Training in progress, step 500
Browse files- README.md +364 -0
- config.json +32 -0
- model.safetensors +3 -0
- training_args.bin +3 -0
README.md
ADDED
|
@@ -0,0 +1,364 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
base_model: aubmindlab/bert-base-arabertv02
|
| 4 |
+
tags:
|
| 5 |
+
- generated_from_trainer
|
| 6 |
+
model-index:
|
| 7 |
+
- name: Arabic_CrossPrompt_FineTuningAraBERT_noAug_TestTask4_vocabulary
|
| 8 |
+
results: []
|
| 9 |
+
---
|
| 10 |
+
|
| 11 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 12 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 13 |
+
|
| 14 |
+
# Arabic_CrossPrompt_FineTuningAraBERT_noAug_TestTask4_vocabulary
|
| 15 |
+
|
| 16 |
+
This model is a fine-tuned version of [aubmindlab/bert-base-arabertv02](https://huggingface.co/aubmindlab/bert-base-arabertv02) on the None dataset.
|
| 17 |
+
It achieves the following results on the evaluation set:
|
| 18 |
+
- Loss: 0.4296
|
| 19 |
+
- Qwk: 0.6237
|
| 20 |
+
- Mse: 0.4296
|
| 21 |
+
- Rmse: 0.6555
|
| 22 |
+
|
| 23 |
+
## Model description
|
| 24 |
+
|
| 25 |
+
More information needed
|
| 26 |
+
|
| 27 |
+
## Intended uses & limitations
|
| 28 |
+
|
| 29 |
+
More information needed
|
| 30 |
+
|
| 31 |
+
## Training and evaluation data
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Training procedure
|
| 36 |
+
|
| 37 |
+
### Training hyperparameters
|
| 38 |
+
|
| 39 |
+
The following hyperparameters were used during training:
|
| 40 |
+
- learning_rate: 2e-05
|
| 41 |
+
- train_batch_size: 8
|
| 42 |
+
- eval_batch_size: 8
|
| 43 |
+
- seed: 42
|
| 44 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 45 |
+
- lr_scheduler_type: linear
|
| 46 |
+
- num_epochs: 100
|
| 47 |
+
|
| 48 |
+
### Training results
|
| 49 |
+
|
| 50 |
+
| Training Loss | Epoch | Step | Validation Loss | Qwk | Mse | Rmse |
|
| 51 |
+
|:-------------:|:------:|:----:|:---------------:|:------:|:------:|:------:|
|
| 52 |
+
| No log | 0.0187 | 2 | 5.1151 | 0.0068 | 5.1151 | 2.2617 |
|
| 53 |
+
| No log | 0.0374 | 4 | 3.5702 | 0.0652 | 3.5702 | 1.8895 |
|
| 54 |
+
| No log | 0.0561 | 6 | 1.9152 | 0.1201 | 1.9152 | 1.3839 |
|
| 55 |
+
| No log | 0.0748 | 8 | 0.9999 | 0.1753 | 0.9999 | 1.0000 |
|
| 56 |
+
| No log | 0.0935 | 10 | 0.8377 | 0.1317 | 0.8377 | 0.9153 |
|
| 57 |
+
| No log | 0.1121 | 12 | 0.7951 | 0.2575 | 0.7951 | 0.8917 |
|
| 58 |
+
| No log | 0.1308 | 14 | 0.8093 | 0.1858 | 0.8093 | 0.8996 |
|
| 59 |
+
| No log | 0.1495 | 16 | 0.7734 | 0.1304 | 0.7734 | 0.8794 |
|
| 60 |
+
| No log | 0.1682 | 18 | 0.7411 | 0.1491 | 0.7411 | 0.8609 |
|
| 61 |
+
| No log | 0.1869 | 20 | 0.7567 | 0.2319 | 0.7567 | 0.8699 |
|
| 62 |
+
| No log | 0.2056 | 22 | 0.7552 | 0.3521 | 0.7552 | 0.8690 |
|
| 63 |
+
| No log | 0.2243 | 24 | 0.7195 | 0.3650 | 0.7195 | 0.8482 |
|
| 64 |
+
| No log | 0.2430 | 26 | 0.7216 | 0.4096 | 0.7216 | 0.8494 |
|
| 65 |
+
| No log | 0.2617 | 28 | 0.6824 | 0.4251 | 0.6824 | 0.8261 |
|
| 66 |
+
| No log | 0.2804 | 30 | 0.6116 | 0.4286 | 0.6116 | 0.7820 |
|
| 67 |
+
| No log | 0.2991 | 32 | 0.6070 | 0.4357 | 0.6070 | 0.7791 |
|
| 68 |
+
| No log | 0.3178 | 34 | 0.5655 | 0.5077 | 0.5655 | 0.7520 |
|
| 69 |
+
| No log | 0.3364 | 36 | 0.5314 | 0.5237 | 0.5314 | 0.7290 |
|
| 70 |
+
