Support OCI "modelcar" sidecars for model loading
Support OCI "modelcar" sidecars for model loading: a task in MiMo-V2.6-RL-oss: Agentic RL Environments (MiMo RL release). Today, when an InferenceService pod is admitted, model data is provisioned by an init container that copies the model from the configured storageUri into a shared volume before…
The task
Today, when an InferenceService pod is admitted, model data is provisioned by an init container that copies the model from the configured storageUri into a shared volume before the serving container starts. For large models this copy dominates the cold-start time, and the data…
Part of XiaomiMiMo/MiMo-V2.6-RL-oss.