CI: Abstract the steps to prepare the LLM

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

@@ -0,0 +1,19 @@
+name: Prepare llama.cpp and Phi 2
+description: Build llama.cpp and load Phi 2
+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
+      shell: bash
+      run: ./llama.cpp/server --host 0.0.0.0 -m ./dolphin-2_6-phi-2.Q3_K_M.gguf &

.github/workflows/test-babashka.yml

@@ -13,16 +13,10 @@ jobs:
 
       - run: bb --version
 
-      - name: Get and build llama.cpp
-        run: git clone https://github.com/ggerganov/llama.cpp.git && cd llama.cpp && make
+      - name: Prepare LLM (Phi 2)
+        uses: ./.github/actions/prepare-llm
 
-      - name: Download Phi 2 model
-        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
-        run: ./llama.cpp/server --host 0.0.0.0 -m ./dolphin-2_6-phi-2.Q3_K_M.gguf &
-
-      - name: Wait until the API server is ready
+      - name: Wait until the LLM server is ready
         run: while ! curl -s 'http://localhost:8080/health' | grep 'ok'; do sleep 1; done
         timeout-minutes: 3
 

.github/workflows/test-bun.yml

@@ -14,16 +14,10 @@ jobs:
 
       - run: bun --version
 
-      - name: Get and build llama.cpp
-        run: git clone https://github.com/ggerganov/llama.cpp.git && cd llama.cpp && make
+      - name: Prepare LLM (Phi 2)
+        uses: ./.github/actions/prepare-llm
 
-      - name: Download Phi 2 model
-        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
-        run: ./llama.cpp/server --host 0.0.0.0 -m ./dolphin-2_6-phi-2.Q3_K_M.gguf &
-
-      - name: Wait until the API server is ready
+      - name: Wait until the LLM server is ready
         run: while ! curl -s 'http://localhost:8080/health' | grep 'ok'; do sleep 1; done
         timeout-minutes: 3
 

.github/workflows/test-cpython.yml

@@ -13,16 +13,10 @@ jobs:
         with:
           python-version: '3.10'
 
-      - name: Get and build llama.cpp
-        run: git clone https://github.com/ggerganov/llama.cpp.git && cd llama.cpp && make
+      - name: Prepare LLM (Phi 2)
+        uses: ./.github/actions/prepare-llm
 
-      - name: Download Phi 2 model
-        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
-        run: ./llama.cpp/server --host 0.0.0.0 -m ./dolphin-2_6-phi-2.Q3_K_M.gguf &
-
-      - name: Wait until the API server is ready
+      - name: Wait until the LLM server is ready
         run: while ! curl -s 'http://localhost:8080/health' | grep 'ok'; do sleep 1; done
         timeout-minutes: 3
 

.github/workflows/test-nodejs.yml

@@ -13,16 +13,10 @@ jobs:
         with:
           node-version: '18'
 
-      - name: Get and build llama.cpp
-        run: git clone https://github.com/ggerganov/llama.cpp.git && cd llama.cpp && make
+      - name: Prepare LLM (Phi 2)
+        uses: ./.github/actions/prepare-llm
 
-      - name: Download Phi 2 model
-        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
-        run: ./llama.cpp/server --host 0.0.0.0 -m ./dolphin-2_6-phi-2.Q3_K_M.gguf &
-
-      - name: Wait until the API server is ready
+      - name: Wait until the LLM server is ready
         run: while ! curl -s 'http://localhost:8080/health' | grep 'ok'; do sleep 1; done
         timeout-minutes: 3
 

.github/workflows/test-pypy.yml

@@ -15,16 +15,10 @@ jobs:
 
       - run: pypy3 --version
 
-      - name: Get and build llama.cpp
-        run: git clone https://github.com/ggerganov/llama.cpp.git && cd llama.cpp && make
+      - name: Prepare LLM (Phi 2)
+        uses: ./.github/actions/prepare-llm
 
-      - name: Download Phi 2 model
-        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
-        run: ./llama.cpp/server --host 0.0.0.0 -m ./dolphin-2_6-phi-2.Q3_K_M.gguf &
-
-      - name: Wait until the API server is ready
+      - name: Wait until the LLM server is ready
         run: while ! curl -s 'http://localhost:8080/health' | grep 'ok'; do sleep 1; done
         timeout-minutes: 3