39 lines
1.4 KiB
Bash
Executable File
39 lines
1.4 KiB
Bash
Executable File
#!/bin/bash
|
||
# llama-server 模型切换开关：3b ↔ 7b ↔ off
|
||
# 用法: llama-switch.sh 3b|7b|off|status
|
||
# 两个服务共用 :8080 端口，同一时刻只能有一个运行
|
||
|
||
case "$1" in
|
||
3b)
|
||
systemctl --user stop llama-server-7b 2>/dev/null
|
||
systemctl --user start llama-server
|
||
echo "✅ 已切换到 3B (Qwen2.5-3B Q4_K_M)"
|
||
;;
|
||
7b)
|
||
systemctl --user stop llama-server
|
||
systemctl --user start llama-server-7b
|
||
echo "✅ 已切换到 7B (Qwen2.5-7B Q3_K_M)"
|
||
;;
|
||
off)
|
||
systemctl --user stop llama-server
|
||
systemctl --user stop llama-server-7b
|
||
echo "🛑 本地 LLM 服务已停止"
|
||
;;
|
||
status)
|
||
_s3b=$(systemctl --user is-active llama-server 2>/dev/null)
|
||
_s7b=$(systemctl --user is-active llama-server-7b 2>/dev/null)
|
||
if [ "$_s3b" = "active" ]; then
|
||
echo "🟢 当前: 3B (Qwen2.5-3B Q4_K_M) @ :8080"
|
||
echo " 显存: $(nvidia-smi --query-gpu=memory.used --format=csv,noheader,nounits) MiB"
|
||
elif [ "$_s7b" = "active" ]; then
|
||
echo "🟢 当前: 7B (Qwen2.5-7B Q3_K_M) @ :8080"
|
||
echo " 显存: $(nvidia-smi --query-gpu=memory.used --format=csv,noheader,nounits) MiB"
|
||
else
|
||
echo "⚪ 本地 LLM 服务未运行"
|
||
fi
|
||
;;
|
||
*)
|
||
echo "用法: llama-switch.sh 3b|7b|off|status"
|
||
;;
|
||
esac
|