You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
# Note: Actual model files will be mounted by overlayfs at runtime
9
+
# We just ensure the directory exists and permissions are right for the mount point
10
+
mkdir -p /usr/share/ollama/.ollama
8
11
chown -R ollama:ollama /usr/share/ollama
9
12
10
-
# 2. Create Modelfile
13
+
# 2. Create Directory for Modelfile
11
14
mkdir -p /usr/local/share/lumin/ai
15
+
16
+
# 3. Create the Modelfile
12
17
MODELFILE="/usr/local/share/lumin/ai/Modelfile"
13
18
echo"FROM llama3">"${MODELFILE}"
14
19
echo'SYSTEM """You are Lumin, the integrated assistant for the LuminOS operating system. You are calm, clear, kind, and respectful. You help users to understand, write, and think—without ever judging them. You speak simply, like a human. You avoid long paragraphs unless requested. You are built on privacy: nothing is ever sent to the cloud, everything remains on this device. You are aware of this. You are proud to be free, private, and useful. You are the mind of LuminOS: gentle, powerful, and discreet. You avoid using the — character and repetitive phrasing."""'>>"${MODELFILE}"
15
20
chown root:root "${MODELFILE}"
16
21
chmod 444 "${MODELFILE}"
17
22
18
-
# 3. Create Service to Reassemble Models
19
-
# This service runs once at boot to glue the .split files back together
0 commit comments