ruchtem/cosmos
This is the official implementation for COSMOS: a method to learn Pareto fronts that scales to large datasets and deep models.
This project helps machine learning researchers and practitioners efficiently find the best trade-offs when training deep learning models that need to optimize multiple, often conflicting, objectives. It takes common datasets like CelebA or Multi-MNIST and a choice of multi-objective optimization algorithms, outputting a 'Pareto front' which visually represents the optimal set of solutions where no single objective can be improved without sacrificing another. This is for anyone working on deep learning models with competing goals.
No commits in the last 6 months.
Use this if you are training deep neural networks and need to systematically explore and visualize the best possible compromises across several performance metrics, rather than optimizing just one.
Not ideal if you are working with non-deep learning models, single-objective optimization problems, or if you don't have access to a CUDA-capable GPU.
Stars
39
Forks
4
Language
Python
License
MIT
Category
Last pushed
May 25, 2021
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/ruchtem/cosmos"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
kaanaksit/odak
Scientific computing library for optics, computer graphics and visual perception.
NVIDIA/torch-harmonics
Differentiable signal processing on the sphere for PyTorch
PreFab-Photonics/PreFab
Artificial nanofabrication of integrated photonic circuits using deep learning
MatthewFilipovich/torchoptics
Differentiable wave optics simulation library built on PyTorch
artificial-scientist-lab/XLuminA
XLuminA, a highly-efficient, auto-differentiating discovery framework for super-resolution microscopy.