search

FedML-AI / FedML

FEDML - The unified and scalable ML library for large-scale distributed training, model serving, and federated learning. FEDML Launch, a cross-cloud scheduler, further enables running any AI jobs on any GPU cloud or on-premise cluster. Built on this library, TensorOpera AI (https://TensorOpera.ai) is your generative AI platform at scale.
https://TensorOpera.ai
Apache License 2.0
4.11k stars 773 forks source link

Deployment of FedGKT on IoT devices and use AWS as server #146

Open shanullah opened 2 years ago

shanullah commented 2 years ago

I have few more questions, I am planning to deploy the FedGKT at IoT devices (Jetson nano, Tx2, and Xavier). I read the paper (fedGKT), where you have used cpu as edge (client) and gpu as server for aggregation. However, in the case of deploying the fedGKT at IoT devices, what kind of issues do you foresee. Although I will try my best to do it, as your IoT example cover nano and raspberry-pi (without fedGKT support), will that work on Xavier and Tx2 too ?and I will deploy server as AWS Cloud for testing in real environment to study the impact of lightweight models on bandwidth and latency . your guidance and recommendation is required. what steps shall I follow to do it smoothly?

KOUDA-AMINE commented 2 years ago

hi @shanullah, did you succeed in the deployment of FedGKT on IoT devices?

fedml-dimitris commented 8 months ago

Hello @shanullah and @KOUDA-AMINE were you able to run: https://github.com/FedML-AI/FedML/blob/master/python/fedml/simulation/mpi/fedgkt/FedGKTAPI.py ?

Also have a look here: https://github.com/FedML-AI/FedML/tree/master/iot