From d160219c6bf7fb3326a7411324750b02d6c891d5 Mon Sep 17 00:00:00 2001
From: Alexander Rogov <i_am@rogov.al>
Date: Tue, 9 Jun 2026 20:28:21 +0300
Subject: [PATCH] adjust llama gpu params

---
 manifests/llama/main.yaml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/manifests/llama/main.yaml b/manifests/llama/main.yaml
index 5b191af..aca6269 100644
--- a/manifests/llama/main.yaml
+++ b/manifests/llama/main.yaml
@@ -93,9 +93,9 @@ spec:
 
             # performance tuning
             - "--ctx-size"
-            - "32768"
+            - "24576"
             - "--parallel"
-            - "4"
+            - "2"
 
             # KV cache quantization
             - "--cache-type-k"