Minerva
Minerva is a PyTorch-Lightning-based framework for training machine learning models, designed for researchers. It provides a robust and flexible framework for those working on machine learning projects.
Minerva includes various utilities and modules for:
Data transformation
Model creation
Analysis metrics
Reproducibility
Designed to be modular and extensible, Minerva allows researchers to easily add new features and functionalities.
Features
Minerva offers a wide range of features to support your machine learning projects:
Model Creation Offers a variety of models and architectures to choose from.
Training and Evaluation Provides tools to train and evaluate models, including loss functions, optimizers, and evaluation metrics.
Data Transformation Includes tools for preprocessing and transforming data, such as data loaders, data augmentation, and normalization.
Modular Design Designed to be modular and extensible, allowing easy integration of new features and functionalities.
Reproducibility Ensures reproducibility with tools for versioning, configuration, and experiment logging.
Self-Supervised Learning (SSL) Support Supports Self-Supervised Learning (SSL) for training models with limited labeled data.
Development Environment Provides a pre-configured development environment with all dependencies installed.
Minerva is a framework for training machine learning models for researchers. We provide a set models, datasets, and tools to help you get started with your research. Also, we provide a set of tools to help you configurate experiments, train models, and analyze results in a reproducible way, logging all the information you need to reproduce your experiments.