https://github.com/googleinterns/paksha
Compiling JAX to WebAssembly for exploring client-side machine learning
https://github.com/googleinterns/paksha
compilers jax ml xla
Last synced: 6 months ago
JSON representation
Compiling JAX to WebAssembly for exploring client-side machine learning
- Host: GitHub
- URL: https://github.com/googleinterns/paksha
- Owner: googleinterns
- Created: 2021-02-17T07:29:04.000Z (over 4 years ago)
- Default Branch: main
- Last Pushed: 2022-03-23T16:25:08.000Z (over 3 years ago)
- Last Synced: 2025-02-01T01:41:25.217Z (8 months ago)
- Topics: compilers, jax, ml, xla
- Language: WebAssembly
- Homepage:
- Size: 836 KB
- Stars: 4
- Watchers: 3
- Forks: 0
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# JAX on the Web
## Why run JAX ML models on the Web?
- ***Privacy***: Running on edge means data doesn't have to be sent back to servers and can power local, privacy-first machine learning, such as federated learning or executing models on private PII information.
- ***Low-Latency***: Interacting with client-side data avoids round-trip of server back and forth and allows for computation to be performed on the client's device itself. You can also consider a combination of low-latency local decisions combined with slower, longer round-trip for more powerful models running in the cloud that are queried at a different time scale.
- ***Run Anywhere***: No need to worry about user's OS or user interaction, and no server costs or scaling when exploring demos with users.