Edit model card

predict-perception-bert-cause-human

This model is a fine-tuned version of dbmdz/bert-base-italian-xxl-cased on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.7139
  • Rmse: 1.2259
  • Rmse Cause::a Causata da un essere umano: 1.2259
  • Mae: 1.0480
  • Mae Cause::a Causata da un essere umano: 1.0480
  • R2: 0.4563
  • R2 Cause::a Causata da un essere umano: 0.4563
  • Cos: 0.4783
  • Pair: 0.0
  • Rank: 0.5
  • Neighbors: 0.3953
  • Rsa: nan

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: 1e-05
  • train_batch_size: 20
  • eval_batch_size: 8
  • seed: 1996
  • 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 Rmse Rmse Cause::a Causata da un essere umano Mae Mae Cause::a Causata da un essere umano R2 R2 Cause::a Causata da un essere umano Cos Pair Rank Neighbors Rsa
1.0874 1.0 15 1.2615 1.6296 1.6296 1.3836 1.3836 0.0393 0.0393 0.0435 0.0 0.5 0.2935 nan
0.9577 2.0 30 1.1988 1.5886 1.5886 1.3017 1.3017 0.0870 0.0870 0.4783 0.0 0.5 0.3944 nan
0.8414 3.0 45 0.9870 1.4414 1.4414 1.1963 1.1963 0.2483 0.2483 0.3913 0.0 0.5 0.3048 nan
0.7291 4.0 60 0.9098 1.3839 1.3839 1.1297 1.1297 0.3071 0.3071 0.4783 0.0 0.5 0.3084 nan
0.5949 5.0 75 0.9207 1.3921 1.3921 1.2079 1.2079 0.2988 0.2988 0.4783 0.0 0.5 0.3084 nan
0.4938 6.0 90 0.8591 1.3448 1.3448 1.1842 1.1842 0.3458 0.3458 0.4783 0.0 0.5 0.3084 nan
0.3611 7.0 105 0.8176 1.3119 1.3119 1.1454 1.1454 0.3774 0.3774 0.5652 0.0 0.5 0.4091 nan
0.2663 8.0 120 0.6879 1.2034 1.2034 1.0300 1.0300 0.4761 0.4761 0.5652 0.0 0.5 0.4091 nan
0.1833 9.0 135 0.7704 1.2735 1.2735 1.1031 1.1031 0.4133 0.4133 0.5652 0.0 0.5 0.3152 nan
0.1704 10.0 150 0.7097 1.2222 1.2222 1.0382 1.0382 0.4596 0.4596 0.4783 0.0 0.5 0.3084 nan
0.1219 11.0 165 0.6872 1.2027 1.2027 1.0198 1.0198 0.4767 0.4767 0.4783 0.0 0.5 0.3084 nan
0.1011 12.0 180 0.7201 1.2312 1.2312 1.0466 1.0466 0.4516 0.4516 0.5652 0.0 0.5 0.3152 nan
0.0849 13.0 195 0.7267 1.2368 1.2368 1.0454 1.0454 0.4466 0.4466 0.4783 0.0 0.5 0.3953 nan
0.0818 14.0 210 0.7361 1.2448 1.2448 1.0565 1.0565 0.4394 0.4394 0.4783 0.0 0.5 0.3953 nan
0.0634 15.0 225 0.7158 1.2275 1.2275 1.0384 1.0384 0.4549 0.4549 0.3913 0.0 0.5 0.3306 nan
0.065 16.0 240 0.7394 1.2475 1.2475 1.0659 1.0659 0.4369 0.4369 0.3913 0.0 0.5 0.3306 nan
0.0541 17.0 255 0.7642 1.2683 1.2683 1.0496 1.0496 0.4181 0.4181 0.4783 0.0 0.5 0.3953 nan
0.0577 18.0 270 0.7137 1.2257 1.2257 1.0303 1.0303 0.4565 0.4565 0.4783 0.0 0.5 0.3953 nan
0.0474 19.0 285 0.7393 1.2475 1.2475 1.0447 1.0447 0.4370 0.4370 0.4783 0.0 0.5 0.3084 nan
0.0494 20.0 300 0.7157 1.2274 1.2274 1.0453 1.0453 0.4550 0.4550 0.4783 0.0 0.5 0.3084 nan
0.0434 21.0 315 0.7248 1.2352 1.2352 1.0462 1.0462 0.4480 0.4480 0.4783 0.0 0.5 0.3953 nan
0.049 22.0 330 0.7384 1.2467 1.2467 1.0613 1.0613 0.4377 0.4377 0.4783 0.0 0.5 0.3953 nan
0.0405 23.0 345 0.7420 1.2498 1.2498 1.0653 1.0653 0.4349 0.4349 0.3913 0.0 0.5 0.3306 nan
0.0398 24.0 360 0.7355 1.2442 1.2442 1.0620 1.0620 0.4399 0.4399 0.4783 0.0 0.5 0.3953 nan
0.0398 25.0 375 0.7570 1.2623 1.2623 1.0698 1.0698 0.4235 0.4235 0.3913 0.0 0.5 0.3306 nan
0.0345 26.0 390 0.7359 1.2446 1.2446 1.0610 1.0610 0.4396 0.4396 0.5652 0.0 0.5 0.3152 nan
0.0345 27.0 405 0.7417 1.2495 1.2495 1.0660 1.0660 0.4352 0.4352 0.4783 0.0 0.5 0.3953 nan
0.0386 28.0 420 0.7215 1.2323 1.2323 1.0514 1.0514 0.4506 0.4506 0.4783 0.0 0.5 0.3084 nan
0.0372 29.0 435 0.7140 1.2260 1.2260 1.0477 1.0477 0.4562 0.4562 0.5652 0.0 0.5 0.4091 nan
0.0407 30.0 450 0.7139 1.2259 1.2259 1.0480 1.0480 0.4563 0.4563 0.4783 0.0 0.5 0.3953 nan

Framework versions

  • Transformers 4.16.2
  • Pytorch 1.10.2+cu113
  • Datasets 1.18.3
  • Tokenizers 0.11.0
Downloads last month
317
Hosted inference API
Text Classification
Examples
Examples
This model can be loaded on the Inference API on-demand.