0.2.0 fix (part 1)

.github/ISSUE_TEMPLATE/bug.yml

@@ -23,6 +23,22 @@ body:
         - label: Other issue
     validations:
       required: true
+  - type: checkboxes
+    attributes:
+      label: OS/Platform(s) Used
+      description: \
+        What OS or platform(s) did you encounter this issue? Specify
+        exact version or distro in the Problem Brief if needed.
+      options:
+        - label: Windows (local)
+        - label: MacOS Intel (local)
+        - label: MacOS Apple Silicon (local)
+        - label: Linux (local)
+        - label: Coder
+        - label: Run:ai VSCode
+        - label: Run:ai JupyterLab
+    validations:
+      required: true
   - type: textarea
     attributes:
       label: Problem Brief

{{cookiecutter.repo_name}}/.gitlab-ci.yml

@@ -1,6 +1,5 @@
 default:
   tags:
-    - dind
     - on-prem
 
 variables:
@@ -40,6 +39,8 @@ build:data-prep-image:
   image:
     name: gcr.io/kaniko-project/executor:debug
     entrypoint: [""]
+  before_script:
+    - "[[ -z ${HARBOR_ROBOT_CREDS_JSON} ]] && exit 1"
   script:
     - mkdir -p /kaniko/.docker
     - cat $HARBOR_ROBOT_CREDS_JSON > /kaniko/.docker/config.json
@@ -63,6 +64,8 @@ build:model-training-image:
   image:
     name: gcr.io/kaniko-project/executor:debug
     entrypoint: [""]
+  before_script:
+    - "[[ -z ${HARBOR_ROBOT_CREDS_JSON} ]] && exit 1"
   script:
     - mkdir -p /kaniko/.docker
     - cat $HARBOR_ROBOT_CREDS_JSON > /kaniko/.docker/config.json
@@ -86,6 +89,8 @@ build:retag-images:
   image:
     name: gcr.io/go-containerregistry/crane:debug
     entrypoint: [""]
+  before_script:
+    - "[[ -z ${HARBOR_ROBOT_CREDS_JSON} ]] && exit 1"
   script:
     - cat $HARBOR_ROBOT_CREDS_JSON > /root/.docker/config.json
     - crane tag {{cookiecutter.registry_project_path}}/data-prep:${CI_COMMIT_SHORT_SHA} ${CI_COMMIT_TAG}

{{cookiecutter.repo_name}}/aisg-context/guide-site/docs/guide-for-user/03-mlops-components-platform.md

@@ -833,7 +833,7 @@ MLflow Tracking server.
 === "Linux/macOS"
 
     ```bash
-    $ conda create -n mlflow-test python=3.10.11
+    $ conda create -n mlflow-test python=3.11.7
     $ conda activate mlflow-test
     $ pip install boto3==1.34.17 mlflow==2.9.2
     $ export MLFLOW_TRACKING_USERNAME=<MLFLOW_TRACKING_USERNAME>
@@ -844,7 +844,7 @@ MLflow Tracking server.
 === "Windows PowerShell"
 
     ```powershell
-    $ conda create -n mlflow-test python=3.10.11
+    $ conda create -n mlflow-test python=3.11.7
     $ conda activate mlflow-test
     $ pip install boto3==1.34.17 mlflow==2.9.2
     $ $MLFLOW_TRACKING_USERNAME=<MLFLOW_TRACKING_USERNAME>

{{cookiecutter.repo_name}}/aisg-context/guide-site/docs/guide-for-user/04-dev-wksp.md

@@ -222,7 +222,7 @@ Now, let's clone your repository from the remote:
 === "VSCode Server Terminal"
 
     ```bash
-    $ cd /{{cookiecutter.proj_name}}/workspaces/<YOUR_HYPHENATED_NAME>
+    $ cd /<PVC_LOCATION>/workspaces/<YOUR_HYPHENATED_NAME>
     $ git clone <REMOTE_URL_HTTPS>
     $ cd {{cookiecutter.repo_name}}
     ```

{{cookiecutter.repo_name}}/aisg-context/guide-site/docs/guide-for-user/05-virtual-env.md

