/NLG_Sentiment_Analysis

Facebook BART Summarizer and LLM Engineering for Stocks/Crypto sentiment analysis from Financial News Outlets

Primary LanguagePythonMIT LicenseMIT

Financial Sentiment Analysis : OpenAI, and Llama and GAI (Amazon Bedrock)

In this project, I am testing the accuracy of all the models and see which one is better for interpreting financial documents/news

BERT & BART

Machine Learning Project that uses BERT Deep Learning and Encoder-Decoder NLP NB : This ML model is Novel and may occasionally generate wrong Data

Theoretical Background

description

Architectures

description description description

Installing & Usage

description

  1. Clone the repo to your local machine ; Open in VScode (Have Jupyter Environment installed)
  2. Open Terminal and Run pip install requirements.txt
  3. In the same terminal window, run trainer.py and then run streamlit run app.py
  4. As on the window above, a stock ticker is the 4-code stock symbol i.e APPL
  5. The model takes a few minutes to train and output per stock ticker
  6. To understand deeply how the model works, Investigate the .ipynb file

Limitations

  1. The Framework maybe inaccurate for some sentiments analysis so the analyst needs to confirm before making final decision

Llama70B

under construction

OpenAI Docureader & sentiment analysis

Under construction