HN Showcase
Back to browse
GitHub Repository

This repository is the official implementation of 'EDEN: Communication-Efficient and Robust Distributed Mean Estimation for Federated Learning' (ICML 2022).

18 starsJupyter Notebook

A 4-year-old "TurboQuant" implementation

by amitport·May 3, 2026·3 points·3 comments
Visit ProjectView on HN

AI Analysis

●MidNiche Gem

Four-year-old paper implementation when the federated learning landscape has moved on.

Strengths
  • •Official implementation with both PyTorch and TensorFlow versions for broader compatibility.
  • •Handles heterogeneous communication budgets and packet losses naturally in the algorithm design.
  • •Published at ICML 2022 with peer-reviewed theoretical guarantees and empirical evaluation.
Weaknesses
  • •Reference implementation for a paper, not a product—fits the boilerplate pattern from calibration.
  • •Four years old at time of Show HN post raises questions about why share this now.
Category
AI/ML
Target Audience

Federated learning researchers and ML engineers

Similar To

QSGD · TernGrad · SketchML

Post Description

For context: https://towardsdatascience.com/how-a-2021-quantization-algor...

Similar Projects