First author: Wei Luo(罗威)

Pre-trained Large Language Models for Financial Sentiment Analysis.

This project contains codes for the research paper title "Pre-trained Large Language Models for Financial Sentiment Analysis". Authors: Wei Luo, Dihong Gong.

Environment setup

• hostfile.txt should contain the IP addresses for distributed training, one line per IP.
• env_a800.sh should contain all the additional custom environment variables for running the codes.

Data preprocessing

• Split the train, val and test sets for the PhraseBank dataset: process_financial_phrasebank.py

Training

• May need to tune some parameters accordingly.
• Execute the training script with ./train.sh

Testing

• Execute the testing script with torchrun test.py

Results

Methods	Accuracy
LSTM	0.71
LSTM with ELMo	0.75
ULMFit	0.83
LPS	0.71
HSC	0.71
FinBERT	0.86
Ours (Few-shots)	0.68
Ours (SFT)	0.90
Ours (3-class classification)	0.90