change to sglang

2026-03-21 17:04:05 +08:00
parent b48d522bb3
commit bd67fc088f
6 changed files with 2468 additions and 1534 deletions
--- a/llamacpp-nemotron-super.sh
+++ b/llamacpp-nemotron-super.sh
@@ -7,10 +7,8 @@ MODEL="/home/kongcunliang/workspace/pretrained-models/unsloth/NVIDIA-Nemotron-3-
 llama-server \
  --model $MODEL \
  --alias "nemotron-3-super" \
-  --prio 1 \
-  --min_p 0.01 \
  --temp 0.6 \
  --top-p 0.95 \
-  --ctx-size 1048576 \
+  --ctx-size 262144 \
  --host 0.0.0.0 \
-  --port 8001
+  --port 8001 \