Edit model card

deberta-v3-large-survey-related_passage_consistency-rater-all

This model is a fine-tuned version of microsoft/deberta-v3-large on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.7600
  • Krippendorff: 0.7377
  • Spearman: 0.6526
  • Absolute Agreement: 0.7546
  • Agreement Within One: 0.9470

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: 6e-06
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 20
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Krippendorff Spearman Absolute Agreement Agreement Within One
No log 1.0 55 1.9111 -0.2279 nan 0.4167 0.8333
No log 2.0 110 1.8873 -0.2279 nan 0.4167 0.8333
No log 3.0 165 1.8559 -0.2279 nan 0.4167 0.8333
No log 4.0 220 1.7568 -0.2279 nan 0.4167 0.8333
No log 5.0 275 1.6846 -0.2279 nan 0.4167 0.8333
No log 6.0 330 1.5749 -0.2279 nan 0.4167 0.8333
No log 7.0 385 1.5052 -0.2279 nan 0.4167 0.8333
No log 8.0 440 1.4261 -0.2279 nan 0.4167 0.8333
No log 9.0 495 1.4044 0.3239 0.3369 0.4167 0.9028
1.2329 10.0 550 1.3616 0.2662 0.2785 0.4167 0.8889
1.2329 11.0 605 1.2619 0.2754 0.2653 0.4722 0.8889
1.2329 12.0 660 1.2314 0.2592 0.3128 0.4306 0.8889
1.2329 13.0 715 1.1903 0.3049 0.3317 0.5417 0.8889
1.2329 14.0 770 1.3300 0.2890 0.3347 0.5278 0.8889
1.2329 15.0 825 1.2220 0.2192 0.3622 0.4722 0.875
1.2329 16.0 880 1.1202 0.3028 0.3719 0.5139 0.8889
1.2329 17.0 935 1.1119 0.1911 0.1013 0.5556 0.875
1.2329 18.0 990 1.0345 0.2221 0.1646 0.5556 0.875
0.6021 19.0 1045 1.1044 0.2988 0.3099 0.5417 0.8889
0.6021 20.0 1100 1.0117 0.2274 0.1782 0.5556 0.875

Framework versions

  • Transformers 4.26.0
  • Pytorch 1.13.1
  • Datasets 2.10.1
  • Tokenizers 0.12.1
Downloads last month
14