Mozilla.ai Lumigator

Lumigator is an open-source platform built by Mozilla.ai for guiding users through the process of selecting the right language model for their needs.

Currently, we support evaluating summarization tasks using sequence-to-sequence models, like BART, and causal architectures, like Mistral.

These models can either be loaded and run by Lumigator (if they are models stored in Huggingface Hub), or the models can be reached through any OpenAI compatible interface (for example, llamafile or OpenAI GPT models). Lumigator will soon support additional tasks such as translation (See #628).

Note

Lumigator is in the early stages of development. It is missing important features and documentation. You should expect breaking changes in the core interfaces and configuration structures as development continues.

Overview

Lumigator contains several different components and can be used in a few different ways.

The core functionality is the backend, which is a Python-based FastAPI server that features REST API endpoints.

These REST API endpoints can be accessed by a variety of clients, including:

The backend is responsible for receiving requests from clients and managine the lifecycle of datasets, experiments, workflows, and jobs. The backend also coordinates interaction with a Ray cluster and a database. The Ray cluster is used to run jobs, and the database is used to store metadata about the platform, jobs, and datasets.

For more information about datasets, see Datasets.

For more information about experiments, see Experiments.

For more information about workflows, see Workflows.

For more information about jobs, see Jobs.

Get Started

Operations Guide

User Guides

Conceptual Guides

Reference