feat: Automated Preset Docker Image Building

.github/workflows/preset-image-build.yml

@@ -0,0 +1,120 @@
+name: Build and Push Preset Models
+
+on:
+  push:
+    branches:
+      - main
+      - Ishaan/auto-image-build
+    paths:
+      - 'pkg/presets/falcon/**'
+      - 'pkg/presets/llama-2/**'
+      - 'pkg/presets/llama-2-chat/**'
+  workflow_dispatch:
+    inputs:
+      release:
+        description: 'Release (yes/no)'
+        required: true
+        default: 'no'
+      image_tag:
+        description: 'Image Tag'
+        required: false
+
+permissions:
+  id-token: write
+  contents: read
+
+jobs:
+  setup:
+    runs-on: ubuntu-20.04
+    outputs:
+      image_tag: ${{ steps.set_tag.outputs.image_tag }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v2
+
+      - name: Install Azure CLI latest
+        run: |
+          curl -sL https://aka.ms/InstallAzureCLIDeb | sudo bash
+
+      - uses: azure/login@v1
+        with:
+          client-id: ${{ secrets.AZURE_CLIENT_ID }}
+          tenant-id: ${{ secrets.AZURE_TENANT_ID }}
+          allow-no-subscriptions: true
+
+      - name: Set Image Tag
+        id: set_tag
+        run: |
+          if [[ "${{ github.event_name }}" == "workflow_dispatch" && -n "${{ github.event.inputs.image_tag }}" ]]; then
+            echo "::set-output name=image_tag::${{ github.event.inputs.image_tag }}"
+          else
+            echo "::set-output name=image_tag::$(git rev-parse --short HEAD)"
+          fi
+
+      - name: 'Login to ACR'
+        run: az acr login --name aimodelsregistry
+
+  falcon:
+    needs: setup
+    runs-on: ubuntu-20.04
+    if: contains(github.event.commits[0].modified, 'pkg/presets/falcon/')
+    steps:
+      - name: Build and push Falcon model
+        run: |
+          cd docker/presets/falcon
+          az acr build -t aimodelsregistry.azurecr.io/falcon:${{ needs.setup.outputs.image_tag }} -r aimodelsregistry .
+
+  llama-2-7b:
+    needs: setup
+    runs-on: ubuntu-20.04
+    if: contains(github.event.commits[0].modified, 'pkg/presets/llama-2/')
+    steps:
+      - name: Build and push Llama model
+        run: |
+          az acr build --build-arg LLAMA_VERSION=llama-2-7b --build-arg SRC_DIR=pkg/presets/llama-2 -t aimodelsregistry.azurecr.io/llama-2-7b:${{ needs.setup.outputs.image_tag }} -r aimodelsregistry .
+
+  llama-2-13b:
+    needs: setup
+    runs-on: ubuntu-20.04
+    if: contains(github.event.commits[0].modified, 'pkg/presets/llama-2/')
+    steps:
+      - name: Build and push Llama model
+        run: |
+          az acr build --build-arg LLAMA_VERSION=llama-2-13b --build-arg SRC_DIR=pkg/presets/llama-2 -t aimodelsregistry.azurecr.io/llama-2-13b:${{ needs.setup.outputs.image_tag }} -r aimodelsregistry .
+
+  llama-2-70b:
+    needs: setup
+    runs-on: ubuntu-20.04
+    if: contains(github.event.commits[0].modified, 'pkg/presets/llama-2/')
+    steps:
+      - name: Build and push Llama model
+        run: |
+          az acr build --build-arg LLAMA_VERSION=llama-2-70b --build-arg SRC_DIR=pkg/presets/llama-2 -t aimodelsregistry.azurecr.io/llama-2-70b:${{ needs.setup.outputs.image_tag }} -r aimodelsregistry .
+
+  llama-2-7b-chat:
+    needs: setup
+    runs-on: ubuntu-20.04
+    if: contains(github.event.commits[0].modified, 'pkg/presets/llama-2-chat/')
+    steps:
+      - name: Build and push Llama chat model
+        run: |
+          az acr build --build-arg LLAMA_VERSION=llama-2-7b-chat --build-arg SRC_DIR=pkg/presets/llama-2-chat -t aimodelsregistry.azurecr.io/llama-2-7b-chat:${{ needs.setup.outputs.image_tag }} -r aimodelsregistry .
+
+  llama-2-13b-chat:
+    needs: setup
+    runs-on: ubuntu-20.04
+    if: contains(github.event.commits[0].modified, 'pkg/presets/llama-2-chat/')
+    steps:
+      - name: Build and push Llama chat model
+        run: |
+          az acr build --build-arg LLAMA_VERSION=llama-2-13b-chat --build-arg SRC_DIR=pkg/presets/llama-2-chat -t aimodelsregistry.azurecr.io/llama-2-13b-chat:${{ needs.setup.outputs.image_tag }} -r aimodelsregistry .
+
+  llama-2-70b-chat:
+    needs: setup
+    runs-on: ubuntu-20.04
+    if: contains(github.event.commits[0].modified, 'pkg/presets/llama-2-chat/')
+    steps:
+      - name: Build and push Llama chat model
+        run: |
+          az acr build --build-arg LLAMA_VERSION=llama-2-70b-chat --build-arg SRC_DIR=pkg/presets/llama-2-chat -t aimodelsregistry.azurecr.io/llama-2-70b-chat:${{ needs.setup.outputs.image_tag }} -r aimodelsregistry .
+

