deberta-v3-base-emobank-vad

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

  • Loss: 0.0617
  • Pearson V: 0.7741
  • Mae V: 0.2133
  • Pearson A: 0.5715
  • Mae A: 0.1914
  • Pearson D: 0.3832
  • Mae D: 0.1748
  • Avg Pearson: 0.5763

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: 2e-05
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 64
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 30
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Pearson V Mae V Pearson A Mae A Pearson D Mae D Avg Pearson
9.2581 1.0 126 8.7946 0.0038 2.9832 0.0521 2.8927 0.0138 2.9830 0.0232
4.3424 2.0 252 1.2829 -0.0945 0.7757 0.0719 1.1714 0.0129 1.2900 -0.0032
0.1384 3.0 378 0.1301 0.2267 0.3005 0.0943 0.2892 0.0542 0.2918 0.1251
0.1016 4.0 504 0.1061 0.5828 0.2776 0.2247 0.2613 0.2420 0.2685 0.3498
0.0834 5.0 630 0.0994 0.6477 0.2818 0.3395 0.2594 0.2599 0.2348 0.4157
0.0732 6.0 756 0.0692 0.6854 0.2302 0.4174 0.2051 0.2845 0.1840 0.4624
0.0691 7.0 882 0.0738 0.7077 0.2209 0.4683 0.2276 0.2771 0.1949 0.4844
0.064 8.0 1008 0.0714 0.7234 0.2169 0.5006 0.2198 0.3041 0.1991 0.5094
0.0646 9.0 1134 0.0674 0.7368 0.2406 0.5197 0.1932 0.3150 0.1802 0.5238
0.0596 10.0 1260 0.0785 0.7349 0.2605 0.5127 0.2013 0.3072 0.2039 0.5182
0.0592 11.0 1386 0.0717 0.7487 0.2448 0.5194 0.1985 0.3308 0.1904 0.5330
0.0553 12.0 1512 0.0684 0.7577 0.2141 0.5358 0.2211 0.3446 0.1831 0.5460
0.056 13.0 1638 0.0662 0.7553 0.2229 0.5366 0.2028 0.3499 0.1755 0.5473
0.055 14.0 1764 0.0622 0.7637 0.1998 0.5556 0.2030 0.3468 0.1791 0.5554
0.0521 15.0 1890 0.0687 0.7623 0.2213 0.5557 0.2033 0.3480 0.1951 0.5553
0.0518 16.0 2016 0.0664 0.7652 0.2149 0.5564 0.1995 0.3486 0.1917 0.5567
0.0492 17.0 2142 0.0672 0.7657 0.2124 0.5581 0.2086 0.3619 0.1884 0.5619
0.0522 18.0 2268 0.0585 0.7705 0.2040 0.5659 0.1877 0.3729 0.1706 0.5698
0.0499 19.0 2394 0.0647 0.7715 0.2166 0.5653 0.2029 0.3710 0.1779 0.5693
0.0491 20.0 2520 0.0596 0.7698 0.2025 0.5676 0.1855 0.3673 0.1801 0.5682
0.0489 21.0 2646 0.0624 0.7742 0.2184 0.5745 0.1887 0.3719 0.1788 0.5735
0.0491 22.0 2772 0.0583 0.7758 0.2038 0.5749 0.1844 0.3728 0.1748 0.5745
0.0479 23.0 2898 0.0608 0.7726 0.2113 0.5715 0.1888 0.3727 0.1747 0.5723
0.0484 24.0 3024 0.0632 0.7716 0.2123 0.5700 0.1970 0.3794 0.1791 0.5737
0.047 25.0 3150 0.0638 0.7727 0.2155 0.5707 0.1969 0.3803 0.1792 0.5746
0.0462 26.0 3276 0.0628 0.7744 0.2188 0.5733 0.1901 0.3811 0.1800 0.5762
0.0474 27.0 3402 0.0633 0.7739 0.2179 0.5717 0.1925 0.3814 0.1802 0.5756
0.0466 28.0 3528 0.0623 0.7743 0.2142 0.5702 0.1951 0.3815 0.1730 0.5753
0.0461 29.0 3654 0.0612 0.7741 0.2108 0.5716 0.1910 0.3827 0.1749 0.5761
0.0464 30.0 3780 0.0617 0.7741 0.2133 0.5715 0.1914 0.3832 0.1748 0.5763

Framework versions

  • PEFT 0.16.0
  • Transformers 4.44.2
  • Pytorch 2.6.0+cu124
  • Datasets 4.1.1
  • Tokenizers 0.19.1
Downloads last month
10
Safetensors
Model size
0.2B params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for BenRongey/deberta-v3-base-emobank-vad

Adapter
(15)
this model

Space using BenRongey/deberta-v3-base-emobank-vad 1