metadata
license: mit
base_model: rinna/japanese-roberta-base
tags:
- generated_from_trainer
datasets:
- ja_qu_ad
model-index:
- name: rinna-roberta-qa-ja
results: []
rinna-roberta-qa-ja
This model is a fine-tuned version of rinna/japanese-roberta-base on the ja_qu_ad dataset. It achieves the following results on the evaluation set:
- Loss: 1.6226
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: 7e-05
- train_batch_size: 2
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 16
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 9
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.1602 | 0.47 | 150 | 0.9559 |
| 1.1212 | 0.95 | 300 | 0.7738 |
| 0.6935 | 1.42 | 450 | 0.8850 |
| 0.6329 | 1.9 | 600 | 0.5839 |
| 0.3988 | 2.37 | 750 | 1.0007 |
| 0.3905 | 2.85 | 900 | 0.9119 |
| 0.2245 | 3.32 | 1050 | 1.2299 |
| 0.2331 | 3.79 | 1200 | 1.2494 |
| 0.1556 | 4.27 | 1350 | 1.3691 |
| 0.1537 | 4.74 | 1500 | 1.2580 |
| 0.0811 | 5.22 | 1650 | 1.3939 |
| 0.1139 | 5.69 | 1800 | 1.1321 |
| 0.0623 | 6.17 | 1950 | 1.3723 |
| 0.1096 | 6.64 | 2100 | 1.1039 |
| 0.0622 | 7.11 | 2250 | 1.4657 |
| 0.0731 | 7.59 | 2400 | 1.5731 |
| 0.0482 | 8.06 | 2550 | 1.7190 |
| 0.0518 | 8.54 | 2700 | 1.6226 |
Framework versions
- Transformers 4.33.0
- Pytorch 2.0.1+cu118
- Datasets 2.14.4
- Tokenizers 0.13.3