| No log | 0.3551 | 38 | 0.5509 | 0.5343 | 0.5509 | 0.7422 |
|
| 71 |
+
| No log | 0.3738 | 40 | 0.5430 | 0.5290 | 0.5430 | 0.7369 |
|
| 72 |
+
| No log | 0.3925 | 42 | 0.4972 | 0.5650 | 0.4972 | 0.7051 |
|
| 73 |
+
| No log | 0.4112 | 44 | 0.5113 | 0.5426 | 0.5113 | 0.7150 |
|
| 74 |
+
| No log | 0.4299 | 46 | 0.5414 | 0.5683 | 0.5414 | 0.7358 |
|
| 75 |
+
| No log | 0.4486 | 48 | 0.5381 | 0.5784 | 0.5381 | 0.7335 |
|
| 76 |
+
| No log | 0.4673 | 50 | 0.5352 | 0.5833 | 0.5352 | 0.7316 |
|
| 77 |
+
| No log | 0.4860 | 52 | 0.5309 | 0.5825 | 0.5309 | 0.7287 |
|
| 78 |
+
| No log | 0.5047 | 54 | 0.5245 | 0.5480 | 0.5245 | 0.7242 |
|
| 79 |
+
| No log | 0.5234 | 56 | 0.5343 | 0.4985 | 0.5343 | 0.7309 |
|
| 80 |
+
| No log | 0.5421 | 58 | 0.5758 | 0.4740 | 0.5758 | 0.7588 |
|
| 81 |
+
| No log | 0.5607 | 60 | 0.5621 | 0.4643 | 0.5621 | 0.7497 |
|
| 82 |
+
| No log | 0.5794 | 62 | 0.5016 | 0.5045 | 0.5016 | 0.7083 |
|
| 83 |
+
| No log | 0.5981 | 64 | 0.5002 | 0.5483 | 0.5002 | 0.7073 |
|
| 84 |
+
| No log | 0.6168 | 66 | 0.4606 | 0.5896 | 0.4606 | 0.6787 |
|
| 85 |
+
| No log | 0.6355 | 68 | 0.4393 | 0.6448 | 0.4393 | 0.6628 |
|
| 86 |
+
| No log | 0.6542 | 70 | 0.4239 | 0.6044 | 0.4239 | 0.6511 |
|
| 87 |
+
| No log | 0.6729 | 72 | 0.4257 | 0.6037 | 0.4257 | 0.6524 |
|
| 88 |
+
| No log | 0.6916 | 74 | 0.4841 | 0.5658 | 0.4841 | 0.6957 |
|
| 89 |
+
| No log | 0.7103 | 76 | 0.5178 | 0.5473 | 0.5178 | 0.7196 |
|
| 90 |
+
| No log | 0.7290 | 78 | 0.4747 | 0.5645 | 0.4747 | 0.6890 |
|
| 91 |
+
| No log | 0.7477 | 80 | 0.4316 | 0.6240 | 0.4316 | 0.6570 |
|
| 92 |
+
| No log | 0.7664 | 82 | 0.4240 | 0.6501 | 0.4240 | 0.6511 |
|
| 93 |
+
| No log | 0.7850 | 84 | 0.4667 | 0.6671 | 0.4667 | 0.6831 |
|
| 94 |
+
| No log | 0.8037 | 86 | 0.4828 | 0.6706 | 0.4828 | 0.6949 |
|
| 95 |
+
| No log | 0.8224 | 88 | 0.4280 | 0.6956 | 0.4280 | 0.6542 |
|
| 96 |
+
| No log | 0.8411 | 90 | 0.4324 | 0.6494 | 0.4324 | 0.6576 |
|
| 97 |
+
| No log | 0.8598 | 92 | 0.3981 | 0.6621 | 0.3981 | 0.6309 |
|
| 98 |
+
| No log | 0.8785 | 94 | 0.4367 | 0.6497 | 0.4367 | 0.6608 |
|
| 99 |
+
| No log | 0.8972 | 96 | 0.4323 | 0.6446 | 0.4323 | 0.6575 |
|
| 100 |
+
| No log | 0.9159 | 98 | 0.4346 | 0.6169 | 0.4346 | 0.6592 |
|
| 101 |
+
| No log | 0.9346 | 100 | 0.4479 | 0.6199 | 0.4479 | 0.6692 |
|
| 102 |
+
| No log | 0.9533 | 102 | 0.4237 | 0.6225 | 0.4237 | 0.6509 |
|
| 103 |
+
| No log | 0.9720 | 104 | 0.4681 | 0.6090 | 0.4681 | 0.6842 |
|
| 104 |
+
| No log | 0.9907 | 106 | 0.5188 | 0.5808 | 0.5188 | 0.7203 |
|
| 105 |
+
| No log | 1.0093 | 108 | 0.4809 | 0.5984 | 0.4809 | 0.6935 |
|
| 106 |
+
| No log | 1.0280 | 110 | 0.4232 | 0.6335 | 0.4232 | 0.6505 |
|
| 107 |
+
| No log | 1.0467 | 112 | 0.4360 | 0.6017 | 0.4360 | 0.6603 |
|
| 108 |
+
| No log | 1.0654 | 114 | 0.4424 | 0.5964 | 0.4424 | 0.6651 |
|
| 109 |
+
| No log | 1.0841 | 116 | 0.4270 | 0.6521 | 0.4270 | 0.6535 |
|
| 110 |
+
| No log | 1.1028 | 118 | 0.4615 | 0.6251 | 0.4615 | 0.6794 |
|
| 111 |
+
| No log | 1.1215 | 120 | 0.5120 | 0.6079 | 0.5120 | 0.7155 |
|
| 112 |
+
| No log | 1.1402 | 122 | 0.4763 | 0.6152 | 0.4763 | 0.6902 |
|
| 113 |
+
| No log | 1.1589 | 124 | 0.4339 | 0.6675 | 0.4339 | 0.6587 |
|
| 114 |
+
| No log | 1.1776 | 126 | 0.4792 | 0.6642 | 0.4792 | 0.6922 |
