Luciano's picture
Update README.md
f5bd3ab
|
raw
history blame
2.36 kB
metadata
language:
  - pt
license: mit
tags:
  - generated_from_trainer
metrics:
  - accuracy
model-index:
  - name: bertimbau-base-finetuned-lener-br-finetuned-brazilian_court_decisions
    results: []

bertimbau-base-finetuned-lener-br-finetuned-brazilian_court_decisions

This model is a fine-tuned version of Luciano/bertimbau-base-finetuned-lener-br on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 1.8017
  • Accuracy: 0.7698

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: 15

Training results

Training Loss Epoch Step Validation Loss Accuracy
No log 1.0 405 0.7790 0.6535
0.8276 2.0 810 0.6739 0.7277
0.5818 3.0 1215 0.8767 0.7302
0.4147 4.0 1620 0.8229 0.7896
0.287 5.0 2025 0.9874 0.7921
0.287 6.0 2430 1.2301 0.7772
0.1727 7.0 2835 1.2864 0.7946
0.1179 8.0 3240 1.5097 0.7772
0.0709 9.0 3645 1.4772 0.7921
0.0437 10.0 4050 1.5581 0.7797
0.0437 11.0 4455 1.6317 0.7896
0.0318 12.0 4860 1.7295 0.7822
0.0158 13.0 5265 1.7333 0.7797
0.0108 14.0 5670 1.8008 0.7772
0.0137 15.0 6075 1.8017 0.7698

Framework versions

  • Transformers 4.22.0
  • Pytorch 1.12.1+cu113
  • Datasets 2.4.0
  • Tokenizers 0.12.1