dfurman's picture
Update README.md
1bbae78 verified
metadata
license: mit
base_model: microsoft/deberta-v3-large
datasets:
  - imdb
tags:
  - generated_from_trainer
metrics:
  - accuracy
  - f1
  - precision
  - recall
model-index:
  - name: deberta-v3-large-imdb-v0.2
    results: []

deberta-v3-large-imdb-v0.2

This model is a fine-tuned version of microsoft/deberta-v3-large on the imdb dataset. It achieves the following results on the evaluation set @ epoch 9 of 10, which is loaded as the best model here:

  • Accuracy: 0.9656
  • F1: 0.9657
  • Precision: 0.9640
  • Recall: 0.9673

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: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_ratio: 0.2
  • num_epochs: 10

Training results

Training Loss Epoch Step Validation Loss Accuracy F1 Precision Recall
0.2279 1.0 3125 0.1466 0.9603 0.9599 0.9693 0.9506
0.2689 2.0 6250 0.1929 0.9550 0.9546 0.9626 0.9467
0.1728 3.0 9375 0.1807 0.9584 0.9579 0.9697 0.9463
0.1937 4.0 12500 0.1734 0.9435 0.9457 0.9102 0.9841
0.2044 5.0 15625 0.2102 0.9510 0.9523 0.9272 0.9788
0.0484 6.0 18750 0.2134 0.9593 0.9599 0.9448 0.9756
0.0336 7.0 21875 0.2278 0.9610 0.9614 0.9524 0.9706
0.0704 8.0 25000 0.2039 0.9648 0.9651 0.9581 0.9721
0.0004 9.0 28125 0.2241 0.9656 0.9657 0.9640 0.9673
0.0004 10.0 31250 0.2233 0.9653 0.9654 0.9637 0.9670

Framework versions

  • Transformers 4.39.2
  • Pytorch 2.2.0+cu121
  • Datasets 2.18.0
  • Tokenizers 0.15.2