<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
EElayoutlmv3_jordyvl_rvl_cdip_100_examples_per_class_2023-09-22_subgraphs_gates
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: 1.0995
- Accuracy: 0.705
- Exit 0 Accuracy: 0.065
- Exit 1 Accuracy: 0.0575
- Exit 2 Accuracy: 0.065
- Exit 3 Accuracy: 0.07
- Exit 4 Accuracy: 0.1575
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: 20
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 12
- total_train_batch_size: 240
- 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 | 0.9 | 3 | 2.7455 | 0.0975 | 0.0625 | 0.0625 | 0.0875 | 0.0625 | 0.0625 |
No log | 1.9 | 6 | 2.6867 | 0.1475 | 0.06 | 0.0625 | 0.0875 | 0.0625 | 0.0625 |
No log | 2.9 | 9 | 2.6527 | 0.18 | 0.0475 | 0.0575 | 0.09 | 0.0625 | 0.0625 |
No log | 3.9 | 12 | 2.6188 | 0.1975 | 0.035 | 0.0525 | 0.095 | 0.0625 | 0.0625 |
No log | 4.9 | 15 | 2.5827 | 0.22 | 0.035 | 0.05 | 0.0875 | 0.0625 | 0.0625 |
No log | 5.9 | 18 | 2.6265 | 0.1725 | 0.0475 | 0.0375 | 0.0775 | 0.0625 | 0.0825 |
No log | 6.9 | 21 | 2.5771 | 0.23 | 0.0525 | 0.0375 | 0.0775 | 0.0625 | 0.0975 |
No log | 7.9 | 24 | 2.4865 | 0.2675 | 0.055 | 0.0375 | 0.0675 | 0.0625 | 0.08 |
No log | 8.9 | 27 | 2.4184 | 0.25 | 0.0475 | 0.04 | 0.0625 | 0.0625 | 0.0775 |
No log | 9.9 | 30 | 2.3451 | 0.28 | 0.06 | 0.045 | 0.0625 | 0.0625 | 0.085 |
No log | 10.9 | 33 | 2.3027 | 0.3175 | 0.0675 | 0.045 | 0.0625 | 0.0625 | 0.0775 |
No log | 11.9 | 36 | 2.2168 | 0.3225 | 0.07 | 0.045 | 0.0625 | 0.0625 | 0.0775 |
No log | 12.9 | 39 | 2.1698 | 0.365 | 0.07 | 0.045 | 0.0625 | 0.0625 | 0.085 |
No log | 13.9 | 42 | 2.0647 | 0.4125 | 0.065 | 0.045 | 0.0625 | 0.0625 | 0.0875 |
No log | 14.9 | 45 | 2.0370 | 0.4125 | 0.0675 | 0.0475 | 0.0625 | 0.06 | 0.09 |
No log | 15.9 | 48 | 1.9523 | 0.4675 | 0.065 | 0.0475 | 0.0625 | 0.0575 | 0.0975 |
No log | 16.9 | 51 | 1.8758 | 0.4925 | 0.0675 | 0.05 | 0.0625 | 0.055 | 0.1075 |
No log | 17.9 | 54 | 1.7940 | 0.52 | 0.0675 | 0.05 | 0.0625 | 0.0525 | 0.0975 |
No log | 18.9 | 57 | 1.7417 | 0.5125 | 0.0675 | 0.05 | 0.0625 | 0.0525 | 0.125 |
No log | 19.9 | 60 | 1.6789 | 0.565 | 0.0675 | 0.05 | 0.0625 | 0.055 | 0.1325 |
No log | 20.9 | 63 | 1.6016 | 0.5825 | 0.0675 | 0.05 | 0.0625 | 0.0575 | 0.1525 |
No log | 21.9 | 66 | 1.5649 | 0.5775 | 0.065 | 0.05 | 0.0625 | 0.07 | 0.155 |
No log | 22.9 | 69 | 1.5164 | 0.6075 | 0.065 | 0.05 | 0.0625 | 0.0775 | 0.155 |
No log | 23.9 | 72 | 1.4614 | 0.6225 | 0.065 | 0.05 | 0.0625 | 0.085 | 0.1475 |
No log | 24.9 | 75 | 1.4301 | 0.62 | 0.065 | 0.05 | 0.0625 | 0.085 | 0.16 |
No log | 25.9 | 78 | 1.3935 | 0.645 | 0.065 | 0.0525 | 0.065 | 0.0775 | 0.1425 |
No log | 26.9 | 81 | 1.3777 | 0.64 | 0.065 | 0.0525 | 0.065 | 0.08 | 0.1625 |
