metadata
tags:
- generated_from_trainer
datasets:
- null
model-index:
- name: bert-base-multilingual-cased-urgency
results:
- task:
name: Masked Language Modeling
type: fill-mask
bert-base-multilingual-cased-urgency
This model is a fine-tuned version of bert-base-multilingual-cased on the mWACH NEO dataset. It achieves the following results on the evaluation set:
- Loss: 2.2797
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: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
4.1408 | 1.0 | 5659 | 3.6705 |
2.8777 | 2.0 | 11318 | 2.5536 |
2.561 | 3.0 | 16977 | 2.2740 |
Framework versions
- Transformers 4.10.0
- Pytorch 1.9.0+cu102
- Datasets 1.11.0
- Tokenizers 0.10.3