@@ -35,7 +35,7 @@ workspace directory:
     (base) $ echo 'alias {{cookiecutter.repo_name}}-conda="conda activate /<NAME_OF_DATA_SOURCE>/workspaces/<YOUR_HYPHENATED_NAME>/conda_envs/{{cookiecutter.repo_name}}"' >> ~/.bashrc
     (base) $ source ~/.bashrc
     (base) $ {{cookiecutter.repo_name}}-conda
-    ({{cookiecutter.repo_name}}-conda) $ # conda environment has been activated
+    ({{cookiecutter.repo_name}}) $ # conda environment has been activated
     ```
 
 !!! tip

{{cookiecutter.repo_name}}/aisg-context/guide-site/docs/guide-for-user/08-deployment.md

@@ -32,20 +32,22 @@ autolog). With that, we have the following pointers to take note of:
   convention similar to the following:
   `<MLFLOW_EXPERIMENT_ARTIFACT_LOCATION>/<MLFLOW_RUN_UUID>/artifacts`.
 - With this path/URI, we can use the AWS CLI to download the predictive
-  model from ECS into a mounted volume when we run the Docker image for
+  model from ECS. Alternatively, we can utilise the MLFlow Client
+  library to retrieve the predictive model. This model can then be
+  propagated into a mounted volume when we run the Docker image for
   the REST APIs.
 
-Here's how you can quickly retrieve the artifact location of a
-specific MLflow experiment within your VSCode server:
+Here's how you can quickly retrieve the artifact of a specific MLflow
+Run within your VSCode server with the MLFlow Client library:
 
 === "Linux/macOS"
 
     ```bash
     $ export MLFLOW_TRACKING_URI=<MLFLOW_TRACKING_URI>
     $ export MLFLOW_TRACKING_USERNAME=<MLFLOW_TRACKING_USERNAME>
     $ export MLFLOW_TRACKING_PASSWORD=<MLFLOW_TRACKING_PASSWORD>
-    $  python -c "import mlflow; mlflow_experiment = mlflow.get_experiment_by_name('<NAME_OF_DEFAULT_MLFLOW_EXPERIMENT>'); print(mlflow_experiment.artifact_location)"
-    s3://<BUCKET_NAME>/subdir/paths
+    $ python -c "import mlflow; mlflow.artifacts.download_artifact(artifact_uri='runs:/<MLFLOW_RUN_UUID>/', dst_path='models/<MLFLOW_RUN_UUID')"
+    Downloading artifacts: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:01<00:00,  1.18it/s]
     ```
 
 === "Windows PowerShell"
@@ -54,57 +56,12 @@ specific MLflow experiment within your VSCode server:
     $ $Env:MLFLOW_TRACKING_URI=<MLFLOW_TRACKING_URI>
     $ $Env:MLFLOW_TRACKING_USERNAME=<MLFLOW_TRACKING_USERNAME>
     $ $Env:MLFLOW_TRACKING_PASSWORD=<MLFLOW_TRACKING_PASSWORD>
-    $  python -c "import mlflow; mlflow_experiment = mlflow.get_experiment_by_name('<NAME_OF_DEFAULT_MLFLOW_EXPERIMENT>'); print(mlflow_experiment.artifact_location)"
-    s3://<BUCKET_NAME>/subdir/paths
+    $ python -c "import mlflow; mlflow.artifacts.download_artifact(artifact_uri='runs:/<MLFLOW_RUN_UUID>/', dst_path='models/<MLFLOW_RUN_UUID')"
+    Downloading artifacts: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:01<00:00,  1.18it/s]
     ```
 
-To list the contents of the artifact location, you can use the AWS CLI
-(installed within the VSCOde server by default) like so:
 
