bert-large-uncased-financial-phrasebank-allagree2

This model is a fine-tuned version of bert-large-uncased on the financial_phrasebank dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0734
  • Accuracy: 0.9912
  • F1: 0.9911

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: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 5

Training results

Training Loss Epoch Step Validation Loss Accuracy F1
0.3209 1.0 227 0.1929 0.9558 0.9551
0.0821 2.0 454 0.0994 0.9867 0.9867
0.04 3.0 681 0.0685 0.9867 0.9866
0.0098 4.0 908 0.0980 0.9867 0.9867
0.0003 5.0 1135 0.0734 0.9912 0.9911

Framework versions

  • Transformers 4.21.1
  • Pytorch 1.12.0+cu113
  • Datasets 2.4.0
  • Tokenizers 0.12.1
Downloads last month
13
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Dataset used to train Farshid/bert-large-uncased-financial-phrasebank-allagree2

Evaluation results