feat: add ai-stack Kubernetes manifests for Qdrant, vLLM, and Triton

2026-11-08 15:03:24 -05:00
parent be820c6b92
commit 7b2dd24526
1 changed files with 7 additions and 0 deletions
--- a/manifests/spacedock/ai-stack/ai-stack.yaml
+++ b/manifests/spacedock/ai-stack/ai-stack.yaml
@@ -249,6 +249,13 @@ spec:
          args:
            - |
              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 \
                --served-model-name Gemma-4-12B-it \
                --dtype half \