Perceval-Quest-Solution

The Perceval Quest Solution by Team Quantum Tree.

One-page Report

View the PDF

Overview

This repository contains the solution for the Perceval Quest by Team Quantum Tree. It implements the GLASE architecture and a classical CNN baseline for partial MNIST classification. The solution uses a BosonSampler to simulate quantum circuits and supports evaluation on a QPU via Scaleway. The project includes scripts for training, plotting results, and QPU validation.

Directory Structure

.
├── requirements.txt
├── README.md
└── src
    ├── train.py          # Training script with simulator.
    ├── plot_result.py    # Script to plot detailed training metrics.
    └── qpu_validation.py # Script to evaluate the trained model on a QPU session.

Installation

Before running the scripts, install the required Python packages by executing:

pip install -r requirements.txt

Usage

Training the Model

Run the training script with default hyperparameters or specify your own using command-line arguments. For example:

python src/train.py --m 20 --n 3 --batch_size 256 --lr 2e-3 --weight_decay 1e-3 --epochs 50 --label_smoothing 0.1

If no arguments are provided, the script will use the default values. Run python src/train.py -h for detailed arguments.

Plotting Training Results

To visualize the training metrics after training both QNN and CNN, use the plotting script. By default, it uses qnn.pkl and cnn.pkl files. You can override these by providing your own file paths:

python src/plot_result.py --files qnn.pkl cnn.pkl

QPU Validation

Evaluate the model using a QPU session with the qpu_validation.py script. You will need to provide your Scaleway credentials and configuration parameters. For example:

python src/qpu_validation.py --scw_project_id YOUR_PROJECT_ID --scw_token YOUR_TOKEN --filename qnn.pkl --platform qpu:ascella --m 12 --n 3 --fraction 0.2 --batch_size 2

Replace YOUR_PROJECT_ID and YOUR_TOKEN with your actual Scaleway credentials.

Requirements

All required packages are listed in the requirements.txt file. Make sure to install them before running any scripts.