alan-turing-institute · rchan26 · Jun 13, 2024 · Jun 12, 2024 · Jun 12, 2024 · Jun 12, 2024
diff --git a/.github/workflows/build_docker_image.yaml b/.github/workflows/build_docker_image.yaml
@@ -20,35 +20,37 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - dockerfile: docker/reginald/Dockerfile
-            image: ghcr.io/${{ github.repository }}_reginald
+          - dockerfile: docker/run_all/Dockerfile
+            image: ghcr.io/${{ github.repository }}_run_all
           - dockerfile: docker/slack_bot/Dockerfile
             image: ghcr.io/${{ github.repository }}_slackbot
           - dockerfile: docker/create_index/Dockerfile
             image: ghcr.io/${{ github.repository }}_create_index
+          - dockerfile: docker/reginald_app/Dockerfile
+            image: ghcr.io/${{ github.repository }}_app
     permissions:
       packages: write
       contents: read
 
     steps:
       - name: Check out the repo
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
 
       - name: Log in to the Container registry
-        uses: docker/login-action@v2
+        uses: docker/login-action@v3.2.0
         with:
           registry: ghcr.io
           username: ${{ github.actor }}
           password: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Extract metadata (tags, labels) for Docker
         id: meta
-        uses: docker/metadata-action@v4
+        uses: docker/metadata-action@v5.5.0
         with:
           images: ${{ matrix.image }}
 
       - name: Build and push Docker images
-        uses: docker/build-push-action@v4
+        uses: docker/build-push-action@v5.4.0
         with:
           file: ${{ matrix.dockerfile }}
           push: true

diff --git a/.github/workflows/lint_code.yaml b/.github/workflows/lint_code.yaml
@@ -13,7 +13,7 @@ jobs:
     name: Lint with pre-commit
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: actions/setup-python@v4
         with:
           python-version: 3.11

diff --git a/azure/hack_week/__main__.py b/azure/hack_week/__main__.py
@@ -7,6 +7,8 @@
     storage,
 )
 
+from reginald.defaults import DEFAULT_ARGS
+
 # Get some configuration variables
 stack_name = pulumi.get_stack()
 config = pulumi.Config()
@@ -98,7 +100,7 @@
     containers=[
         # Reginald chat completion container
         containerinstance.ContainerArgs(
-            image="ghcr.io/alan-turing-institute/reginald_reginald:main",
+            image="ghcr.io/alan-turing-institute/reginald_run_all:main",
             name="reginald-completion",  # maximum of 63 characters
             environment_variables=[
                 containerinstance.EnvironmentVariableArgs(
@@ -141,7 +143,7 @@
         ),
         # Reginald (public) container
         containerinstance.ContainerArgs(
-            image="ghcr.io/alan-turing-institute/reginald_reginald:main",
+            image="ghcr.io/alan-turing-institute/reginald_run_all:main",
             name="reginald-gpt-azure",  # maximum of 63 characters
             environment_variables=[
                 containerinstance.EnvironmentVariableArgs(
@@ -166,23 +168,23 @@
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_MAX_INPUT_SIZE",
-                    value="4096",
+                    value=str(DEFAULT_ARGS["max_input_size"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_K",
-                    value="3",
+                    value=str(DEFAULT_ARGS["k"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_CHUNK_SIZE",
-                    value="512",
+                    value=str(DEFAULT_ARGS["chunk_size"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_CHUNK_OVERLAP_RATIO",
-                    value="0.1",
+                    value=str(DEFAULT_ARGS["chunk_overlap_ratio"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_NUM_OUTPUT",
-                    value="512",
+                    value=str(DEFAULT_ARGS["num_output"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="OPENAI_AZURE_API_BASE",
@@ -204,8 +206,8 @@
             ports=[],
             resources=containerinstance.ResourceRequirementsArgs(
                 requests=containerinstance.ResourceRequestsArgs(
-                    cpu=1,
-                    memory_in_gb=12,
+                    cpu=3,
+                    memory_in_gb=14,
                 ),
             ),
             volume_mounts=[
@@ -253,30 +255,30 @@
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_MAX_INPUT_SIZE",
-                    value="4096",
+                    value=str(DEFAULT_ARGS["max_input_size"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_K",
-                    value="3",
+                    value=str(DEFAULT_ARGS["k"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_CHUNK_SIZE",
-                    value="512",
+                    value=str(DEFAULT_ARGS["chunk_size"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_CHUNK_OVERLAP_RATIO",
-                    value="0.1",
+                    value=str(DEFAULT_ARGS["chunk_overlap_ratio"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_NUM_OUTPUT",
-                    value="512",
+                    value=str(DEFAULT_ARGS["num_output"]),
                 ),
             ],
             ports=[],
             resources=containerinstance.ResourceRequirementsArgs(
                 requests=containerinstance.ResourceRequestsArgs(
-                    cpu=4,
-                    memory_in_gb=16,
+                    cpu=1,
+                    memory_in_gb=8,
                 ),
             ),
             volume_mounts=[

