Define four-profile production matrix with Medium default
This commit is contained in:
@@ -21,7 +21,8 @@ install -m 0644 "$PLATFORM/systemd/mike-ai-llama-ui.service" \
|
||||
install -m 0755 "$PLATFORM/scripts/llama-profile" /usr/local/bin/llama-profile
|
||||
install -m 0644 "$PLATFORM/profiles/profile-fast.conf" "$PROFILE_TARGET/profile-fast.conf"
|
||||
install -m 0644 "$PLATFORM/profiles/profile-medium.conf" "$PROFILE_TARGET/profile-medium.conf"
|
||||
install -m 0644 "$PLATFORM/profiles/profile-long.conf" "$PROFILE_TARGET/profile-long.conf"
|
||||
install -m 0644 "$PLATFORM/profiles/profile-large.conf" "$PROFILE_TARGET/profile-large.conf"
|
||||
install -m 0644 "$PLATFORM/profiles/profile-ultra.conf" "$PROFILE_TARGET/profile-ultra.conf"
|
||||
rsync -a --delete "$PLATFORM/mcp/" "$MCP_TARGET/"
|
||||
rsync -a --delete --exclude searxng-settings.yml \
|
||||
"$PLATFORM/web-search/" "$MCP_SEARCH_TARGET/"
|
||||
@@ -39,7 +40,7 @@ Kernkonfiguration installiert, aber noch nicht gestartet.
|
||||
|
||||
Vor dem Start:
|
||||
1. Modellpfade und Hashes gegen manifest.local.yaml prüfen.
|
||||
2. llama.cpp bauen und mit `llama-profile fast` starten.
|
||||
2. llama.cpp bauen und mit `llama-profile medium` starten.
|
||||
3. Fach-Secrets unter /etc/mike-ai ablegen.
|
||||
4. Danach: /opt/mike-ai/mcp-containers/platform/mcp/install-tools.sh
|
||||
EOF
|
||||
|
||||
Reference in New Issue
Block a user