|
| 115 |
+
| No log | 1.1963 | 128 | 0.4370 | 0.6533 | 0.4370 | 0.6610 |
|
| 116 |
+
| No log | 1.2150 | 130 | 0.4735 | 0.6585 | 0.4735 | 0.6881 |
|
| 117 |
+
| No log | 1.2336 | 132 | 0.5326 | 0.6054 | 0.5326 | 0.7298 |
|
| 118 |
+
| No log | 1.2523 | 134 | 0.4769 | 0.6112 | 0.4769 | 0.6905 |
|
| 119 |
+
| No log | 1.2710 | 136 | 0.4556 | 0.6353 | 0.4556 | 0.6750 |
|
| 120 |
+
| No log | 1.2897 | 138 | 0.7194 | 0.5264 | 0.7194 | 0.8482 |
|
| 121 |
+
| No log | 1.3084 | 140 | 0.7037 | 0.5179 | 0.7037 | 0.8389 |
|
| 122 |
+
| No log | 1.3271 | 142 | 0.6053 | 0.4987 | 0.6053 | 0.7780 |
|
| 123 |
+
| No log | 1.3458 | 144 | 0.5330 | 0.5247 | 0.5330 | 0.7301 |
|
| 124 |
+
| No log | 1.3645 | 146 | 0.5008 | 0.5601 | 0.5008 | 0.7077 |
|
| 125 |
+
| No log | 1.3832 | 148 | 0.4687 | 0.5768 | 0.4687 | 0.6846 |
|
| 126 |
+
| No log | 1.4019 | 150 | 0.4397 | 0.6651 | 0.4397 | 0.6631 |
|
| 127 |
+
| No log | 1.4206 | 152 | 0.4456 | 0.6656 | 0.4456 | 0.6676 |
|
| 128 |
+
| No log | 1.4393 | 154 | 0.4633 | 0.6863 | 0.4633 | 0.6807 |
|
| 129 |
+
| No log | 1.4579 | 156 | 0.5210 | 0.6235 | 0.5210 | 0.7218 |
|
| 130 |
+
| No log | 1.4766 | 158 | 0.6128 | 0.5996 | 0.6128 | 0.7828 |
|
| 131 |
+
| No log | 1.4953 | 160 | 0.5522 | 0.6054 | 0.5522 | 0.7431 |
|
| 132 |
+
| No log | 1.5140 | 162 | 0.4503 | 0.6364 | 0.4503 | 0.6710 |
|
| 133 |
+
| No log | 1.5327 | 164 | 0.5143 | 0.5957 | 0.5143 | 0.7172 |
|
| 134 |
+
| No log | 1.5514 | 166 | 0.5861 | 0.5595 | 0.5861 | 0.7655 |
|
| 135 |
+
| No log | 1.5701 | 168 | 0.4833 | 0.6169 | 0.4833 | 0.6952 |
|
| 136 |
+
| No log | 1.5888 | 170 | 0.4341 | 0.6043 | 0.4341 | 0.6589 |
|
| 137 |
+
| No log | 1.6075 | 172 | 0.4456 | 0.5935 | 0.4456 | 0.6675 |
|
| 138 |
+
| No log | 1.6262 | 174 | 0.4673 | 0.5670 | 0.4673 | 0.6836 |
|
| 139 |
+
| No log | 1.6449 | 176 | 0.4628 | 0.5890 | 0.4628 | 0.6803 |
|
| 140 |
+
| No log | 1.6636 | 178 | 0.4717 | 0.6367 | 0.4717 | 0.6868 |
|
| 141 |
+
| No log | 1.6822 | 180 | 0.6264 | 0.6031 | 0.6264 | 0.7914 |
|
| 142 |
+
| No log | 1.7009 | 182 | 0.8957 | 0.5608 | 0.8957 | 0.9464 |
|
| 143 |
+
| No log | 1.7196 | 184 | 0.9124 | 0.5436 | 0.9124 | 0.9552 |
|
| 144 |
+
| No log | 1.7383 | 186 | 0.6478 | 0.5944 | 0.6478 | 0.8049 |
|
| 145 |
+
| No log | 1.7570 | 188 | 0.4457 | 0.6657 | 0.4457 | 0.6676 |
|
| 146 |
+
| No log | 1.7757 | 190 | 0.4624 | 0.6249 | 0.4624 | 0.6800 |
|
| 147 |
+
| No log | 1.7944 | 192 | 0.5610 | 0.5215 | 0.5610 | 0.7490 |
|
| 148 |
+
| No log | 1.8131 | 194 | 0.4946 | 0.5463 | 0.4946 | 0.7033 |
|
| 149 |
+
| No log | 1.8318 | 196 | 0.4401 | 0.6351 | 0.4401 | 0.6634 |
|
| 150 |
+
| No log | 1.8505 | 198 | 0.4069 | 0.6567 | 0.4069 | 0.6379 |
|
| 151 |
+
| No log | 1.8692 | 200 | 0.3985 | 0.7022 | 0.3985 | 0.6312 |
|
| 152 |
+
| No log | 1.8879 | 202 | 0.3985 | 0.6996 | 0.3985 | 0.6313 |
|
| 153 |
+
| No log | 1.9065 | 204 | 0.4031 | 0.6838 | 0.4031 | 0.6349 |
|
| 154 |
+
| No log | 1.9252 | 206 | 0.3788 | 0.7174 | 0.3788 | 0.6155 |
|
| 155 |
+
| No log | 1.9439 | 208 | 0.3671 | 0.7107 | 0.3671 | 0.6059 |
|
| 156 |
+
| No log | 1.9626 | 210 | 0.3665 | 0.6840 | 0.3665 | 0.6054 |
|
| 157 |
+
| No log | 1.9813 | 212 | 0.3654 | 0.7132 | 0.3654 | 0.6045 |
|
| 158 |
+
| No log | 2.0 | 214 | 0.3746 | 0.6697 | 0.3746 | 0.6120 |