No log | 27.9 | 84 | 1.3301 | 0.6525 | 0.065 | 0.0525 | 0.065 | 0.0825 | 0.16 |
No log | 28.9 | 87 | 1.3064 | 0.6775 | 0.065 | 0.0525 | 0.065 | 0.0825 | 0.1625 |
No log | 29.9 | 90 | 1.2689 | 0.675 | 0.065 | 0.0525 | 0.065 | 0.085 | 0.1625 |
No log | 30.9 | 93 | 1.2549 | 0.6575 | 0.065 | 0.0525 | 0.065 | 0.075 | 0.175 |
No log | 31.9 | 96 | 1.2440 | 0.66 | 0.065 | 0.055 | 0.065 | 0.08 | 0.1525 |
No log | 32.9 | 99 | 1.2250 | 0.685 | 0.065 | 0.055 | 0.065 | 0.09 | 0.1675 |
No log | 33.9 | 102 | 1.2243 | 0.665 | 0.065 | 0.055 | 0.065 | 0.0775 | 0.155 |
No log | 34.9 | 105 | 1.1960 | 0.6625 | 0.065 | 0.055 | 0.065 | 0.0825 | 0.1675 |
No log | 35.9 | 108 | 1.1879 | 0.685 | 0.065 | 0.055 | 0.065 | 0.08 | 0.155 |
No log | 36.9 | 111 | 1.1710 | 0.6825 | 0.065 | 0.055 | 0.065 | 0.0825 | 0.155 |
No log | 37.9 | 114 | 1.1739 | 0.6925 | 0.065 | 0.055 | 0.065 | 0.0675 | 0.1525 |
No log | 38.9 | 117 | 1.1414 | 0.7 | 0.065 | 0.055 | 0.065 | 0.0725 | 0.1475 |
No log | 39.9 | 120 | 1.1476 | 0.695 | 0.065 | 0.0575 | 0.065 | 0.0775 | 0.1525 |
No log | 40.9 | 123 | 1.1435 | 0.6875 | 0.065 | 0.0575 | 0.065 | 0.0825 | 0.15 |
No log | 41.9 | 126 | 1.1196 | 0.695 | 0.065 | 0.0575 | 0.065 | 0.075 | 0.165 |
No log | 42.9 | 129 | 1.1454 | 0.7 | 0.065 | 0.0575 | 0.065 | 0.0775 | 0.155 |
No log | 43.9 | 132 | 1.1133 | 0.7 | 0.065 | 0.0575 | 0.065 | 0.075 | 0.1575 |
No log | 44.9 | 135 | 1.1205 | 0.7125 | 0.065 | 0.0575 | 0.065 | 0.08 | 0.1625 |
No log | 45.9 | 138 | 1.1188 | 0.705 | 0.065 | 0.0575 | 0.065 | 0.0775 | 0.1525 |
No log | 46.9 | 141 | 1.1076 | 0.7025 | 0.065 | 0.0575 | 0.065 | 0.0775 | 0.155 |
No log | 47.9 | 144 | 1.1304 | 0.7075 | 0.065 | 0.0575 | 0.065 | 0.08 | 0.16 |
No log | 48.9 | 147 | 1.1029 | 0.7125 | 0.065 | 0.0575 | 0.065 | 0.08 | 0.16 |
No log | 49.9 | 150 | 1.0979 | 0.705 | 0.065 | 0.0575 | 0.065 | 0.08 | 0.16 |
No log | 50.9 | 153 | 1.1081 | 0.71 | 0.065 | 0.0575 | 0.065 | 0.0775 | 0.155 |
No log | 51.9 | 156 | 1.1023 | 0.71 | 0.065 | 0.0575 | 0.065 | 0.0725 | 0.16 |
No log | 52.9 | 159 | 1.1029 | 0.705 | 0.065 | 0.0575 | 0.065 | 0.07 | 0.16 |
No log | 53.9 | 162 | 1.1015 | 0.71 | 0.065 | 0.0575 | 0.065 | 0.07 | 0.155 |
No log | 54.9 | 165 | 1.0968 | 0.7125 | 0.065 | 0.0575 | 0.065 | 0.0725 | 0.155 |
No log | 55.9 | 168 | 1.0968 | 0.7125 | 0.065 | 0.0575 | 0.065 | 0.0725 | 0.16 |
No log | 56.9 | 171 | 1.1006 | 0.7025 | 0.065 | 0.0575 | 0.065 | 0.075 | 0.155 |
No log | 57.9 | 174 | 1.1024 | 0.7025 | 0.065 | 0.0575 | 0.065 | 0.07 | 0.1575 |
No log | 58.9 | 177 | 1.1006 | 0.705 | 0.065 | 0.0575 | 0.065 | 0.07 | 0.1575 |
No log | 59.9 | 180 | 1.0995 | 0.705 | 0.065 | 0.0575 | 0.065 | 0.07 | 0.1575 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1.post200
- Datasets 2.9.0
- Tokenizers 0.13.2