bert-base-multilingual-cased-finetuned-multilingual-nli_newdata_oneepoch
This model is a fine-tuned version of bert-base-multilingual-cased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.7647
- Accuracy: 0.6853
- Precision: 0.6932
- Recall: 0.6853
- F1: 0.6847
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: 3e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 1
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 |
---|---|---|---|---|---|---|---|
0.9394 | 0.04 | 500 | 0.9044 | 0.592 | 0.5985 | 0.592 | 0.5917 |
0.8603 | 0.08 | 1000 | 0.9159 | 0.579 | 0.6210 | 0.579 | 0.5739 |
0.8293 | 0.11 | 1500 | 0.8520 | 0.6214 | 0.6278 | 0.6214 | 0.6215 |
0.8042 | 0.15 | 2000 | 0.8085 | 0.6418 | 0.6439 | 0.6418 | 0.6414 |
0.7945 | 0.19 | 2500 | 0.8251 | 0.6319 | 0.6575 | 0.6319 | 0.6262 |
0.7768 | 0.23 | 3000 | 0.8298 | 0.6383 | 0.6556 | 0.6383 | 0.6365 |
0.753 | 0.27 | 3500 | 0.8225 | 0.6464 | 0.6684 | 0.6464 | 0.6436 |
0.754 | 0.3 | 4000 | 0.7979 | 0.6529 | 0.6750 | 0.6529 | 0.6523 |
0.7466 | 0.34 | 4500 | 0.7644 | 0.6718 | 0.6727 | 0.6718 | 0.6713 |
0.7331 | 0.38 | 5000 | 0.7861 | 0.6591 | 0.6757 | 0.6591 | 0.6581 |
0.72 | 0.42 | 5500 | 0.7972 | 0.6595 | 0.6815 | 0.6595 | 0.6582 |
0.7103 | 0.46 | 6000 | 0.7652 | 0.6702 | 0.6728 | 0.6702 | 0.6688 |
0.7103 | 0.49 | 6500 | 0.7732 | 0.6684 | 0.6796 | 0.6684 | 0.6670 |
0.7023 | 0.53 | 7000 | 0.7921 | 0.6657 | 0.6834 | 0.6657 | 0.6663 |
0.6827 | 0.57 | 7500 | 0.7672 | 0.6733 | 0.6824 | 0.6733 | 0.6726 |
0.6826 | 0.61 | 8000 | 0.7665 | 0.6755 | 0.6789 | 0.6755 | 0.6747 |
0.6705 | 0.65 | 8500 | 0.7659 | 0.6755 | 0.6815 | 0.6755 | 0.6748 |
0.662 | 0.68 | 9000 | 0.7738 | 0.6767 | 0.6833 | 0.6767 | 0.6757 |
0.6556 | 0.72 | 9500 | 0.7623 | 0.6805 | 0.6906 | 0.6805 | 0.6799 |
0.6462 | 0.76 | 10000 | 0.7863 | 0.6719 | 0.6849 | 0.6719 | 0.6701 |
0.6405 | 0.8 | 10500 | 0.7523 | 0.681 | 0.6845 | 0.681 | 0.6805 |
0.6407 | 0.84 | 11000 | 0.7661 | 0.6807 | 0.6856 | 0.6807 | 0.6801 |
0.6341 | 0.87 | 11500 | 0.7672 | 0.6787 | 0.6904 | 0.6787 | 0.6770 |
0.6292 | 0.91 | 12000 | 0.7742 | 0.682 | 0.6922 | 0.682 | 0.6803 |
0.6238 | 0.95 | 12500 | 0.7584 | 0.6855 | 0.6926 | 0.6855 | 0.6850 |
0.6201 | 0.99 | 13000 | 0.7647 | 0.6853 | 0.6932 | 0.6853 | 0.6847 |
Framework versions
- Transformers 4.21.0
- Pytorch 1.12.0+cu102
- Datasets 2.4.0
- Tokenizers 0.12.1
- Downloads last month
- 1