CoolFace
Modelpublic

whataboutyou-ai/financial_bert

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face
1likes235downloads
Model Card

financial_bert

This model is a fine-tuned version of bert-base-uncased on the BUSTER dataset. This model is ready to use for Named Entity Recognition (NER).

It achieves the following results on the evaluation set:

  • —Loss: 0.0201
  • —Precision: 0.7977
  • —Recall: 0.8532
  • —F1: 0.8245
  • —Accuracy: 0.9937

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

This model was fine-tuned on the BUSTER dataset.

The training dataset distinguishes between the beginning and continuation of an entity so that if there are back-to-back entities of the same type, the model can output where the second entity begins. As in the dataset, each token will be classified as one of the following classes:

EntityDescription
OOutside of a named entity
B-GenericInfo.ANNUALREVENUESBeginning of annual revenues entity
I-GenericInfo.ANNUALREVENUESContinuation of annual revenues entity
B-Parties.ACQUIRED_COMPANYBeginning of acquired company entity
I-Parties.ACQUIRED_COMPANYContinuation of acquired company entity
B-Parties.BUYING_COMPANYBeginning of buying company entity
I-Parties.BUYING_COMPANYContinuation of buying company entity
B-Parties.SELLING_COMPANYBeginning of selling company entity
I-Parties.SELLING_COMPANYContinuation of selling company entity
B-Advisors.GENERICCONSULTINGCOMPANYBeginning of generic consulting company entity
I-Advisors.GENERICCONSULTINGCOMPANYContinuation of generic consulting company entity
B-Advisors.LEGALCONSULTINGCOMPANYBeginning of legal consulting company entity
I-Advisors.LEGALCONSULTINGCOMPANYContinuation of legal consulting company entity

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • —learning_rate: 5e-05
  • —trainbatchsize: 8
  • —evalbatchsize: 8
  • —seed: 42
  • —optimizer: Use OptimizerNames.ADAMWTORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
  • —lrschedulertype: linear
  • —num_epochs: 3.0

Training results

Framework versions

  • —Transformers 4.48.0.dev0
  • —Pytorch 2.5.1
  • —Datasets 3.1.0
  • —Tokenizers 0.21.0