Hi folks,
On behalf of LWS team, we're very excited to announce that we just released lws v0.7.0, including several major features like:
- Partition update support for LWS, which is benefit to canary release and xPyD rolling update for inference service
- Resize capacity support to tune the group size, no longer needed to recreate the lws anymore
- Gang scheduling support with the first integration of volcano, more on the way, like the coscheduling plugin.
Meanwhile, we have new project adopters like Nvidia Dynamio, SGLang OME joined us, 8 new contributors from different companies and communities.
Thanks for all these contributions. Welcome to join us to discuss more about the inference demands, like the xPyD framework on top of LWS.
Best,
Kante