Use container right now

ramalama run/serve right now require the container, it has the version
of llama.cpp that works.

Long-term we may be able to remove this.

Signed-off-by: Eric Curtin <[email protected]>

ci.sh

@@ -43,6 +43,13 @@ main() {
   ./ramalama list | grep tinyllama
   ./ramalama list | grep tiny-vicuna-1b
   ./ramalama list | grep NAME
+
+  if [ "$os" = "Linux" ]; then # no macos support for running/serving models yet
+    set +o pipefail
+    timeout -s9 16 ./ramalama serve tinyllama | grep -m1 -i listen
+    set -o pipefail
+  fi
+
 #  ramalama list | grep granite-code
 #  ramalama rm granite-code
 }

install.sh

@@ -49,6 +49,7 @@ main() {
   local url="raw.githubusercontent.com/containers/ramalama/main/$from"
   local from="$TMP/$from"
   download
+  pip install "huggingface_hub[cli]==0.24.2"
   install -D -m755 "$from" "$bindir/"
 
   if false; then # to be done

ramalama

@@ -215,6 +215,7 @@ def list_cli(ramalama_store, args):
 
 
 funcDict["list"] = list_cli
+funcDict["ls"] = list_cli
 
 
 def pull_huggingface(model, ramalama_store):
@@ -343,6 +344,11 @@ def main(args):
     ramalama_store = get_ramalama_store()
 
     try:
+        cmd = args[0]
+        if conman and (cmd == 'serve' or cmd == 'run'):
+            conman_args = [conman, "run", "--rm", "-it", "--security-opt=label=disable", f"-v{ramalama_store}:/var/lib/ramalama", f"-v{os.path.expanduser('~')}:{os.path.expanduser('~')}", "-v/tmp:/tmp",
+                           f"-v{__file__}:{__file__}", "quay.io/ramalama/ramalama:latest", __file__] + args
+            os.execvp(conman, conman_args)
         cmd = args.pop(0)
         funcDict[cmd](ramalama_store, args)
     except IndexError: