0.17 | 0.08 | 0.0575 | 0.1375 | 0.0625 | 0.0625 |
No log | 4.72 | 10 | 2.6201 | 0.19 | 0.09 | 0.0525 | 0.1175 | 0.0625 | 0.0625 |
No log | 5.72 | 12 | 2.5813 | 0.2175 | 0.095 | 0.0825 | 0.1125 | 0.0675 | 0.0625 |
No log | 6.72 | 14 | 2.5503 | 0.215 | 0.0925 | 0.08 | 0.12 | 0.0825 | 0.0625 |
No log | 7.72 | 16 | 2.5289 | 0.23 | 0.09 | 0.0925 | 0.15 | 0.1025 | 0.0625 |
No log | 8.72 | 18 | 2.5344 | 0.245 | 0.0975 | 0.1 | 0.165 | 0.105 | 0.0675 |
No log | 9.72 | 20 | 2.5533 | 0.265 | 0.1 | 0.0975 | 0.185 | 0.09 | 0.1025 |
No log | 10.72 | 22 | 2.4567 | 0.29 | 0.0975 | 0.13 | 0.2 | 0.1 | 0.095 |
No log | 11.72 | 24 | 2.3982 | 0.3 | 0.1 | 0.12 | 0.205 | 0.1125 | 0.09 |
No log | 12.72 | 26 | 2.3722 | 0.3075 | 0.1025 | 0.1175 | 0.195 | 0.13 | 0.0825 |
No log | 13.72 | 28 | 2.3546 | 0.31 | 0.105 | 0.1225 | 0.1825 | 0.1425 | 0.085 |
No log | 14.72 | 30 | 2.3287 | 0.315 | 0.11 | 0.125 | 0.195 | 0.1775 | 0.095 |
No log | 15.72 | 32 | 2.2970 | 0.32 | 0.1075 | 0.13 | 0.2175 | 0.2275 | 0.1 |
No log | 16.72 | 34 | 2.2763 | 0.325 | 0.1075 | 0.14 | 0.225 | 0.2375 | 0.1075 |
No log | 17.72 | 36 | 2.3456 | 0.3075 | 0.105 | 0.14 | 0.2375 | 0.18 | 0.1275 |
No log | 18.72 | 38 | 2.3160 | 0.325 | 0.115 | 0.14 | 0.24 | 0.175 | 0.16 |
No log | 19.72 | 40 | 2.2257 | 0.33 | 0.1225 | 0.14 | 0.245 | 0.225 | 0.17 |
No log | 20.72 | 42 | 2.1769 | 0.355 | 0.125 | 0.1425 | 0.26 | 0.2725 | 0.135 |
No log | 21.72 | 44 | 2.1449 | 0.355 | 0.125 | 0.14 | 0.2725 | 0.3125 | 0.1175 |
No log | 22.72 | 46 | 2.1200 | 0.3675 | 0.125 | 0.1425 | 0.27 | 0.3125 | 0.115 |
No log | 23.72 | 48 | 2.0995 | 0.3725 | 0.1225 | 0.1425 | 0.2625 | 0.31 | 0.115 |
No log | 24.72 | 50 | 2.0769 | 0.3825 | 0.12 | 0.1425 | 0.2725 | 0.3375 | 0.1125 |
No log | 25.72 | 52 | 2.0473 | 0.3975 | 0.115 | 0.14 | 0.285 | 0.335 | 0.1325 |
No log | 26.72 | 54 | 2.0094 | 0.4075 | 0.115 | 0.14 | 0.2925 | 0.3075 | 0.1525 |
No log | 27.72 | 56 | 1.9660 | 0.435 | 0.1175 | 0.14 | 0.29 | 0.2725 | 0.21 |
No log | 28.72 | 58 | 1.9271 | 0.46 | 0.11 | 0.1425 | 0.3025 | 0.27 | 0.235 |
No log | 29.72 | 60 | 1.8910 | 0.4825 | 0.11 | 0.145 | 0.305 | 0.27 | 0.2525 |
No log | 30.72 | 62 | 1.8619 | 0.475 | 0.11 | 0.1475 | 0.3 | 0.2875 | 0.27 |
No log | 31.72 | 64 | 1.8215 | 0.5025 | 0.11 | 0.15 | 0.3025 | 0.305 | 0.325 |
No log | 32.72 | 66 | 1.7845 | 0.52 | 0.1125 | 0.15 | 0.3175 | 0.3225 | 0.3625 |
