anniew666's picture
Update README.md
7fff545
metadata
license: mit
base_model: roberta-large
tags:
  - generated_from_trainer
metrics:
  - accuracy
  - recall
  - f1
model-index:
  - name: lora-roberta-large-no-roller
    results: []
library_name: peft

lora-roberta-large-no-roller

This model is a fine-tuned version of roberta-large on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6492
  • Accuracy: 0.7693
  • Prec: 0.6783
  • Recall: 0.6473
  • F1: 0.6599
  • B Acc: 0.6473
  • Micro F1: 0.7693
  • Prec Joy: 0.7606
  • Recall Joy: 0.7609
  • F1 Joy: 0.7607
  • Prec Anger: 0.6490
  • Recall Anger: 0.6564
  • F1 Anger: 0.6527
  • Prec Disgust: 0.4785
  • Recall Disgust: 0.5179
  • F1 Disgust: 0.4974
  • Prec Fear: 0.7204
  • Recall Fear: 0.6690
  • F1 Fear: 0.6938
  • Prec Neutral: 0.8316
  • Recall Neutral: 0.8715
  • F1 Neutral: 0.8511
  • Prec Sadness: 0.7426
  • Recall Sadness: 0.6653
  • F1 Sadness: 0.7018
  • Prec Surprise: 0.5654
  • Recall Surprise: 0.3903
  • F1 Surprise: 0.4618

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: 0.001
  • train_batch_size: 32
  • eval_batch_size: 16
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 128
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.05
  • num_epochs: 15.0