|
| 159 |
+
| No log | 2.0187 | 216 | 0.4697 | 0.6798 | 0.4697 | 0.6853 |
|
| 160 |
+
| No log | 2.0374 | 218 | 0.4888 | 0.6470 | 0.4888 | 0.6992 |
|
| 161 |
+
| No log | 2.0561 | 220 | 0.3865 | 0.7110 | 0.3865 | 0.6217 |
|
| 162 |
+
| No log | 2.0748 | 222 | 0.5405 | 0.6280 | 0.5405 | 0.7352 |
|
| 163 |
+
| No log | 2.0935 | 224 | 0.6085 | 0.6104 | 0.6085 | 0.7801 |
|
| 164 |
+
| No log | 2.1121 | 226 | 0.4186 | 0.7247 | 0.4186 | 0.6470 |
|
| 165 |
+
| No log | 2.1308 | 228 | 0.4729 | 0.6658 | 0.4729 | 0.6876 |
|
| 166 |
+
| No log | 2.1495 | 230 | 0.5579 | 0.6153 | 0.5579 | 0.7470 |
|
| 167 |
+
| No log | 2.1682 | 232 | 0.5132 | 0.6208 | 0.5132 | 0.7164 |
|
| 168 |
+
| No log | 2.1869 | 234 | 0.4816 | 0.6515 | 0.4816 | 0.6940 |
|
| 169 |
+
| No log | 2.2056 | 236 | 0.4244 | 0.6734 | 0.4244 | 0.6515 |
|
| 170 |
+
| No log | 2.2243 | 238 | 0.4065 | 0.7224 | 0.4065 | 0.6376 |
|
| 171 |
+
| No log | 2.2430 | 240 | 0.5097 | 0.6560 | 0.5097 | 0.7139 |
|
| 172 |
+
| No log | 2.2617 | 242 | 0.4721 | 0.6718 | 0.4721 | 0.6871 |
|
| 173 |
+
| No log | 2.2804 | 244 | 0.4052 | 0.6489 | 0.4052 | 0.6366 |
|
| 174 |
+
| No log | 2.2991 | 246 | 0.4194 | 0.6121 | 0.4194 | 0.6476 |
|
| 175 |
+
| No log | 2.3178 | 248 | 0.4305 | 0.5678 | 0.4305 | 0.6562 |
|
| 176 |
+
| No log | 2.3364 | 250 | 0.4314 | 0.5619 | 0.4314 | 0.6568 |
|
| 177 |
+
| No log | 2.3551 | 252 | 0.4499 | 0.5918 | 0.4499 | 0.6707 |
|
| 178 |
+
| No log | 2.3738 | 254 | 0.4893 | 0.6071 | 0.4893 | 0.6995 |
|
| 179 |
+
| No log | 2.3925 | 256 | 0.4782 | 0.6235 | 0.4782 | 0.6915 |
|
| 180 |
+
| No log | 2.4112 | 258 | 0.4547 | 0.6885 | 0.4547 | 0.6743 |
|
| 181 |
+
| No log | 2.4299 | 260 | 0.3911 | 0.7325 | 0.3911 | 0.6254 |
|
| 182 |
+
| No log | 2.4486 | 262 | 0.3953 | 0.7072 | 0.3953 | 0.6288 |
|
| 183 |
+
| No log | 2.4673 | 264 | 0.4589 | 0.6617 | 0.4589 | 0.6774 |
|
| 184 |
+
| No log | 2.4860 | 266 | 0.4169 | 0.6659 | 0.4169 | 0.6457 |
|
| 185 |
+
| No log | 2.5047 | 268 | 0.4307 | 0.6563 | 0.4307 | 0.6563 |
|
| 186 |
+
| No log | 2.5234 | 270 | 0.5538 | 0.6284 | 0.5538 | 0.7442 |
|
| 187 |
+
| No log | 2.5421 | 272 | 0.5301 | 0.6229 | 0.5301 | 0.7280 |
|
| 188 |
+
| No log | 2.5607 | 274 | 0.5289 | 0.6397 | 0.5289 | 0.7272 |
|
| 189 |
+
| No log | 2.5794 | 276 | 0.4600 | 0.6356 | 0.4600 | 0.6782 |
|
| 190 |
+
| No log | 2.5981 | 278 | 0.4485 | 0.6435 | 0.4485 | 0.6697 |
|
| 191 |
+
| No log | 2.6168 | 280 | 0.4442 | 0.6564 | 0.4442 | 0.6665 |
|
| 192 |
+
| No log | 2.6355 | 282 | 0.4288 | 0.6609 | 0.4288 | 0.6548 |
|
| 193 |
+
| No log | 2.6542 | 284 | 0.4716 | 0.6707 | 0.4716 | 0.6868 |
|
| 194 |
+
| No log | 2.6729 | 286 | 0.5568 | 0.6183 | 0.5568 | 0.7462 |
|
| 195 |
+
| No log | 2.6916 | 288 | 0.5851 | 0.6510 | 0.5851 | 0.7650 |
|
| 196 |
+
| No log | 2.7103 | 290 | 0.5592 | 0.6731 | 0.5592 | 0.7478 |
|
| 197 |
+
| No log | 2.7290 | 292 | 0.5077 | 0.6882 | 0.5077 | 0.7125 |
|
| 198 |
+
| No log | 2.7477 | 294 | 0.5106 | 0.7038 | 0.5106 | 0.7146 |
|
| 199 |
+
| No log | 2.7664 | 296 | 0.4709 | 0.7127 | 0.4709 | 0.6862 |
|
| 200 |
+
| No log | 2.7850 | 298 | 0.4347 | 0.7074 | 0.4347 | 0.6593 |
|
| 201 |
+
| No log | 2.8037 | 300 | 0.4121 | 0.6976 | 0.4121 | 0.6420 |
|
| 202 |
+
