Skip to content

Commit 880dbb6

Browse files
committed
Update LoRAs env
Signed-off-by: Shiva Krishna, Merla <smerla@nvidia.com>
1 parent 8b90c91 commit 880dbb6

File tree

2 files changed

+12
-0
lines changed

2 files changed

+12
-0
lines changed

config/samples/nim/serving/standalone/lora/llm.yaml

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -16,6 +16,12 @@ spec:
1616
env:
1717
- name: NIM_PEFT_SOURCE
1818
value: "/model-store/<lora-directory>" # Update with the absolute <lora-directory> path inside the PVC
19+
- name: NIM_PEFT_REFRESH_INTERVAL
20+
value: "180"
21+
- name: NIM_MAX_CPU_LORAS
22+
value: "16"
23+
- name: NIM_MAX_GPU_LORAS
24+
value: "8"
1925
storage:
2026
pvc: # Pre-created PVC with downloaded loras
2127
name: meta-llama-3-2-1b-instruct

config/samples/nim/serving/standalone/lora/multi-llm.yaml

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -17,6 +17,12 @@ spec:
1717
env:
1818
- name: NIM_PEFT_SOURCE
1919
value: "/model-store/<lora-dir>" # Update with the absolute <lora-dir> path inside the PVC
20+
- name: NIM_PEFT_REFRESH_INTERVAL
21+
value: "180"
22+
- name: NIM_MAX_CPU_LORAS
23+
value: "16"
24+
- name: NIM_MAX_GPU_LORAS
25+
value: "8"
2026
storage:
2127
pvc: # Pre-created PVC with downloaded loras
2228
name: nim-cache-multi-llm

0 commit comments

Comments
 (0)