docker/presets/falcon/Dockerfile

@@ -2,7 +2,7 @@
 FROM nvcr.io/nvidia/pytorch:23.06-py3
 
 # Set the working directory
-WORKDIR /workspace/huggingface
+WORKDIR /workspace/falcon
 
 # First, copy just the requirements.txt file and install dependencies
 # This is done before copying the code to utilize Docker's layer caching and

docker/presets/llama-2/Dockerfile

@@ -1,3 +1,15 @@
+# Build text completion model
+# docker build \
+#   --build-arg LLAMA_VERSION=llama-2-7b \
+#   --build-arg SRC_DIR=pkg/presets/llama-2 \
+#   -t llama-2-7b:latest .
+
+# Build chat completion model
+# docker build \
+#   --build-arg LLAMA_VERSION=llama-2-7b-chat \
+#   --build-arg SRC_DIR=pkg/presets/llama-2-chat \
+#   -t llama-2-7b-chat:latest .
+
 FROM nvcr.io/nvidia/pytorch:23.06-py3
 WORKDIR /workspace
 
@@ -9,4 +21,25 @@ RUN pip install -e .
 RUN pip install fastapi pydantic
 RUN pip install 'uvicorn[standard]'
 
-ADD pkg/presets/llama-2 /workspace/llama/llama-2
+ARG LLAMA_VERSION
+ARG SRC_DIR
+
+# Conditional logic based on LLAMA_VERSION argument
+RUN if [ "$LLAMA_VERSION" = "llama-2-7b" ]; then \
+      wget -P ${SRC_DIR}/weights <URL_for_7b_chat_model>; \
+    elif [ "$LLAMA_VERSION" = "llama-2-13b" ]; then \
+      wget -P ${SRC_DIR}/weights <URL_for_13b_chat_model>; \
+    elif [ "$LLAMA_VERSION" = "llama-2-70b" ]; then \
+      wget -P ${SRC_DIR}/weights <URL_for_70b_chat_model>; \
+    elif [ "$LLAMA_VERSION" = "llama-2-7b-chat" ]; then \
+      wget -P ${SRC_DIR}/weights <URL_for_7b_model>; \
+    elif [ "$LLAMA_VERSION" = "llama-2-13b-chat" ]; then \
+      wget -P ${SRC_DIR}/weights <URL_for_13b_model>; \
+    elif [ "$LLAMA_VERSION" = "llama-2-70b-chat" ]; then \
+      wget -P ${SRC_DIR}/weights <URL_for_70b_model>; \
+    else \
+      echo "Invalid or missing LLAMA_VERSION"; \
+      exit 1; \
+    fi
+
+ADD ${SRC_DIR} /workspace/llama/llama-2