| No log | 2.8224 | 302 | 0.3938 | 0.7243 | 0.3938 | 0.6275 |
|
| 203 |
+
| No log | 2.8411 | 304 | 0.3892 | 0.6857 | 0.3892 | 0.6238 |
|
| 204 |
+
| No log | 2.8598 | 306 | 0.4401 | 0.6372 | 0.4401 | 0.6634 |
|
| 205 |
+
| No log | 2.8785 | 308 | 0.4208 | 0.6078 | 0.4208 | 0.6487 |
|
| 206 |
+
| No log | 2.8972 | 310 | 0.4404 | 0.6034 | 0.4404 | 0.6636 |
|
| 207 |
+
| No log | 2.9159 | 312 | 0.4830 | 0.6108 | 0.4830 | 0.6950 |
|
| 208 |
+
| No log | 2.9346 | 314 | 0.4747 | 0.6250 | 0.4747 | 0.6890 |
|
| 209 |
+
| No log | 2.9533 | 316 | 0.3987 | 0.6449 | 0.3987 | 0.6314 |
|
| 210 |
+
| No log | 2.9720 | 318 | 0.3882 | 0.6814 | 0.3882 | 0.6231 |
|
| 211 |
+
| No log | 2.9907 | 320 | 0.3927 | 0.6684 | 0.3927 | 0.6266 |
|
| 212 |
+
| No log | 3.0093 | 322 | 0.4357 | 0.6411 | 0.4357 | 0.6601 |
|
| 213 |
+
| No log | 3.0280 | 324 | 0.5752 | 0.6158 | 0.5752 | 0.7584 |
|
| 214 |
+
| No log | 3.0467 | 326 | 0.6273 | 0.6003 | 0.6273 | 0.7920 |
|
| 215 |
+
| No log | 3.0654 | 328 | 0.5258 | 0.6445 | 0.5258 | 0.7251 |
|
| 216 |
+
| No log | 3.0841 | 330 | 0.4486 | 0.6853 | 0.4486 | 0.6698 |
|
| 217 |
+
| No log | 3.1028 | 332 | 0.4256 | 0.6661 | 0.4256 | 0.6524 |
|
| 218 |
+
| No log | 3.1215 | 334 | 0.4146 | 0.6973 | 0.4146 | 0.6439 |
|
| 219 |
+
| No log | 3.1402 | 336 | 0.4124 | 0.7023 | 0.4124 | 0.6422 |
|
| 220 |
+
| No log | 3.1589 | 338 | 0.4016 | 0.7063 | 0.4016 | 0.6337 |
|
| 221 |
+
| No log | 3.1776 | 340 | 0.3812 | 0.7011 | 0.3812 | 0.6174 |
|
| 222 |
+
| No log | 3.1963 | 342 | 0.4411 | 0.6541 | 0.4411 | 0.6641 |
|
| 223 |
+
| No log | 3.2150 | 344 | 0.6437 | 0.6371 | 0.6437 | 0.8023 |
|
| 224 |
+
| No log | 3.2336 | 346 | 0.6617 | 0.6283 | 0.6617 | 0.8134 |
|
| 225 |
+
| No log | 3.2523 | 348 | 0.4898 | 0.6699 | 0.4898 | 0.6999 |
|
| 226 |
+
| No log | 3.2710 | 350 | 0.3686 | 0.7194 | 0.3686 | 0.6071 |
|
| 227 |
+
| No log | 3.2897 | 352 | 0.4297 | 0.6561 | 0.4297 | 0.6555 |
|
| 228 |
+
| No log | 3.3084 | 354 | 0.4180 | 0.6669 | 0.4180 | 0.6465 |
|
| 229 |
+
| No log | 3.3271 | 356 | 0.3951 | 0.6398 | 0.3951 | 0.6286 |
|
| 230 |
+
| No log | 3.3458 | 358 | 0.4012 | 0.6495 | 0.4012 | 0.6334 |
|
| 231 |
+
| No log | 3.3645 | 360 | 0.4840 | 0.6226 | 0.4840 | 0.6957 |
|
| 232 |
+
| No log | 3.3832 | 362 | 0.5195 | 0.5963 | 0.5195 | 0.7208 |
|
| 233 |
+
| No log | 3.4019 | 364 | 0.4473 | 0.6482 | 0.4473 | 0.6688 |
|
| 234 |
+
| No log | 3.4206 | 366 | 0.4205 | 0.6628 | 0.4205 | 0.6484 |
|
| 235 |
+
| No log | 3.4393 | 368 | 0.4265 | 0.6626 | 0.4265 | 0.6531 |
|
| 236 |
+
| No log | 3.4579 | 370 | 0.4358 | 0.6450 | 0.4358 | 0.6602 |
|
| 237 |
+
| No log | 3.4766 | 372 | 0.4570 | 0.6250 | 0.4570 | 0.6760 |
|
| 238 |
+
| No log | 3.4953 | 374 | 0.4516 | 0.6302 | 0.4516 | 0.6720 |
|
| 239 |
+
| No log | 3.5140 | 376 | 0.4279 | 0.7192 | 0.4279 | 0.6542 |
|
| 240 |
+
| No log | 3.5327 | 378 | 0.4179 | 0.7316 | 0.4179 | 0.6465 |
|
| 241 |
+
| No log | 3.5514 | 380 | 0.4167 | 0.7198 | 0.4167 | 0.6455 |
|
| 242 |
+
| No log | 3.5701 | 382 | 0.4853 | 0.6814 | 0.4853 | 0.6967 |
|
| 243 |
+
| No log | 3.5888 | 384 | 0.4450 | 0.6744 | 0.4450 | 0.6671 |
|
| 244 |
+
| No log | 3.6075 | 386 | 0.3983 | 0.7165 | 0.3983 | 0.6311 |
|
| 245 |
+
| No log | 3.6262 | 388 | 0.4075 | 0.6999 | 0.4075 | 0.6384 |
|
| 246 |
+
