Edit model card

BERT-tiny-sst2

This model is a fine-tuned version of google/bert_uncased_L-2_H-128_A-2 on the glue dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4422
  • Accuracy: 0.8372

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: 16
  • eval_batch_size: 16
  • seed: 33
  • 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
0.3914 1.0 4210 0.4383 0.8211
0.2577 2.0 8420 0.4422 0.8372
0.212 3.0 12630 0.5460 0.8085
0.1862 4.0 16840 0.5885 0.8245
0.1671 5.0 21050 0.7159 0.8096

Framework versions

  • Transformers 4.22.1
  • Pytorch 1.12.1+cu113
  • Datasets 2.5.1
  • Tokenizers 0.12.1
Downloads last month
3

Dataset used to train gokuls/BERT-tiny-sst2

Evaluation results