No log | 33.72 | 68 | 1.7509 | 0.5375 | 0.1125 | 0.15 | 0.325 | 0.3525 | 0.3975 |
No log | 34.72 | 70 | 1.7237 | 0.545 | 0.1075 | 0.15 | 0.3325 | 0.365 | 0.4275 |
No log | 35.72 | 72 | 1.6970 | 0.555 | 0.11 | 0.15 | 0.3275 | 0.4 | 0.4475 |
No log | 36.72 | 74 | 1.6512 | 0.57 | 0.1075 | 0.15 | 0.3225 | 0.4125 | 0.465 |
No log | 37.72 | 76 | 1.6212 | 0.5875 | 0.11 | 0.1525 | 0.3375 | 0.42 | 0.4775 |
No log | 38.72 | 78 | 1.5995 | 0.595 | 0.1125 | 0.15 | 0.34 | 0.4275 | 0.4975 |
No log | 39.72 | 80 | 1.5713 | 0.5925 | 0.115 | 0.15 | 0.35 | 0.4375 | 0.525 |
No log | 40.72 | 82 | 1.5551 | 0.5875 | 0.115 | 0.15 | 0.3525 | 0.4375 | 0.5325 |
No log | 41.72 | 84 | 1.5276 | 0.59 | 0.115 | 0.15 | 0.35 | 0.4575 | 0.5425 |
No log | 42.72 | 86 | 1.5050 | 0.5925 | 0.115 | 0.15 | 0.355 | 0.46 | 0.5425 |
No log | 43.72 | 88 | 1.4871 | 0.595 | 0.1125 | 0.1525 | 0.3625 | 0.47 | 0.5625 |
No log | 44.72 | 90 | 1.4712 | 0.5975 | 0.1125 | 0.1525 | 0.3675 | 0.4775 | 0.5525 |
No log | 45.72 | 92 | 1.4615 | 0.5975 | 0.1125 | 0.155 | 0.365 | 0.4825 | 0.555 |
No log | 46.72 | 94 | 1.4449 | 0.6075 | 0.1125 | 0.155 | 0.3625 | 0.4875 | 0.5575 |
No log | 47.72 | 96 | 1.4273 | 0.6175 | 0.1125 | 0.155 | 0.365 | 0.5025 | 0.565 |
No log | 48.72 | 98 | 1.4127 | 0.6225 | 0.1125 | 0.155 | 0.365 | 0.505 | 0.5725 |
No log | 49.72 | 100 | 1.4005 | 0.63 | 0.1125 | 0.155 | 0.3675 | 0.5125 | 0.575 |
No log | 50.72 | 102 | 1.3925 | 0.625 | 0.1125 | 0.155 | 0.37 | 0.5125 | 0.5725 |
No log | 51.72 | 104 | 1.3847 | 0.6325 | 0.1125 | 0.155 | 0.38 | 0.5175 | 0.57 |
No log | 52.72 | 106 | 1.3772 | 0.64 | 0.1125 | 0.155 | 0.38 | 0.515 | 0.57 |
No log | 53.72 | 108 | 1.3679 | 0.6425 | 0.1125 | 0.155 | 0.3775 | 0.52 | 0.5825 |
No log | 54.72 | 110 | 1.3595 | 0.6475 | 0.1125 | 0.155 | 0.3775 | 0.525 | 0.5825 |
No log | 55.72 | 112 | 1.3544 | 0.6425 | 0.1125 | 0.155 | 0.3775 | 0.5225 | 0.58 |
No log | 56.72 | 114 | 1.3515 | 0.6425 | 0.1125 | 0.155 | 0.375 | 0.52 | 0.5875 |
No log | 57.72 | 116 | 1.3500 | 0.6425 | 0.1125 | 0.155 | 0.3775 | 0.52 | 0.5925 |
No log | 58.72 | 118 | 1.3495 | 0.6425 | 0.1125 | 0.155 | 0.3775 | 0.5225 | 0.59 |
No log | 59.72 | 120 | 1.3493 | 0.645 | 0.1125 | 0.155 | 0.3775 | 0.5225 | 0.5875 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1.post200
- Datasets 2.9.0
- Tokenizers 0.13.2
- Downloads last month
- 9
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.