| No log | 3.6449 | 390 | 0.4164 | 0.7053 | 0.4164 | 0.6453 |
|
| 247 |
+
| No log | 3.6636 | 392 | 0.4562 | 0.6958 | 0.4562 | 0.6755 |
|
| 248 |
+
| No log | 3.6822 | 394 | 0.4705 | 0.6739 | 0.4705 | 0.6859 |
|
| 249 |
+
| No log | 3.7009 | 396 | 0.4286 | 0.6753 | 0.4286 | 0.6547 |
|
| 250 |
+
| No log | 3.7196 | 398 | 0.4178 | 0.6713 | 0.4178 | 0.6464 |
|
| 251 |
+
| No log | 3.7383 | 400 | 0.4059 | 0.6994 | 0.4059 | 0.6371 |
|
| 252 |
+
| No log | 3.7570 | 402 | 0.4110 | 0.6991 | 0.4110 | 0.6411 |
|
| 253 |
+
| No log | 3.7757 | 404 | 0.4145 | 0.7093 | 0.4145 | 0.6438 |
|
| 254 |
+
| No log | 3.7944 | 406 | 0.4293 | 0.7072 | 0.4293 | 0.6552 |
|
| 255 |
+
| No log | 3.8131 | 408 | 0.4543 | 0.6910 | 0.4543 | 0.6740 |
|
| 256 |
+
| No log | 3.8318 | 410 | 0.4774 | 0.6191 | 0.4774 | 0.6909 |
|
| 257 |
+
| No log | 3.8505 | 412 | 0.4401 | 0.6362 | 0.4401 | 0.6634 |
|
| 258 |
+
| No log | 3.8692 | 414 | 0.4164 | 0.6687 | 0.4164 | 0.6453 |
|
| 259 |
+
| No log | 3.8879 | 416 | 0.4056 | 0.6838 | 0.4056 | 0.6368 |
|
| 260 |
+
| No log | 3.9065 | 418 | 0.3971 | 0.7096 | 0.3971 | 0.6302 |
|
| 261 |
+
| No log | 3.9252 | 420 | 0.4302 | 0.7090 | 0.4302 | 0.6559 |
|
| 262 |
+
| No log | 3.9439 | 422 | 0.4382 | 0.7258 | 0.4382 | 0.6620 |
|
| 263 |
+
| No log | 3.9626 | 424 | 0.5416 | 0.7074 | 0.5416 | 0.7360 |
|
| 264 |
+
| No log | 3.9813 | 426 | 0.6486 | 0.6516 | 0.6486 | 0.8054 |
|
| 265 |
+
| No log | 4.0 | 428 | 0.5780 | 0.6419 | 0.5780 | 0.7603 |
|
| 266 |
+
| No log | 4.0187 | 430 | 0.4314 | 0.6825 | 0.4314 | 0.6568 |
|
| 267 |
+
| No log | 4.0374 | 432 | 0.4034 | 0.6911 | 0.4034 | 0.6351 |
|
| 268 |
+
| No log | 4.0561 | 434 | 0.4087 | 0.6692 | 0.4087 | 0.6393 |
|
| 269 |
+
| No log | 4.0748 | 436 | 0.4253 | 0.6452 | 0.4253 | 0.6522 |
|
| 270 |
+
| No log | 4.0935 | 438 | 0.4497 | 0.6673 | 0.4497 | 0.6706 |
|
| 271 |
+
| No log | 4.1121 | 440 | 0.5068 | 0.6194 | 0.5068 | 0.7119 |
|
| 272 |
+
| No log | 4.1308 | 442 | 0.5921 | 0.6564 | 0.5921 | 0.7695 |
|
| 273 |
+
| No log | 4.1495 | 444 | 0.6122 | 0.6776 | 0.6122 | 0.7824 |
|
| 274 |
+
| No log | 4.1682 | 446 | 0.4810 | 0.7068 | 0.4810 | 0.6935 |
|
| 275 |
+
| No log | 4.1869 | 448 | 0.4615 | 0.7046 | 0.4615 | 0.6793 |
|
| 276 |
+
| No log | 4.2056 | 450 | 0.5127 | 0.6824 | 0.5127 | 0.7160 |
|
| 277 |
+
| No log | 4.2243 | 452 | 0.5892 | 0.6763 | 0.5892 | 0.7676 |
|
| 278 |
+
| No log | 4.2430 | 454 | 0.4735 | 0.6674 | 0.4735 | 0.6881 |
|
| 279 |
+
| No log | 4.2617 | 456 | 0.4087 | 0.7050 | 0.4087 | 0.6393 |
|
| 280 |
+
| No log | 4.2804 | 458 | 0.4068 | 0.6862 | 0.4068 | 0.6378 |
|
| 281 |
+
| No log | 4.2991 | 460 | 0.3943 | 0.6731 | 0.3943 | 0.6279 |
|
| 282 |
+
| No log | 4.3178 | 462 | 0.4512 | 0.6465 | 0.4512 | 0.6717 |
|
| 283 |
+
| No log | 4.3364 | 464 | 0.5694 | 0.6233 | 0.5694 | 0.7546 |
|
| 284 |
+
| No log | 4.3551 | 466 | 0.5487 | 0.6255 | 0.5487 | 0.7407 |
|
| 285 |
+
| No log | 4.3738 | 468 | 0.4411 | 0.6840 | 0.4411 | 0.6641 |
|
| 286 |
+
| No log | 4.3925 | 470 | 0.3963 | 0.6964 | 0.3963 | 0.6295 |
|
| 287 |
+
| No log | 4.4112 | 472 | 0.3996 | 0.7308 | 0.3996 | 0.6321 |
|
| 288 |
+
| No log | 4.4299 | 474 | 0.4298 | 0.7266 | 0.4298 | 0.6556 |
|
| 289 |
+
| No log | 4.4486 | 476 | 0.5479 | 0.7124 | 0.5479 | 0.7402 |
|
| 290 |
