FLANG is a set of large language models for Financial LANGuage tasks. These models use domain specific pre-training with preferential masking to build more robust representations for the domain. The models in the set are:
FLANG-BERT
FLANG-SpanBERT
FLANG-DistilBERT
FLANG-Roberta
FLANG-ELECTRAFLANG-DistilBERT is a pre-trained language model which uses financial keywords and phrases for preferential masking of domain specific terms. It is built by further training the DistilBERT language model in the finance domain with improved performance over previous models due to the use of domain knowledge and vocabulary.
FLUE (Financial Language Understanding Evaluation) is a comprehensive and heterogeneous benchmark that has been built from 5 diverse financial domain specific datasets.
Sentiment Classification:
Financial PhraseBank
Sentiment Analysis, Question Answering:
FiQA 2018
New Headlines Classification:
Headlines
Named Entity Recognition:
NER
Structure Boundary Detection:
FinSBD31@INPROCEEDINGS{shah-etal-2022-flang,
2 author = {Shah, Raj Sanjay and
3 Chawla, Kunal and
4 Eidnani, Dheeraj and
5 Shah, Agam and
6 Du, Wendi and
7 Chava, Sudheer and
8 Raman, Natraj and
9 Smiley, Charese and
10 Chen, Jiaao and
11 Yang, Diyi },
12 title = {When FLUE Meets FLANG: Benchmarks and Large Pretrained Language Model for Financial Domain},
13 booktitle = {Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
14 year = {2022},
15 publisher = {Association for Computational Linguistics}
16}
Please contact Raj Sanjay Shah (rajsanjayshah[at]gatech[dot]edu) or Sudheer Chava (schava6[at]gatech[dot]edu) or Diyi Yang (diyiy[at]stanford[dot]edu) about any FLANG-DistilBERT related issues and questions.
