Files
ubicloud/rhizome/inference_endpoint/bin/setup-replica
Benjamin Satzger 2da7d5aea3 Use cpu version of vLLM if gpu count is zero
In a previous commit we extended the AI boot image to also include a cpu
version of vLLM. In this change, we start to use the cpu version if the
gpu count is zero.
2025-02-06 09:16:55 +01:00

816 B
Executable File