+
| No log | 4.4673 | 478 | 0.7202 | 0.6823 | 0.7202 | 0.8486 |
|
| 291 |
+
| No log | 4.4860 | 480 | 0.7451 | 0.6525 | 0.7451 | 0.8632 |
|
| 292 |
+
| No log | 4.5047 | 482 | 0.5639 | 0.6767 | 0.5639 | 0.7509 |
|
| 293 |
+
| No log | 4.5234 | 484 | 0.3987 | 0.7034 | 0.3987 | 0.6314 |
|
| 294 |
+
| No log | 4.5421 | 486 | 0.4199 | 0.6867 | 0.4199 | 0.6480 |
|
| 295 |
+
| No log | 4.5607 | 488 | 0.4477 | 0.6663 | 0.4477 | 0.6691 |
|
| 296 |
+
| No log | 4.5794 | 490 | 0.4484 | 0.6542 | 0.4484 | 0.6696 |
|
| 297 |
+
| No log | 4.5981 | 492 | 0.4205 | 0.6265 | 0.4205 | 0.6484 |
|
| 298 |
+
| No log | 4.6168 | 494 | 0.4341 | 0.6127 | 0.4341 | 0.6589 |
|
| 299 |
+
| No log | 4.6355 | 496 | 0.5222 | 0.5907 | 0.5222 | 0.7226 |
|
| 300 |
+
| No log | 4.6542 | 498 | 0.5172 | 0.6166 | 0.5172 | 0.7191 |
|
| 301 |
+
| 0.5012 | 4.6729 | 500 | 0.4727 | 0.6292 | 0.4727 | 0.6875 |
|
| 302 |
+
| 0.5012 | 4.6916 | 502 | 0.3990 | 0.6992 | 0.3990 | 0.6317 |
|
| 303 |
+
| 0.5012 | 4.7103 | 504 | 0.3947 | 0.6996 | 0.3947 | 0.6282 |
|
| 304 |
+
| 0.5012 | 4.7290 | 506 | 0.4120 | 0.7279 | 0.4120 | 0.6419 |
|
| 305 |
+
| 0.5012 | 4.7477 | 508 | 0.4700 | 0.7198 | 0.4700 | 0.6856 |
|
| 306 |
+
| 0.5012 | 4.7664 | 510 | 0.6670 | 0.6538 | 0.6670 | 0.8167 |
|
| 307 |
+
| 0.5012 | 4.7850 | 512 | 0.7010 | 0.6284 | 0.7010 | 0.8372 |
|
| 308 |
+
| 0.5012 | 4.8037 | 514 | 0.6348 | 0.6636 | 0.6348 | 0.7967 |
|
| 309 |
+
| 0.5012 | 4.8224 | 516 | 0.5567 | 0.6783 | 0.5567 | 0.7461 |
|
| 310 |
+
| 0.5012 | 4.8411 | 518 | 0.4841 | 0.6988 | 0.4841 | 0.6958 |
|
| 311 |
+
| 0.5012 | 4.8598 | 520 | 0.4408 | 0.6900 | 0.4408 | 0.6640 |
|
| 312 |
+
| 0.5012 | 4.8785 | 522 | 0.4594 | 0.6874 | 0.4594 | 0.6778 |
|
| 313 |
+
| 0.5012 | 4.8972 | 524 | 0.4504 | 0.6723 | 0.4504 | 0.6711 |
|
| 314 |
+
| 0.5012 | 4.9159 | 526 | 0.4226 | 0.6992 | 0.4226 | 0.6501 |
|
| 315 |
+
| 0.5012 | 4.9346 | 528 | 0.4024 | 0.7080 | 0.4024 | 0.6344 |
|
| 316 |
+
| 0.5012 | 4.9533 | 530 | 0.4302 | 0.6822 | 0.4302 | 0.6559 |
|
| 317 |
+
| 0.5012 | 4.9720 | 532 | 0.4002 | 0.7310 | 0.4002 | 0.6326 |
|
| 318 |
+
| 0.5012 | 4.9907 | 534 | 0.4026 | 0.7421 | 0.4026 | 0.6345 |
|
| 319 |
+
| 0.5012 | 5.0093 | 536 | 0.4240 | 0.7018 | 0.4240 | 0.6511 |
|
| 320 |
+
| 0.5012 | 5.0280 | 538 | 0.4530 | 0.7052 | 0.4530 | 0.6731 |
|
| 321 |
+
| 0.5012 | 5.0467 | 540 | 0.4630 | 0.6924 | 0.4630 | 0.6804 |
|
| 322 |
+
| 0.5012 | 5.0654 | 542 | 0.4842 | 0.6902 | 0.4842 | 0.6958 |
|
| 323 |
+
| 0.5012 | 5.0841 | 544 | 0.4241 | 0.6801 | 0.4241 | 0.6512 |
|
| 324 |
+
| 0.5012 | 5.1028 | 546 | 0.4143 | 0.6889 | 0.4143 | 0.6437 |
|
| 325 |
+
| 0.5012 | 5.1215 | 548 | 0.4461 | 0.6622 | 0.4461 | 0.6679 |
|
| 326 |
+
| 0.5012 | 5.1402 | 550 | 0.6099 | 0.6130 | 0.6099 | 0.7810 |
|
| 327 |
+
| 0.5012 | 5.1589 | 552 | 0.6845 | 0.5782 | 0.6845 | 0.8274 |
|
| 328 |
+
| 0.5012 | 5.1776 | 554 | 0.5880 | 0.6234 | 0.5880 | 0.7668 |
|
| 329 |
+
| 0.5012 | 5.1963 | 556 | 0.4428 | 0.6640 | 0.4428 | 0.6654 |
|
| 330 |
+
| 0.5012 | 5.2150 | 558 | 0.4055 | 0.6661 | 0.4055 | 0.6368 |
|
| 331 |
+
| 0.5012 | 5.2336 | 560 | 0.4284 | 0.6819 | 0.4284 | 0.6545 |
|
| 332 |
+
| 0.5012 | 5.2523 | 562 | 0.5121 | 0.6582 | 0.5121 | 0.7156 |
|
| 333 |
+
| 0.5012 | 5.2710 | 564 | 0.5140 | 0.6611 | 0.5140 | 0.7170 |
