deberta-v3-large-survey-related_passage_old_facts-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: 0.4433
- Krippendorff: 0.9228
- Spearman: 0.9300
- Absolute Agreement: 0.8687
- Agreement Within One: 0.9401
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 | 1.8838 | -0.0125 | 0.1227 | 0.25 | 0.7361 |
No log | 2.0 | 110 | 1.8809 | -0.0125 | 0.1227 | 0.25 | 0.7361 |
No log | 3.0 | 165 | 1.8774 | -0.1004 | nan | 0.2778 | 0.7083 |
No log | 4.0 | 220 | 1.8714 | -0.0611 | 0.0419 | 0.25 | 0.7222 |
No log | 5.0 | 275 | 1.8795 | 0.2639 | 0.3723 | 0.3056 | 0.875 |
No log | 6.0 | 330 | 1.8539 | -0.2880 | 0.1227 | 0.3056 | 0.5833 |
No log | 7.0 | 385 | 1.8261 | 0.4776 | 0.4872 | 0.3472 | 0.7917 |
No log | 8.0 | 440 | 1.7614 | 0.4363 | 0.4580 | 0.3611 | 0.75 |
No log | 9.0 | 495 | 1.7462 | 0.4363 | 0.4977 | 0.3611 | 0.8056 |
1.7495 | 10.0 | 550 | 1.7122 | 0.4028 | 0.4614 | 0.375 | 0.7639 |
1.7495 | 11.0 | 605 | 1.7387 | 0.3566 | 0.4272 | 0.3611 | 0.8056 |
1.7495 | 12.0 | 660 | 1.7000 | 0.4377 | 0.4815 | 0.3611 | 0.8056 |
1.7495 | 13.0 | 715 | 1.5995 | 0.5463 | 0.5639 | 0.4028 | 0.8056 |
1.7495 | 14.0 | 770 | 1.5903 | 0.5178 | 0.5343 | 0.3889 | 0.7639 |
1.7495 | 15.0 | 825 | 1.6389 | 0.4298 | 0.4835 | 0.4028 | 0.7222 |
1.7495 | 16.0 | 880 | 1.5450 | 0.5642 | 0.5777 | 0.4583 | 0.7917 |
1.7495 | 17.0 | 935 | 1.5418 | 0.5031 | 0.5172 | 0.4444 | 0.7361 |
1.7495 | 18.0 | 990 | 1.5372 | 0.4530 | 0.4907 | 0.4583 | 0.75 |
1.1446 | 19.0 | 1045 | 1.4613 | 0.5787 | 0.6183 | 0.4861 | 0.8194 |
1.1446 | 20.0 | 1100 | 1.4501 | 0.5984 | 0.6308 | 0.4861 | 0.8056 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.13.1
- Datasets 2.10.1
- Tokenizers 0.12.1
- Downloads last month
- 13