eProsima is proud to announce the release of SustainML v0.1.0, the first official version of their backend and developer framework for sustainable machine learning. A comprehensive backend library and end-to-end developer framework designed to bring environmental awareness to every stage of machine learning. This inaugural release delivers all the core plumbing for orchestrating tasks, querying metadata, profiling energy and latency, and shipping a rich, Docker-based front-end.
The v0.1.0 release delivers a complete, end-to-end framework, including:
- Intuitive Qt-powered Front-End
A fully Docker-deployed GUI for task submission, iterative experiment control, and side-by-side result comparison. - Iteration & Multi-Output Workflows
Built-in support for reiterating non-final tasks and emitting multiple outputs per request, both in code and via the UI. - Containerized Backend Orchestrator
Fine-grained control over individual SustainML nodes, all deployable via Docker Compose or bash script. - DDS-Based Request-Reply Services
On-demand RPC interfaces for querying available ML models, available hardware, carbon-footprint estimators, and more. - Rich Python & C++ Node APIs
SWIG-generated Python bindings alongside native C++ interfaces for rapid prototyping and production deployment. - Secure Hugging Face Integration
Token-protected endpoints for loading any HF model—LLMs, vision, audio, etc.—directly from the hub. - Automated Hardware Profiling
Out-of-the-box energy & latency measurement across all supported devices, with power-aware layer mappings thanks to UPMEM PIM hardware. - Intelligent Model Selection & Metadata Graph
Browse and query ML problems, modalities, models, and associated metrics via an RDF-backed knowledge graph—powered by DFKI—enabling precise, tag-based model search and reiteration. - Carbon Footprint Estimation
Real-time CO₂ emissions tracking for each epoch and inference run, courtesy of the University of Copenhagen’s CarbonTracker library, to quantify and minimize environmental impact. - First-Class CI & Testing
Initial black-box and communication tests, CI pipelines, and reproducible Docker images to keep your builds rock-solid. - High quality documentation
Thorough, up-to-date ReadTheDocs detailing dependencies, installation steps, configuration, and end-to-end usage of the SustainML Framework. Also, in-depth documentation for developers inside the code repositories.
With these foundational building blocks, eProsima equips ML teams to prototype, deploy and monitor sustainable workflows—from backend services to a user-friendly front end—right out of the box.
For further information and to get started, visit the GitHub repository SustainML Framework and SustainML Library or consult the official docs.
MORE INFORMATION ABOUT SUSTAINML:
If you have any questions, please get in touch with
This project has received funding from the European Union’s Horizon Europe research and innovation programme under grant agreement No 101070408.