Tomor0720's picture
update model card README.md
507f2f8
metadata
license: mit
tags:
  - generated_from_trainer
datasets:
  - glue
metrics:
  - accuracy
  - f1
model-index:
  - name: deberta-large-finetuned-qqp
    results:
      - task:
          name: Text Classification
          type: text-classification
        dataset:
          name: glue
          type: glue
          config: qqp
          split: train
          args: qqp
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.8985901558248826
          - name: F1
            type: f1
            value: 0.8648292232625608

deberta-large-finetuned-qqp

This model is a fine-tuned version of microsoft/deberta-large on the glue dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2635
  • Accuracy: 0.8986
  • F1: 0.8648

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: linear
  • num_epochs: 2

Training results

Training Loss Epoch Step Validation Loss Accuracy F1
0.4058 1.0 22741 0.3923 0.8496 0.8108
0.2347 2.0 45482 0.2635 0.8986 0.8648

Framework versions

  • Transformers 4.25.1
  • Pytorch 1.13.1+cu116
  • Datasets 2.8.0
  • Tokenizers 0.13.2