extended_distilBERT-finetuned-resumes-sections

This model is a fine-tuned version of Geotrend/distilbert-base-en-fr-cased on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0321
  • F1: 0.9735
  • Roc Auc: 0.9850
  • Accuracy: 0.9715

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: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 20

Training results

Training Loss Epoch Step Validation Loss F1 Roc Auc Accuracy
0.0283 1.0 2213 0.0247 0.9610 0.9763 0.9539
0.0153 2.0 4426 0.0223 0.9634 0.9789 0.9593
0.01 3.0 6639 0.0199 0.9702 0.9835 0.9675
0.0073 4.0 8852 0.0218 0.9710 0.9838 0.9690
0.0063 5.0 11065 0.0244 0.9706 0.9835 0.9684
0.0037 6.0 13278 0.0251 0.9700 0.9833 0.9684
0.004 7.0 15491 0.0273 0.9712 0.9837 0.9693
0.003 8.0 17704 0.0266 0.9719 0.9841 0.9695
0.0027 9.0 19917 0.0294 0.9697 0.9831 0.9679
0.0014 10.0 22130 0.0275 0.9714 0.9844 0.9690
0.0016 11.0 24343 0.0299 0.9714 0.9839 0.9697
0.0013 12.0 26556 0.0297 0.9719 0.9852 0.9697
0.0006 13.0 28769 0.0312 0.9711 0.9843 0.9697
0.0004 14.0 30982 0.0305 0.9731 0.9849 0.9720
0.0004 15.0 33195 0.0312 0.9723 0.9845 0.9704
0.0005 16.0 35408 0.0331 0.9716 0.9843 0.9697
0.0006 17.0 37621 0.0321 0.9735 0.9850 0.9715
0.0004 18.0 39834 0.0322 0.9731 0.9850 0.9711
0.0003 19.0 42047 0.0332 0.9722 0.9847 0.9706
0.0004 20.0 44260 0.0334 0.9720 0.9846 0.9704

Framework versions

  • Transformers 4.21.3
  • Pytorch 1.12.1+cu113
  • Datasets 2.4.0
  • Tokenizers 0.12.1
Downloads last month
115
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.