Comments (1)
Hi,
Thanks for trying out FedJAX and filing this issue!
I think this would mainly be an optax functionality since FedJAX optimizers just wrap around existing optax optimizers.
If you want a different client learning rate for each round of federated averaging, here are few suggestions:
- Pass round number as part of the server_state in the federated algorithm and then create a new client optimizer with learning rate based on it at each round of federated training. (caveat: potentially slow due to effectively recompiling client optimizer apply function each round)
- Optax has built-in support for learning rate decay e.g., optax.exponential_decay. Some of these schedulers can be used directly / tweaked to support your use cases.
Some potentially helpful links:
- https://github.com/deepmind/optax#schedules-schedulepy (intro to schedulers in optax)
- https://github.com/google/fedjax/blob/main/fedjax/experimental/notebooks/emnist_compression.ipynb (example using scheduler for server optimizer)
If you have more concrete examples of what you're trying to do, we'd be happy to look into it!
from fedjax.
Related Issues (17)
- Clarifying the meaning of "weight" HOT 1
- Centralized (server-only) algorithm HOT 1
- Implementing SCAFFOLD HOT 2
- Feature request: Convert standard dataset into a federated dataset HOT 5
- PatentMime^
- Full EMNIST example does not exhibit parallelization HOT 2
- FedJax depends on TensorFlow Federated? HOT 7
- How to create a validation dataset? HOT 4
- Problem of Quick Start in Readme.md HOT 2
- Support for haiku models with non-trainable state HOT 2
- Support for gldv2 and inaturalist datasets HOT 7
- CIFAR 100 Questions HOT 5
- Implement standard CIFAR-100 model in fedjax.models.cifar100 HOT 1
- Add support for stateful clients HOT 8
- Error of the Stackoverflow Tokernizer example HOT 1
- External contributions? HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from fedjax.