UNITES-Lab/HEXA-MoE
Official code for the paper "HEXA-MoE: Efficient and Heterogeneous-Aware MoE Acceleration with Zero Computation Redundancy"
This project offers an optimized way to run large AI models that use a 'Mixture of Experts' (MoE) architecture. It takes your existing MoE model and processes it more efficiently, especially on systems with different types of computing hardware. The result is faster model execution and lower memory use, benefiting AI researchers and engineers working with large-scale deep learning models.
No commits in the last 6 months.
Use this if you are training or deploying large Mixture-of-Experts (MoE) models and need to reduce memory consumption or speed up computation, especially on systems with a mix of different GPUs or accelerators.
Not ideal if you are working with smaller, non-MoE deep learning models or do not have access to heterogeneous computing environments.
Stars
15
Forks
2
Language
Python
License
—
Category
Last pushed
Mar 06, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/transformers/UNITES-Lab/HEXA-MoE"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
EfficientMoE/MoE-Infinity
PyTorch library for cost-effective, fast and easy serving of MoE models.
raymin0223/mixture_of_recursions
Mixture-of-Recursions: Learning Dynamic Recursive Depths for Adaptive Token-Level Computation...
AviSoori1x/makeMoE
From scratch implementation of a sparse mixture of experts language model inspired by Andrej...
thu-nics/MoA
[CoLM'25] The official implementation of the paper
jaisidhsingh/pytorch-mixtures
One-stop solutions for Mixture of Expert modules in PyTorch.