|
| 334 |
+
| 0.5012 | 5.2897 | 566 | 0.4347 | 0.7092 | 0.4347 | 0.6593 |
|
| 335 |
+
| 0.5012 | 5.3084 | 568 | 0.4080 | 0.7014 | 0.4080 | 0.6387 |
|
| 336 |
+
| 0.5012 | 5.3271 | 570 | 0.4263 | 0.6988 | 0.4263 | 0.6529 |
|
| 337 |
+
| 0.5012 | 5.3458 | 572 | 0.5351 | 0.6726 | 0.5351 | 0.7315 |
|
| 338 |
+
| 0.5012 | 5.3645 | 574 | 0.5778 | 0.6509 | 0.5778 | 0.7601 |
|
| 339 |
+
| 0.5012 | 5.3832 | 576 | 0.4587 | 0.6751 | 0.4587 | 0.6773 |
|
| 340 |
+
| 0.5012 | 5.4019 | 578 | 0.3949 | 0.6931 | 0.3949 | 0.6284 |
|
| 341 |
+
| 0.5012 | 5.4206 | 580 | 0.3902 | 0.7102 | 0.3902 | 0.6246 |
|
| 342 |
+
| 0.5012 | 5.4393 | 582 | 0.3968 | 0.6880 | 0.3968 | 0.6299 |
|
| 343 |
+
| 0.5012 | 5.4579 | 584 | 0.4255 | 0.6605 | 0.4255 | 0.6523 |
|
| 344 |
+
| 0.5012 | 5.4766 | 586 | 0.4429 | 0.6609 | 0.4429 | 0.6655 |
|
| 345 |
+
| 0.5012 | 5.4953 | 588 | 0.4504 | 0.6833 | 0.4504 | 0.6711 |
|
| 346 |
+
| 0.5012 | 5.5140 | 590 | 0.5034 | 0.6570 | 0.5034 | 0.7095 |
|
| 347 |
+
| 0.5012 | 5.5327 | 592 | 0.5353 | 0.6700 | 0.5353 | 0.7316 |
|
| 348 |
+
| 0.5012 | 5.5514 | 594 | 0.5068 | 0.6587 | 0.5068 | 0.7119 |
|
| 349 |
+
| 0.5012 | 5.5701 | 596 | 0.4610 | 0.6931 | 0.4610 | 0.6790 |
|
| 350 |
+
| 0.5012 | 5.5888 | 598 | 0.4478 | 0.6768 | 0.4478 | 0.6692 |
|
| 351 |
+
| 0.5012 | 5.6075 | 600 | 0.4716 | 0.6652 | 0.4716 | 0.6867 |
|
| 352 |
+
| 0.5012 | 5.6262 | 602 | 0.5286 | 0.6250 | 0.5286 | 0.7270 |
|
| 353 |
+
| 0.5012 | 5.6449 | 604 | 0.5020 | 0.6095 | 0.5020 | 0.7085 |
|
| 354 |
+
| 0.5012 | 5.6636 | 606 | 0.4523 | 0.5837 | 0.4523 | 0.6726 |
|
| 355 |
+
| 0.5012 | 5.6822 | 608 | 0.4292 | 0.6267 | 0.4292 | 0.6551 |
|
| 356 |
+
| 0.5012 | 5.7009 | 610 | 0.4296 | 0.6237 | 0.4296 | 0.6555 |
|
| 357 |
+
|
| 358 |
+
|
| 359 |
+
### Framework versions
|
| 360 |
+
|
| 361 |
+
- Transformers 4.44.2
|
| 362 |
+
- Pytorch 2.4.0+cu118
|
| 363 |
+
- Datasets 2.21.0
|
| 364 |
+
- Tokenizers 0.19.1
|
config.json
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "aubmindlab/bert-base-arabertv02",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"BertForSequenceClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"classifier_dropout": null,
|
| 8 |
+
"hidden_act": "gelu",
|
| 9 |
+
"hidden_dropout_prob": 0.1,
|
| 10 |
+
"hidden_size": 768,
|
| 11 |
+
"id2label": {
|
| 12 |
+
"0": "LABEL_0"
|
| 13 |
+
},
|
| 14 |
+
"initializer_range": 0.02,
|
| 15 |
+
"intermediate_size": 3072,
|
| 16 |
+
"label2id": {
|
| 17 |
+
"LABEL_0": 0
|
| 18 |
+
},
|
| 19 |
+
"layer_norm_eps": 1e-12,
|
| 20 |
+
"max_position_embeddings": 512,
|
| 21 |
+
"model_type": "bert",
|
| 22 |
+
"num_attention_heads": 12,
|
| 23 |
+
"num_hidden_layers": 12,
|
| 24 |
+
"pad_token_id": 0,
|
| 25 |
+
"position_embedding_type": "absolute",
|
| 26 |
+
"problem_type": "regression",
|
| 27 |
+
"torch_dtype": "float32",
|
| 28 |
+
"transformers_version": "4.44.2",
|
| 29 |
+
"type_vocab_size": 2,
|
| 30 |
+
"use_cache": true,
|
| 31 |
+
"vocab_size": 64000
|
| 32 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62c3968238146fde6c0987ec92e863304675b10fdbcccd980a0aef56933059b7
|
| 3 |
+
size 540799996
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:57c19cab095c6d4169e0eb2af1e2e2892982a4fb859831abfd1e6db798be13c3
|
| 3 |
+
size 5240
|