chunwoolee0's picture
Update README.md
fb5847e
---
tags:
- generated_from_trainer
datasets:
- klue
metrics:
- f1
model-index:
- name: klue_ynat_roberta_base_model
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: klue
type: klue
config: ynat
split: validation
args: ynat
metrics:
- name: F1
type: f1
value: 0.872014500465787
---
<!-- 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. -->
# klue_ynat_roberta_base_model
This model is a fine-tuned version of [klue/roberta-base](https://huggingface.co/klue/roberta-base) on the klue dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3747
- F1: 0.8720
## Model description
Pretrained RoBERTa Model on Korean Language. See [Github](https://github.com/KLUE-benchmark/KLUE) and [Paper](https://arxiv.org/abs/2105.09680) for more details.
## Intended uses & limitations
Pretrained RoBERTa Model on Korean Language. See Github and Paper for more details.
## Training and evaluation data
## How to use
_NOTE:_ Use `BertTokenizer` instead of RobertaTokenizer. (`AutoTokenizer` will load `BertTokenizer`)
```python
from transformers import AutoModel, AutoTokenizer
model = AutoModel.from_pretrained("klue/roberta-base")
tokenizer = AutoTokenizer.from_pretrained("klue/roberta-base")
```
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 256
- eval_batch_size: 256
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
### Training results
| Training Loss | Epoch | Step | Validation Loss | F1 |
|:-------------:|:-----:|:----:|:---------------:|:------:|
| No log | 1.0 | 179 | 0.4838 | 0.8444 |
| No log | 2.0 | 358 | 0.3848 | 0.8659 |
| 0.4203 | 3.0 | 537 | 0.3778 | 0.8690 |
| 0.4203 | 4.0 | 716 | 0.3762 | 0.8702 |
| 0.4203 | 5.0 | 895 | 0.3747 | 0.8720 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3