Finboost / finboost-ml

Finboost Machine Learning is a financial chatbot project designed to answer finance-related questions in Indonesian. Utilizing advanced deep learning models, including a fine-tuned GPT-2 for generative AI and an LSTM for question suggestions, it offers personalized assistance and enhances user engagement.

Finboost Machine Learning

               

Finboost Machine Learning is a financial chatbot project that can answer finance-related questions in Indonesian. The project leverages deep learning models with fine-tuning on the `cahya/gpt2-small-indonesian-522M` model for generative AI and uses an LSTM model for question suggestion.

---

Table of Contents

• Tech Stack
• Architecture
• File and Folder Structure
• Flow Diagrams
• Setup
• Usage
• Contributing

Tech Stack

• Python: Primary programming language used for the project.
• Flask: Web framework for building the REST API.
• TensorFlow: Deep learning framework used for model training and inference.
• Pandas: Library for data manipulation and analysis.
• Transformers: Hugging Face library for working with transformer models.
• NumPy: Library for numerical operations.
• Matplotlib: Library for creating visualizations.
• Google Colab: Cloud service for running Jupyter notebooks.
• GPT-2: Transformer model fine-tuned for generating responses.
• LSTM: Model used for question suggestion based on user input and profile data.

Architecture

This project uses the following architecture:

1. Data Preparation: Preparing question and answer data in CSV format.
2. Model Fine-Tuning: Fine-tuning the cahya/gpt2-small-indonesian-522M model using the dataset for generative AI.
3. LSTM Model: Using an LSTM model for question suggestion.

File and Folder Structure

File/Folder Name	Description
data/generative-ai/finansial-dataset-v2.csv	Example dataset generative-ai in CSV format
data/question-suggestion/data.csv	Example dataset question suggestion in CSV format
notebooks/generative_ai.ipynb	Notebook for fine-tuning and using the generative AI model
notebooks/question_suggestion.ipynb	Notebook for training and using the LSTM model for question suggestion
models/gen-ai	Directory to store the fine-tuned generative AI model (create it yourself by running notebooks/generative_ai.ipynb)
models/question-suggestion	Directory to store the question suggestion model (create it yourself by running notebooks/question_suggestion.ipynb)
preprocessing/combined_dataset.ipynb	Notebook for preprocessing the collected dataset
scripts/	Folder to save the script for future features
generative-ai/	Folder to deploy generative-ai model
generative-ai-v2/	Folder to deploy generative-ai-v2 (Coming soon)
generative-ai-v3/	Folder to deploy generative-ai-v3 (Coming soon)
question-suggestion/	Folder to deploy question-suggestion model
requirements.txt	List of dependencies for this project

Flow Diagrams

Flow Generative AI Model

Flow Question Suggestion Model

Setup

1. Clone the repository:

   git clone https://github.com/username/finboost-ml.git
   cd finboost-ml

2. Create a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`

3. Install dependencies:

   pip install -r requirements.txt

4. Download and place the datasets:

   • Place finansial-dataset-v2.csv in the data/generative-ai/ directory.
   • Place data.csv in the data/question-suggestion/ directory.

5. Create Folder Models:

• Create a models folder in the root directory /:

  mkdir models

  NOTE: When running the Flask application, you must also create model folders within the generative-ai and question-suggestion

• Create models folders in the generative-ai/models and question-suggestion/models directories:

  mkdir -p generative-ai/models
  mkdir -p question-suggestion/models

6. Run the notebooks to obtain the models

   • Open notebooks/generative_ai.ipynb in Google Colab or Jupyter Notebook and execute all the cells to fine-tune the cahya/gpt2-small-indonesian-522M model and generate the model files.
   • Open notebooks/question_suggestion.ipynb in Google Colab or Jupyter Notebook and execute all the cells to train the LSTM model and generate the model files.

7. Download and place the models:

   • Place the fine-tuned generative AI model in the each models/gen-ai/ directory.
   • Place the question suggestion model files (tokenizer.pickle, label_dict.pickle, and model_question_suggestion.h5) in the each models/question-suggestion/ directory.

8. Run the Flask application:

   python generative-ai/app/main.py
   python question-suggestion/app/main.py

Usage

Run the notebook:

1. Run the Generative AI notebook:

• Execute all cells in the generative_ai.ipynb notebook to fine-tune the cahya/gpt2-small-indonesian-522Mmodel and generate responses.

2. Run the Question Suggestion notebook:

• Execute all cells in the question_suggestion.ipynb notebook to train the LSTM model and make question suggestions.

3. Fine-Tuning:

• Use the fine-tuning script generative_ai.py to fine-tune the cahya/gpt2-small-indonesian-522M model on new data.

4. Question Suggestion:

• Use the question_suggestion.py script to train and use the LSTM model for question suggestion.

Contributing

1. Fork the repository
2. Create a new branch (git checkout -b feature-branch)
3. Commit your changes (git commit -am 'Add new feature')
4. Push to the branch (git push origin feature-branch)
5. Create a new Pull Request

---

You can save the above content into the README.md file for your project. This file includes all the essential information about the project, including folder structure, setup steps, and usage instructions.