Lets keep those models loaded for a bit longer.

This commit is contained in:
Joseph Hanson 2024-07-17 14:10:25 -05:00
parent ba9afaad72
commit 09642a8de9
Signed by: jahanson
SSH key fingerprint: SHA256:vy6dKBECV522aPAwklFM3ReKAVB086rT3oWwiuiFG7o

View file

@ -43,6 +43,8 @@ spec:
value: "*"
- name: OLLAMA_MODELS
value: &modelPath "/models"
- name: OLLAMA_KEEP_ALIVE
value: "24h"
resources:
requests:
nvidia.com/gpu: 1 # requesting 1 GPU