koelectra-base-v3-discriminator-finetuned-ner
This model is a fine-tuned version of monologg/koelectra-base-v3-discriminator on the klue dataset. It achieves the following results on the evaluation set:
- Loss: 0.1957
- Precision: 0.6665
- Recall: 0.7350
- F1: 0.6991
- Accuracy: 0.9396
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: 48
- eval_batch_size: 48
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
Training results
Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
---|---|---|---|---|---|---|---|
No log | 1.0 | 438 | 0.2588 | 0.5701 | 0.6655 | 0.6141 | 0.9212 |
0.4333 | 2.0 | 876 | 0.2060 | 0.6671 | 0.7134 | 0.6895 | 0.9373 |
0.1944 | 3.0 | 1314 | 0.1957 | 0.6665 | 0.7350 | 0.6991 | 0.9396 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.12.0+cu102
- Datasets 1.14.0
- Tokenizers 0.10.3
- Downloads last month
- 21
Inference Providers
NEW
This model is not currently available via any of the supported Inference Providers.
Dataset used to train JoonJoon/koelectra-base-v3-discriminator-finetuned-ner
Evaluation results
- Precision on klueself-reported0.667
- Recall on klueself-reported0.735
- F1 on klueself-reported0.699
- Accuracy on klueself-reported0.940