hBERTv1_qnli / README.md
gokuls's picture
End of training
b1261c8
metadata
language:
  - en
tags:
  - generated_from_trainer
datasets:
  - glue
metrics:
  - accuracy
model-index:
  - name: hBERTv1_qnli
    results:
      - task:
          name: Text Classification
          type: text-classification
        dataset:
          name: GLUE QNLI
          type: glue
          config: qnli
          split: validation
          args: qnli
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.8112758557569101

hBERTv1_qnli

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

  • Loss: 0.4198
  • Accuracy: 0.8113

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
0.6667 1.0 410 0.5955 0.6874
0.4998 2.0 820 0.4486 0.7948
0.3985 3.0 1230 0.4198 0.8113
0.3106 4.0 1640 0.4841 0.7866
0.2286 5.0 2050 0.5340 0.7906
0.1662 6.0 2460 0.6282 0.7728
0.1237 7.0 2870 0.6678 0.7752
0.0945 8.0 3280 0.7668 0.7752

Framework versions

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