Edit model card

VN_ja-en_byt5_small

This model is a fine-tuned version of google/byt5-small on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 1.0552

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: 0.0003
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 3

Training results

Training Loss Epoch Step Validation Loss
1.1687 0.1 2000 1.1805
0.9685 0.19 4000 1.1384
0.8989 0.29 6000 1.1207
0.8583 0.39 8000 1.1046
0.833 0.49 10000 1.1290
0.8102 0.58 12000 1.1225
0.7932 0.68 14000 1.0956
0.7776 0.78 16000 1.0970
0.762 0.88 18000 1.0992
0.7522 0.97 20000 1.0760
0.7318 1.07 22000 1.0579
0.7197 1.17 24000 1.0780
0.7142 1.27 26000 1.0748
0.7093 1.36 28000 1.0781
0.7005 1.46 30000 1.0756
0.6938 1.56 32000 1.0702
0.6896 1.65 34000 1.0563
0.6846 1.75 36000 1.0603
0.6807 1.85 38000 1.0626
0.6766 1.95 40000 1.0666
0.6649 2.04 42000 1.0694
0.6532 2.14 44000 1.0564
0.6501 2.24 46000 1.0715
0.6476 2.34 48000 1.0551
0.646 2.43 50000 1.0601
0.6445 2.53 52000 1.0595
0.6404 2.63 54000 1.0494
0.6378 2.72 56000 1.0584
0.636 2.82 58000 1.0531
0.6345 2.92 60000 1.0552

Framework versions

  • Transformers 4.19.2
  • Pytorch 1.11.0+cu113
  • Datasets 2.2.2
  • Tokenizers 0.12.1
Downloads last month
2