feat: add ai-stack Kubernetes manifests for Qdrant, vLLM, and Triton
This commit is contained in:
@@ -249,6 +249,13 @@ spec:
|
|||||||
args:
|
args:
|
||||||
- |
|
- |
|
||||||
rm -rf /usr/local/lib/python3.12/dist-packages/torchcodec*
|
rm -rf /usr/local/lib/python3.12/dist-packages/torchcodec*
|
||||||
|
python3 -c '
|
||||||
|
p = "/usr/local/lib/python3.12/dist-packages/vllm/v1/attention/ops/triton_unified_attention.py"
|
||||||
|
s = open(p).read()
|
||||||
|
s = s.replace("return 32", "return 16")
|
||||||
|
s = s.replace("launch_num_stages: int | None = None", "launch_num_stages: int | None = 1")
|
||||||
|
open(p, "w").write(s)
|
||||||
|
'
|
||||||
exec vllm serve google/gemma-4-12B-it-qat-w4a16-ct \
|
exec vllm serve google/gemma-4-12B-it-qat-w4a16-ct \
|
||||||
--served-model-name Gemma-4-12B-it \
|
--served-model-name Gemma-4-12B-it \
|
||||||
--dtype half \
|
--dtype half \
|
||||||
|
|||||||
Reference in New Issue
Block a user