diff --git a/azure/production/__main__.py b/azure/production/__main__.py
@@ -7,6 +7,8 @@
     storage,
 )
 
+from reginald.defaults import DEFAULT_ARGS
+
 # Get some configuration variables
 stack_name = pulumi.get_stack()
 config = pulumi.Config()
@@ -142,7 +144,7 @@
         # all_data index creation container
         containerinstance.ContainerArgs(
             image="ghcr.io/alan-turing-institute/reginald_create_index:main",
-            name="reginald-create-index",  # maximum of 63 characters
+            name="reginald-create-index-all-data",  # maximum of 63 characters
             environment_variables=[
                 containerinstance.EnvironmentVariableArgs(
                     name="GITHUB_TOKEN",
@@ -154,30 +156,126 @@
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_MAX_INPUT_SIZE",
-                    value="4096",
+                    value=str(DEFAULT_ARGS["max_input_size"]),
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_K",
+                    value=str(DEFAULT_ARGS["k"]),
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_CHUNK_SIZE",
+                    value=str(DEFAULT_ARGS["chunk_size"]),
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_CHUNK_OVERLAP_RATIO",
+                    value=str(DEFAULT_ARGS["chunk_overlap_ratio"]),
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_NUM_OUTPUT",
+                    value=str(DEFAULT_ARGS["num_output"]),
+                ),
+            ],
+            ports=[],
+            resources=containerinstance.ResourceRequirementsArgs(
+                requests=containerinstance.ResourceRequestsArgs(
+                    cpu=2,
+                    memory_in_gb=6,
+                ),
+            ),
+            volume_mounts=[
+                containerinstance.VolumeMountArgs(
+                    mount_path="/app/data",
+                    name="llama-data",
+                ),
+            ],
+        ),
+        # reg index creation container
+        containerinstance.ContainerArgs(
+            image="ghcr.io/alan-turing-institute/reginald_create_index:main",
+            name="reginald-create-index-reg",  # maximum of 63 characters
+            environment_variables=[
+                containerinstance.EnvironmentVariableArgs(
+                    name="GITHUB_TOKEN",
+                    secure_value=config.get_secret("GITHUB_TOKEN"),
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_WHICH_INDEX",
+                    value="reg",
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_MAX_INPUT_SIZE",
+                    value=str(DEFAULT_ARGS["max_input_size"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_K",
-                    value="3",
+                    value=str(DEFAULT_ARGS["k"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_CHUNK_SIZE",
-                    value="512",
+                    value=str(DEFAULT_ARGS["chunk_size"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_CHUNK_OVERLAP_RATIO",
-                    value="0.1",
+                    value=str(DEFAULT_ARGS["chunk_overlap_ratio"]),
                 ),
                 containerinstance.EnvironmentVariableArgs(
                     name="LLAMA_INDEX_NUM_OUTPUT",
-                    value="512",
+                    value=str(DEFAULT_ARGS["num_output"]),
                 ),
             ],
             ports=[],
             resources=containerinstance.ResourceRequirementsArgs(
                 requests=containerinstance.ResourceRequestsArgs(
-                    cpu=4,
-                    memory_in_gb=16,
+                    cpu=1,
+                    memory_in_gb=5,
+                ),
+            ),
+            volume_mounts=[
+                containerinstance.VolumeMountArgs(
+                    mount_path="/app/data",
+                    name="llama-data",
+                ),
+            ],
+        ),
+        # public index creation container
+        containerinstance.ContainerArgs(
+            image="ghcr.io/alan-turing-institute/reginald_create_index:main",
+            name="reginald-create-index-public",  # maximum of 63 characters
+            environment_variables=[
+                containerinstance.EnvironmentVariableArgs(
+                    name="GITHUB_TOKEN",
+                    secure_value=config.get_secret("GITHUB_TOKEN"),
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_WHICH_INDEX",
+                    value="public",
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_MAX_INPUT_SIZE",
+                    value=str(DEFAULT_ARGS["max_input_size"]),
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_K",
+                    value=str(DEFAULT_ARGS["k"]),
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_CHUNK_SIZE",
+                    value=str(DEFAULT_ARGS["chunk_size"]),
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_CHUNK_OVERLAP_RATIO",
+                    value=str(DEFAULT_ARGS["chunk_overlap_ratio"]),
+                ),
+                containerinstance.EnvironmentVariableArgs(
+                    name="LLAMA_INDEX_NUM_OUTPUT",
+                    value=str(DEFAULT_ARGS["num_output"]),
+                ),
+            ],
+            ports=[],
+            resources=containerinstance.ResourceRequirementsArgs(
+                requests=containerinstance.ResourceRequestsArgs(
+                    cpu=1,
+                    memory_in_gb=5,
                 ),
             ),
             volume_mounts=[
@@ -190,7 +288,7 @@
     ],
     os_type=containerinstance.OperatingSystemTypes.LINUX,
     resource_group_name=resource_group.name,
-    restart_policy=containerinstance.ContainerGroupRestartPolicy.NEVER,
+    restart_policy=containerinstance.ContainerGroupRestartPolicy.ON_FAILURE,
     sku=containerinstance.ContainerGroupSku.STANDARD,
     volumes=[
         containerinstance.VolumeArgs(

