Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
hyperbovine
on Oct 21, 2022
|
parent
|
context
|
favorite
| on:
Building the Future of TensorFlow
I believe it is, whether they realize it or not. JAX/PyTorch + XLA is just so much better in so many ways. Development on XLA will (thankfully) continue, and JAX and PyTorch will continue to cannibalize the TF userbase.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: