Hugging Face's logo

hcene
/
results

PEFT
TensorBoard
Safetensors
Generated from Trainer
Model card Files Metrics Community
results / README.md
hcene's picture
hcene
hcene/finetuned-mDEBERTa-v3-mnli-xnli
f545141 verified
preview code
|
raw
history blame contribute delete
6.99 kB
metadata
license: mit
library_name: peft
tags:
  - generated_from_trainer
metrics:
  - accuracy
  - precision
  - recall
  - f1
base_model: MoritzLaurer/mDeBERTa-v3-base-mnli-xnli
model-index:
  - name: results
    results: []

results

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

  • Loss: 0.6765
  • Accuracy: 0.7634
  • Precision: 0.7675
  • Recall: 0.7644
  • F1: 0.7627
  • Ratio: 0.3297

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: 0.0005
  • train_batch_size: 20
  • eval_batch_size: 20
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • lr_scheduler_warmup_steps: 4
  • num_epochs: 20
  • label_smoothing_factor: 0.1

Training results

Training Loss Epoch Step Validation Loss Accuracy Precision Recall F1 Ratio
1.7741 0.17 10 1.0961 0.7061 0.7103 0.7073 0.7078 0.3262
1.2149 0.34 20 0.8783 0.7025 0.7066 0.7038 0.7044 0.3262
0.959 0.52 30 0.8413 0.6774 0.6981 0.6784 0.6854 0.2939
0.9582 0.69 40 0.7705 0.7312 0.7417 0.7321 0.7314 0.3190
0.8706 0.86 50 0.6728 0.7419 0.7545 0.7437 0.7385 0.3190
0.8804 1.03 60 0.6933 0.7133 0.7402 0.7157 0.6919 0.3190
0.8999 1.21 70 0.7167 0.7133 0.7208 0.7144 0.7158 0.3190
0.8914 1.38 80 0.6910 0.7384 0.7549 0.7390 0.7325 0.3226
0.8578 1.55 90 0.6862 0.7348 0.7533 0.7369 0.7243 0.3262
0.8755 1.72 100 0.6889 0.7240 0.7449 0.7263 0.7095 0.3262
0.8551 1.9 110 0.7090 0.7133 0.7408 0.7158 0.6899 0.3262
0.8736 2.07 120 0.7019 0.7097 0.7236 0.7120 0.6975 0.3262
0.8647 2.24 130 0.7078 0.7240 0.7354 0.7261 0.7174 0.3262
0.8755 2.41 140 0.7023 0.7527 0.7716 0.7533 0.7448 0.3262
0.858 2.59 150 0.6745 0.7384 0.7450 0.7393 0.7372 0.3262
0.8912 2.76 160 0.6842 0.7491 0.7635 0.7511 0.7424 0.3297
0.8294 2.93 170 0.6623 0.7599 0.7624 0.7609 0.7602 0.3297
0.8481 3.1 180 0.6652 0.7599 0.7715 0.7617 0.7551 0.3333
0.8488 3.28 190 0.6782 0.7312 0.7609 0.7335 0.7131 0.3297
0.8418 3.45 200 0.6884 0.7706 0.7738 0.7719 0.7720 0.3262
0.8774 3.62 210 0.7066 0.7419 0.7523 0.7438 0.7381 0.3262
0.8496 3.79 220 0.6687 0.7133 0.7214 0.7154 0.7060 0.3333
0.825 3.97 230 0.6618 0.7634 0.7833 0.7639 0.7547 0.3297
0.8933 4.14 240 0.6946 0.7419 0.7692 0.7424 0.7278 0.3262
0.8579 4.31 250 0.6795 0.7491 0.7786 0.7495 0.7353 0.3262
0.8023 4.48 260 0.6595 0.7563 0.7727 0.7569 0.7501 0.3262
0.8736 4.66 270 0.6703 0.7491 0.7558 0.7508 0.7482 0.3262
0.8291 4.83 280 0.7102 0.6989 0.7630 0.7019 0.6499 0.3262
0.8923 5.0 290 0.7004 0.7097 0.7571 0.7124 0.6756 0.3262
0.8571 5.17 300 0.6739 0.7634 0.7717 0.7642 0.7621 0.3262
0.8521 5.34 310 0.6666 0.7563 0.7710 0.7569 0.7511 0.3262
0.8369 5.52 320 0.6815 0.7455 0.7487 0.7467 0.7472 0.3262
0.7897 5.69 330 0.6731 0.7097 0.7343 0.7122 0.6871 0.3262
0.8801 5.86 340 0.6773 0.7419 0.7631 0.7441 0.7304 0.3297
0.891 6.03 350 0.7107 0.7491 0.7556 0.7509 0.7473 0.3297
0.8444 6.21 360 0.6805 0.7634 0.7879 0.7639 0.7543 0.3262
0.8375 6.38 370 0.6562 0.7599 0.7725 0.7605 0.7560 0.3262
0.8141 6.55 380 0.6578 0.7276 0.7409 0.7296 0.7217 0.3262
0.8792 6.72 390 0.6790 0.7204 0.7355 0.7226 0.7121 0.3262
0.8868 6.9 400 0.7063 0.7384 0.7411 0.7397 0.7404 0.3262
0.8767 7.07 410 0.7074 0.7240 0.7440 0.7262 0.7126 0.3262
0.8545 7.24 420 0.6725 0.7276 0.7520 0.7300 0.7108 0.3297
0.8589 7.41 430 0.6712 0.7276 0.7473 0.7299 0.7139 0.3297
0.8522 7.59 440 0.6853 0.7634 0.7655 0.7649 0.7644 0.3297
0.777 7.76 450 0.6623 0.7634 0.7714 0.7642 0.7604 0.3297
0.8903 7.93 460 0.6629 0.7599 0.7629 0.7609 0.7598 0.3297
0.8168 8.1 470 0.6714 0.7599 0.7650 0.7608 0.7584 0.3297
0.7979 8.28 480 0.6469 0.7491 0.7505 0.7505 0.7504 0.3297
0.8674 8.45 490 0.6553 0.7455 0.7603 0.7475 0.7382 0.3297
0.8475 8.62 500 0.6788 0.7563 0.7576 0.7576 0.7576 0.3297
0.8723 8.79 510 0.6862 0.7599 0.7613 0.7612 0.7611 0.3297
0.8684 8.97 520 0.6938 0.7563 0.7604 0.7579 0.7560 0.3297
0.8278 9.14 530 0.6765 0.7634 0.7675 0.7644 0.7627 0.3297

Framework versions

  • PEFT 0.9.0
  • Transformers 4.39.0.dev0
  • Pytorch 2.1.0+cu121
  • Datasets 2.17.1
  • Tokenizers 0.15.2