Edit model card

DNADebertaSentencepiece30k

This model is a fine-tuned version of on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 6.3257

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: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 64
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 15
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
7.9373 0.41 5000 7.8263
7.8005 0.81 10000 7.7871
7.7704 1.22 15000 7.7630
7.7477 1.62 20000 7.6857
7.6058 2.03 25000 7.5543
7.5281 2.44 30000 7.4839
7.4487 2.84 35000 7.3801
7.3368 3.25 40000 7.2603
7.1923 3.66 45000 7.0365
6.9858 4.06 50000 6.8793
6.8639 4.47 55000 6.7839
6.7877 4.87 60000 6.7176
6.728 5.28 65000 6.6680
6.6826 5.69 70000 6.6258
6.6414 6.09 75000 6.5847
6.6057 6.5 80000 6.5571
6.5794 6.91 85000 6.5279
6.5525 7.31 90000 6.5059
6.5354 7.72 95000 6.4816
6.5125 8.12 100000 6.4674
6.4958 8.53 105000 6.4486
6.4817 8.94 110000 6.4317
6.4674 9.34 115000 6.4195
6.4549 9.75 120000 6.4072
6.4409 10.16 125000 6.3945
6.4302 10.56 130000 6.3861
6.4214 10.97 135000 6.3755
6.4118 11.37 140000 6.3659
6.4058 11.78 145000 6.3604
6.3985 12.19 150000 6.3560
6.3899 12.59 155000 6.3473
6.3837 13.0 160000 6.3417
6.3782 13.41 165000 6.3361
6.3753 13.81 170000 6.3309
6.3733 14.22 175000 6.3285
6.3706 14.62 180000 6.3277

Framework versions

  • Transformers 4.19.2
  • Pytorch 1.11.0
  • Datasets 2.2.2
  • Tokenizers 0.12.1
Downloads last month
1