hBERTv1_wnli / README.md
gokuls's picture
End of training
d9b3c14
metadata
language:
  - en
tags:
  - generated_from_trainer
datasets:
  - glue
metrics:
  - accuracy
model-index:
  - name: hBERTv1_wnli
    results:
      - task:
          name: Text Classification
          type: text-classification
        dataset:
          name: GLUE WNLI
          type: glue
          config: wnli
          split: validation
          args: wnli
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.5633802816901409

hBERTv1_wnli

This model is a fine-tuned version of gokuls/bert_12_layer_model_v1 on the GLUE WNLI dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6877
  • Accuracy: 0.5634

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: 5e-05
  • train_batch_size: 256
  • eval_batch_size: 256
  • seed: 10
  • distributed_type: multi-GPU
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 50
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Accuracy
0.7359 1.0 3 0.7194 0.4366
0.6989 2.0 6 0.6899 0.5634
0.7031 3.0 9 0.7028 0.4366
0.7012 4.0 12 0.6889 0.5634
0.697 5.0 15 0.6894 0.5634
0.6971 6.0 18 0.7015 0.4366
0.7 7.0 21 0.6882 0.5634
0.6928 8.0 24 0.6890 0.5634
0.6932 9.0 27 0.6897 0.5634
0.6954 10.0 30 0.6956 0.4366
0.6962 11.0 33 0.6913 0.5634
0.6956 12.0 36 0.6877 0.5634
0.6973 13.0 39 0.6926 0.5070
0.6978 14.0 42 0.6933 0.4930
0.6945 15.0 45 0.6883 0.5634
0.6974 16.0 48 0.6881 0.5634
0.6936 17.0 51 0.6925 0.5211

Framework versions

  • Transformers 4.26.1
  • Pytorch 1.14.0a0+410ce96
  • Datasets 2.10.1
  • Tokenizers 0.13.2