diff --git a/docker/create_index/Dockerfile b/docker/create_index/Dockerfile
@@ -1,14 +1,22 @@
-FROM python:3.11.4
+FROM python:3.11.9
 
 WORKDIR /app
 
 # Setup Python prerequisites
-RUN pip3 install --upgrade pip poetry setuptools wheel
+RUN pip3 install --upgrade pip poetry==1.8.3 setuptools wheel
+
+ENV POETRY_NO_INTERACTION=1 \
+    POETRY_VIRTUALENVS_IN_PROJECT=1 \
+    POETRY_VIRTUALENVS_CREATE=1 \
+    POETRY_CACHE_DIR=/tmp/poetry_cache
 
 # Build Python project
-COPY reginald reginald
 COPY pyproject.toml .
-COPY README.md .
-RUN poetry install
+RUN touch README.md
+RUN poetry install --without dev --no-root  && rm -rf $POETRY_CACHE_DIR
+
+COPY reginald reginald
+
+RUN poetry install --without dev
 
-CMD ["poetry", "run", "reginald_create_index"]
+CMD ["poetry", "run", "reginald", "create-index"]
diff --git a/docker/reginald/Dockerfile b/docker/reginald/Dockerfile
diff --git a/docker/reginald_app/Dockerfile b/docker/reginald_app/Dockerfile
@@ -0,0 +1,22 @@
+FROM python:3.11.9
+
+WORKDIR /app
+
+# Setup Python prerequisites
+RUN pip3 install --upgrade pip poetry setuptools wheel
+
+ENV POETRY_NO_INTERACTION=1 \
+    POETRY_VIRTUALENVS_IN_PROJECT=1 \
+    POETRY_VIRTUALENVS_CREATE=1 \
+    POETRY_CACHE_DIR=/tmp/poetry_cache
+
+# Build Python project
+COPY pyproject.toml .
+RUN touch README.md
+RUN poetry install --without dev --no-root  && rm -rf $POETRY_CACHE_DIR
+
+COPY reginald reginald
+
+RUN poetry install --extras api_bot --without dev
+
+CMD ["poetry", "run", "reginald", "app"]