NG: llamacpp on the ai node (Qwen3.8-27B) with the AMD GPU device plugin
Update Kubernetes Services Wiki / Generate and Update K8s Wiki (push) Successful in 44s
Update Kubernetes Services Wiki / Generate and Update K8s Wiki (push) Successful in 44s
Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
1 parent
bd88daa109
commit
373ce77ac8
7 files changed
+169
No files matched your search
@@ -0,0 +1,17 @@
|
||||
apiVersion: v1
|
||||
kind: ConfigMap
|
||||
metadata:
|
||||
name: llamacpp-config
|
||||
data:
|
||||
LLAMA_CACHE: /models
|
||||
LLAMA_ARG_HOST: 0.0.0.0
|
||||
LLAMA_ARG_PORT: "8080"
|
||||
# The model is chosen here: <hf-user>/<repo>:<quant>. It is downloaded into
|
||||
# /models on first start; changing it restarts the pod (reloader).
|
||||
LLAMA_ARG_HF_REPO: "unsloth/Qwen3.8-27B-GGUF:UD-Q6_K"
|
||||
LLAMA_ARG_CTX_SIZE: "131072"
|
||||
LLAMA_ARG_N_GPU_LAYERS: all
|
||||
LLAMA_ARG_FLASH_ATTN: "on"
|
||||
LLAMA_ARG_FIT: "on"
|
||||
# Strix Halo: mmap-ed weights hang or corrupt on ROCm; read them directly.
|
||||
LLAMA_ARG_LOAD_MODE: dio
|
||||
Reference in new issue
Block a user