CI: Use pre-built Llamafile

.github/actions/prepare-llm/action.yml

@@ -1,19 +1,23 @@
-name: Prepare llama.cpp and Phi 2
-description: Build llama.cpp and load Phi 2
+name: Prepare Phi 2
+description: Download Phi 2 and launch it
 runs:
   using: "composite"
   steps:
-    - name: Get and build llama.cpp
-      shell: bash
-      run: |
-        git clone https://github.com/ggerganov/llama.cpp.git
-        cd llama.cpp
-        make server
-
     - name: Download Phi 2 model
       shell: bash
       run: curl -OL https://huggingface.co/TheBloke/dolphin-2_6-phi-2-GGUF/resolve/main/dolphin-2_6-phi-2.Q3_K_M.gguf
 
-    - name: Run llama.cpp with Phi 2
+    - name: Setup binfmt with APE
       shell: bash
-      run: ./llama.cpp/server --host 0.0.0.0 -m ./dolphin-2_6-phi-2.Q3_K_M.gguf &
+      run: |
+        sudo wget -O /usr/bin/ape https://cosmo.zip/pub/cosmos/bin/ape-$(uname -m).elf
+        sudo chmod +x /usr/bin/ape
+        sudo sh -c "echo ':APE:M::MZqFpD::/usr/bin/ape:' >/proc/sys/fs/binfmt_misc/register"
+        sudo sh -c "echo ':APE-jart:M::jartsr::/usr/bin/ape:' >/proc/sys/fs/binfmt_misc/register"
+
+    - name: Run Phi 2 with Llamafile
+      shell: bash
+      run: |
+        curl -OL https://github.com/Mozilla-Ocho/llamafile/releases/download/0.6.2/llamafile-0.6.2.zip
+        unzip llamafile-0.6.2.zip
+        ./llamafile-0.6.2/bin/llamafile -m ./dolphin-2_6-phi-2.Q3_K_M.gguf &