Edit model card

lilt-xlm-roberta-base-finetuned-funsd-iob-original

This model is a fine-tuned version of nielsr/lilt-xlm-roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 2.1573
  • Precision: 0.7252
  • Recall: 0.7718
  • F1: 0.7478
  • Accuracy: 0.7676

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: 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: 30

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
No log 1.33 100 0.8309 0.5157 0.6673 0.5818 0.6594
No log 2.67 200 1.0045 0.6080 0.6699 0.6374 0.7387
No log 4.0 300 0.9127 0.6177 0.7310 0.6696 0.7427
No log 5.33 400 0.9808 0.6478 0.7300 0.6865 0.7521
0.6318 6.67 500 1.2169 0.6863 0.7376 0.7110 0.7547
0.6318 8.0 600 1.1830 0.6918 0.7580 0.7234 0.7326
0.6318 9.33 700 1.3537 0.6955 0.7504 0.7219 0.7426
0.6318 10.67 800 1.3888 0.6994 0.7611 0.7290 0.7507
0.6318 12.0 900 1.5929 0.7204 0.7560 0.7378 0.7553
0.1082 13.33 1000 1.7679 0.6891 0.7397 0.7135 0.7452
0.1082 14.67 1100 1.7197 0.7003 0.7570 0.7275 0.7530
0.1082 16.0 1200 1.8053 0.7188 0.7448 0.7315 0.7616
0.1082 17.33 1300 1.9315 0.7109 0.7728 0.7405 0.7643
0.1082 18.67 1400 2.0142 0.7240 0.7789 0.7504 0.7676
0.0312 20.0 1500 2.0475 0.7264 0.7478 0.7369 0.7654
0.0312 21.33 1600 2.0463 0.7251 0.7539 0.7393 0.7599
0.0312 22.67 1700 2.0648 0.7289 0.7753 0.7514 0.7623
0.0312 24.0 1800 2.1301 0.7272 0.7606 0.7435 0.7667
0.0312 25.33 1900 2.1319 0.7274 0.7585 0.7426 0.7694
0.0064 26.67 2000 2.1499 0.7247 0.7723 0.7477 0.7673
0.0064 28.0 2100 2.1627 0.7235 0.7733 0.7476 0.7670
0.0064 29.33 2200 2.1573 0.7252 0.7718 0.7478 0.7676

Framework versions

  • Transformers 4.25.1
  • Pytorch 1.13.0+cu116
  • Datasets 2.8.0
  • Tokenizers 0.13.2
Downloads last month
3
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.