substratusai / kubeai

AI Inference Operator for Kubernetes
https://www.kubeai.org
Apache License 2.0
560 stars 46 forks source link

Configurable start up timeout #176

Open samos123 opened 3 months ago

samos123 commented 3 months ago

Mike: Any chance this could be a configurable component? Networks will all differ when using this on any system outside of a major cloud. Llama3.1 8B is 4.6GB (https://ollama.com/library/llama3.1:8b) of storage. 20Mbps (which used to be the standard for FCC "Broadband") would be 30min of download if you have the whole bandwidth to work with (Calculator)

nstogner commented 3 months ago

We should likely make this more generic: How do we want to surface how admins can customize Pod created by KubeAI?