Symbolic Calculations Project

This project focuses on symbolic calculation tasks using deep learning models, specifically Long Short-Term Memory (LSTM) and Transformer models. The tasks include dataset preprocessing, training LSTM and Transformer models to learn Taylor expansions of mathematical functions, and providing predictions.

symbolicAI/
│
├── datasets/
│   ├── Data.py           # Code for dataset cleaning, tokenization, etc.
│   ├── _init_.py
│   ├── registry.py       # All datasets must be registered
│   └── utils.py          # Helper modules
│
├── engine/
│   ├── _init_.py
│   ├── config.py         # Configuration for model training
│   ├── plotter.py        # Plotting utility for loss and accuracy
│   ├── predictor.py      # Prediction utility
│   ├── trainer.py        # Model training utility
│   └── utils.py          # Helper modules
│
├── models/
│   ├── BART.py           # Code for BART model
│   ├── LED.py            # Code for Longformer Encoder Decoder model
│   ├── _init_.py
│   ├── registry.py       # All models must be registered
│   └── seq2seq_transformer.py  # Code for Sequence-to-Sequence Transformer model
│
├── runs/
│   ├── bart-base_trainer.sh    # Script to run BART-base model from terminal
│   ├── seq2seq_trainer.sh      # Script to run Sequence-to-Sequence Transformer from terminal
│
├── symba_trainer.py   # Trainer script for use inside bash scripts
├── symba_tuner.py     # Hyperparameter optimization using Optuna
├── symba_example.ipynb   # Example notebook
└── README.md           # Project documentation

Components

1. Dataset Preprocessing (Common Task 1)

The dataset.py module contains functionalities for dataset creation and tokenization. Data Class: Defines functions to generate datasets using Sympy, tokenize the dataset, and obtain the token dictionary.

2. LSTM Model (Common Task 2)

The LSTM model is implemented in the model.py module. LSTMModel Class: Defines the architecture and training process for the LSTM model.

3. Transformer Model (Specific Task 3)

The Transformer model is also implemented in the model.py module. TransformerModel Class: Defines the architecture and training process for the Transformer model.

4. Training (train.py)

The train.py module contains functionalities for training the LSTM and Transformer models. Train Class: Facilitates training and obtaining trained models for prediction.

5. Utilities (utils.py)

Utility functions for dataset creation and processing are defined in utils.py. TrainDataset Class: Creates PyTorch dataset for training. TestDataset Class: Creates PyTorch dataset for testing. Predict Class: Provides prediction for any function using trained models.

6. Symbolic_AI.ipynb

The Jupyter notebook demonstrates the entire project workflow, including dataset creation, tokenization, model training, and prediction. Usage Ensure necessary dependencies are installed (PyTorch, Sympy, etc.). Run the notebook Symbolic_AI.ipynb to execute the project workflow. Follow the instructions provided in the notebook for dataset creation, model training, and prediction.