Training results

Training Loss Epoch Step Validation Loss Accuracy Prec Recall F1 B Acc Micro F1 Prec Joy Recall Joy F1 Joy Prec Anger Recall Anger F1 Anger Prec Disgust Recall Disgust F1 Disgust Prec Fear Recall Fear F1 Fear Prec Neutral Recall Neutral F1 Neutral Prec Sadness Recall Sadness F1 Sadness Prec Surprise Recall Surprise F1 Surprise
0.7936 0.75 1159 0.7811 0.7151 0.6488 0.5609 0.5848 0.5609 0.7151 0.6199 0.8341 0.7112 0.5064 0.6239 0.5590 0.6071 0.2732 0.3768 0.8054 0.4889 0.6084 0.8561 0.7799 0.8162 0.6749 0.5951 0.6325 0.4717 0.3312 0.3891
0.7404 1.5 2318 0.7121 0.7385 0.6664 0.5694 0.6008 0.5694 0.7385 0.7692 0.7050 0.7357 0.5220 0.6354 0.5732 0.6652 0.2732 0.3873 0.7082 0.5789 0.6371 0.7929 0.8857 0.8367 0.7494 0.5369 0.6256 0.4576 0.3710 0.4097
0.713 2.25 3477 0.7058 0.7460 0.6427 0.6191 0.6269 0.6191 0.7460 0.7403 0.7506 0.7454 0.5511 0.6387 0.5917 0.4204 0.5 0.4568 0.7208 0.5766 0.6407 0.8344 0.8447 0.8395 0.7142 0.6476 0.6793 0.5178 0.3753 0.4352
0.7173 3.0 4636 0.6973 0.7482 0.6831 0.5963 0.6211 0.5963 0.7482 0.7181 0.7529 0.7351 0.7473 0.4054 0.5256 0.4491 0.575 0.5043 0.7169 0.6070 0.6574 0.7877 0.8961 0.8384 0.7280 0.6387 0.6804 0.6347 0.2989 0.4064
0.6988 3.75 5795 0.6958 0.7445 0.6635 0.6000 0.6212 0.6000 0.7445 0.6616 0.8326 0.7373 0.6363 0.5851 0.6096 0.6263 0.3321 0.4341 0.6422 0.6047 0.6229 0.8472 0.8195 0.8331 0.6973 0.6644 0.6805 0.5333 0.3613 0.4308
0.661 4.5 6954 0.6763 0.7543 0.6901 0.6081 0.6333 0.6081 0.7543 0.7145 0.7805 0.7461 0.6601 0.5769 0.6157 0.7074 0.2893 0.4106 0.7588 0.5813 0.6583 0.8369 0.8462 0.8415 0.6341 0.7364 0.6815 0.5188 0.4462 0.4798
0.6632 5.25 8113 0.6745 0.7543 0.6671 0.6177 0.6386 0.6177 0.7543 0.7283 0.7653 0.7463 0.6232 0.6109 0.6170 0.4665 0.4732 0.4699 0.7629 0.6058 0.6754 0.8178 0.8689 0.8426 0.7752 0.5902 0.6702 0.4961 0.4097 0.4488
0.6427 6.0 9272 0.6729 0.7514 0.6607 0.6306 0.6403 0.6306 0.7514 0.8023 0.6707 0.7306 0.6623 0.5827 0.6199 0.4785 0.4768 0.4776 0.7312 0.6269 0.6751 0.8013 0.8919 0.8441 0.7613 0.6124 0.6788 0.3879 0.5527 0.4559
0.6363 6.75 10431 0.6584 0.7579 0.6635 0.6367 0.6475 0.6367 0.7579 0.7349 0.7887 0.7608 0.5808 0.6818 0.6273 0.5335 0.4268 0.4742 0.7103 0.6222 0.6633 0.8489 0.8333 0.8410 0.7227 0.6764 0.6988 0.5135 0.4280 0.4669
0.6134 7.5 11590 0.6490 0.7634 0.6768 0.6351 0.6538 0.6351 0.7634 0.7710 0.7383 0.7543 0.6355 0.6258 0.6306 0.5637 0.4268 0.4858 0.7412 0.6433 0.6888 0.8244 0.8720 0.8476 0.7001 0.6889 0.6944 0.5018 0.4505 0.4748
0.6045 8.25 12749 0.6494 0.7612 0.6648 0.6525 0.6555 0.6525 0.7612 0.7566 0.7571 0.7569 0.6401 0.6085 0.6239 0.4371 0.5893 0.5019 0.7557 0.6222 0.6825 0.8329 0.8613 0.8469 0.7556 0.6498 0.6987 0.4760 0.4796 0.4778
0.6139 9.0 13908 0.6585 0.7561 0.6730 0.6388 0.6499 0.6388 0.7561 0.7070 0.8144 0.7570 0.5686 0.7264 0.6379 0.5846 0.4071 0.48 0.7307 0.6187 0.6700 0.8639 0.8125 0.8374 0.7312 0.6698 0.6991 0.5247 0.4226 0.4681
0.5942 9.75 15067 0.6422 0.7661 0.6838 0.6436 0.6605 0.6436 0.7661 0.7413 0.7812 0.7607 0.6120 0.6938 0.6503 0.5619 0.4375 0.4920 0.7531 0.6351 0.6891 0.8407 0.8505 0.8455 0.7505 0.6684 0.7071 0.5271 0.4387 0.4789
0.5798 10.5 16226 0.6553 0.7614 0.6828 0.6358 0.6495 0.6358 0.7614 0.7338 0.7856 0.7588 0.5979 0.6905 0.6409 0.6692 0.3179 0.4310 0.7168 0.6632 0.6889 0.8489 0.8389 0.8439 0.7273 0.672 0.6985 0.4854 0.4828 0.4841
0.5513 11.25 17385 0.6538 0.7612 0.6640 0.6499 0.6550 0.6499 0.7612 0.7121 0.8059 0.7561 0.6487 0.6574 0.6530 0.5348 0.4393 0.4824 0.6504 0.6877 0.6686 0.8594 0.8270 0.8429 0.7018 0.6924 0.6971 0.5410 0.4398 0.4852
0.5544 12.0 18544 0.6492 0.7693 0.6783 0.6473 0.6599 0.6473 0.7693 0.7606 0.7609 0.7607 0.6490 0.6564 0.6527 0.4785 0.5179 0.4974 0.7204 0.6690 0.6938 0.8316 0.8715 0.8511 0.7426 0.6653 0.7018 0.5654 0.3903 0.4618
0.5229 12.75 19703 0.6428 0.7701 0.6822 0.6449 0.6612 0.6449 0.7701 0.7439 0.7820 0.7625 0.6647 0.6584 0.6615 0.5302 0.4393 0.4805 0.7452 0.6327 0.6844 0.8446 0.8563 0.8504 0.6988 0.7147 0.7067 0.5478 0.4312 0.4826
0.5222 13.5 20862 0.6565 0.7656 0.6705 0.6486 0.6586 0.6486 0.7656 0.7366 0.7862 0.7606 0.6581 0.6531 0.6556 0.5108 0.4643 0.4864 0.7322 0.6269 0.6755 0.8479 0.8465 0.8472 0.7176 0.6933 0.7052 0.4899 0.4699 0.4797
0.4972 14.25 22021 0.6574 0.7672 0.6725 0.6502 0.6596 0.6502 0.7672 0.7537 0.7703 0.7619 0.6368 0.6737 0.6547 0.4859 0.4929 0.4894 0.7302 0.6363 0.68 0.8447 0.8525 0.8486 0.7057 0.7044 0.7051 0.5506 0.4215 0.4775

Framework versions

  • Transformers 4.32.0.dev0
  • Pytorch 2.0.1
  • Datasets 2.12.0
  • Tokenizers 0.11.0