Model composition with serve deployments, why?

Ray Libraries (Data, Train, Tune, Serve)

Li_Bin April 1, 2023, 1:07pm 1

since I can use ingress depoyment + plain tasks/actors, what is the advantage to use ingress deployment + nested depolyments ? Are there any cases where the later is the best choice?

Topic		Replies	Views
[Ray Serve] using GRPC and DAG to host multiple models(or actors) in the same deployment Ray Libraries (Data, Train, Tune, Serve)	3	401	February 2, 2023
Ray serve with dynamic deployments Ray Libraries (Data, Train, Tune, Serve)	0	535	September 23, 2022
Ray serve on Kubernetes Ray Serve	14	834	March 27, 2024
[Serve] New API not as good as old one for programmatic deployment Ray Libraries (Data, Train, Tune, Serve)	0	294	October 5, 2022
Running 10+ models on a ray cluster Kubernetes	1	552	February 27, 2022

Model composition with serve deployments, why?

Related topics