DNADebertaSentencepiece30k_continuation_continuation_continuation

This model is a fine-tuned version of Vlasta/DNADebertaSentencepiece30k_continuation_continuation on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 5.9319

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
6.0844 0.41 5000 6.0623
6.0962 0.81 10000 6.0659
6.0903 1.22 15000 6.0566
6.0874 1.62 20000 6.0550
6.082 2.03 25000 6.0485
6.0756 2.44 30000 6.0446
6.0722 2.84 35000 6.0429
6.0698 3.25 40000 6.0317
6.0627 3.66 45000 6.0297
6.0606 4.06 50000 6.0301
6.0521 4.47 55000 6.0224
6.0526 4.87 60000 6.0159
6.0473 5.28 65000 6.0140
6.0435 5.69 70000 6.0076
6.039 6.09 75000 6.0022
6.032 6.5 80000 6.0037
6.0319 6.91 85000 5.9979
6.0232 7.31 90000 5.9937
6.0279 7.72 95000 5.9844
6.0198 8.12 100000 5.9854
6.0165 8.53 105000 5.9796
6.0153 8.94 110000 5.9741
6.0111 9.34 115000 5.9722
6.0082 9.75 120000 5.9679
6.0035 10.16 125000 5.9654
5.999 10.56 130000 5.9624
5.998 10.97 135000 5.9572
5.9926 11.37 140000 5.9535
5.9927 11.78 145000 5.9533
5.9903 12.19 150000 5.9517
5.986 12.59 155000 5.9459
5.9816 13.0 160000 5.9439
5.9786 13.41 165000 5.9390
5.9781 13.81 170000 5.9357
5.9779 14.22 175000 5.9346
5.9756 14.62 180000 5.9339

Framework versions

  • Transformers 4.19.2
  • Pytorch 1.11.0
  • Datasets 2.2.2
  • Tokenizers 0.12.1
Downloads last month
6
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.