-=== "Linux/macOS"
-
-    ```bash
-    $ export AWS_ACCESS_KEY_ID=<AWS_ACCESS_KEY_ID>
-    $ export AWS_SECRET_ACCESS_KEY=<AWS_SECRET_ACCESS_KEY>
-    $ aws s3 ls --endpoint-url "https://necs.nus.edu.sg" <MLFLOW_EXPERIMENT_ARTIFACT_LOCATION>/
-                           PRE XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX/
-                           PRE YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY/
-                           PRE ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ/
-    ```
-
-=== "Windows PowerShell"
-
-    ```powershell
-    $ $Env:AWS_ACCESS_KEY_ID=<AWS_ACCESS_KEY_ID>
-    $ $Env:AWS_SECRET_ACCESS_KEY=<AWS_SECRET_ACCESS_KEY>
-    $ aws s3 ls --endpoint-url "https://necs.nus.edu.sg" <MLFLOW_EXPERIMENT_ARTIFACT_LOCATION>/
-                           PRE XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX/
-                           PRE YYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY/
-                           PRE ZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZZ/
-    ```
-
-What would be listed are subdirectories named after MLflow experiment
-unique IDs. Within each of these subdirectories, we will find the
-artifacts uploaded to ECS.To list the artifacts of a specific run,
-we can run a command like the following:
-
-=== "Linux/macOS"
-
-    ```bash
-    $ aws s3 ls --recursive --endpoint-url "https://necs.nus.edu.sg" <MLFLOW_EXPERIMENT_ARTIFACT_LOCATION>/<MLFLOW_RUN_UUID>
-    YYYY-MM-DD hh:mm:ss   XXXXXXXX <MLFLOW_EXPERIMENT_ARTIFACT_LOCATION>/<MLFLOW_RUN_UUID>/artifacts/model/model.pt
-    YYYY-MM-DD hh:mm:ss   XXXXXXXX <MLFLOW_EXPERIMENT_ARTIFACT_LOCATION>/<MLFLOW_RUN_UUID>/artifacts/train_model_config.json
-    ```
-
-=== "Windows PowerShell"
-
-    ```powershell
-    $ aws s3 ls --recursive --endpoint-url "https://necs.nus.edu.sg" <MLFLOW_EXPERIMENT_ARTIFACT_LOCATION>/<MLFLOW_RUN_UUID>
-    YYYY-MM-DD hh:mm:ss   XXXXXXXX <MLFLOW_EXPERIMENT_ARTIFACT_LOCATION>/<MLFLOW_RUN_UUID>/artifacts/model/model.pt
-    YYYY-MM-DD hh:mm:ss   XXXXXXXX <MLFLOW_EXPERIMENT_ARTIFACT_LOCATION>/<MLFLOW_RUN_UUID>/artifacts/train_model_config.json
-    ```
-
-Now that we have established on how we are to obtain the models for the
+Now that we have established on how we are to obtain a specific model for the
 API server, let's look into the servers themselves.
 
 ## Model Serving (FastAPI)
@@ -116,7 +73,7 @@ among other things. These factors have made it a popular framework
 within AI Singapore across many projects.
 
 If you were to inspect the `src` folder, you would notice that there
-exist more than one package:
+exists more than one package:
 
 - `{{cookiecutter.src_package_name}}`
 - `{{cookiecutter.src_package_name}}_fastapi`
@@ -150,16 +107,20 @@ following commands:
 
     ```bash
     $ export PRED_MODEL_UUID=<MLFLOW_RUN_UUID>
-    $ export PRED_MODEL_ECS_S3_URI=<MLFLOW_EXPERIMENT_ARTIFACT_LOCATION>/$PRED_MODEL_UUID
-    $ aws s3 cp --recursive --endpoint-url "https://necs.nus.edu.sg" $PRED_MODEL_ECS_S3_URI ./models/$PRED_MODEL_UUID
+    $ export MLFLOW_TRACKING_URI=<MLFLOW_TRACKING_URI>
+    $ export MLFLOW_TRACKING_USERNAME=<MLFLOW_TRACKING_USERNAME>
+    $ export MLFLOW_TRACKING_PASSWORD=<MLFLOW_TRACKING_PASSWORD>
+    $ python -c "import mlflow; mlflow.artifacts.download_artifact(artifact_uri='runs:/$PRED_MODEL_UUID/', dst_path='models/$PRED_MODEL_UUID')"
     ```
 
 === "Windows PowerShell"
 
     ```powershell
     $ $Env:PRED_MODEL_UUID=<MLFLOW_RUN_UUID>
-    $ $Env:PRED_MODEL_ECS_S3_URI=<MLFLOW_EXPERIMENT_ARTIFACT_LOCATION>/$Env:MLFLOW_RUN_UUID
-    $ aws s3 cp --recursive --endpoint-url "https://necs.nus.edu.sg" $Env:PRED_MODEL_ECS_S3_URI .\models\$Env:PRED_MODEL_UUID
+    $ $Env:MLFLOW_TRACKING_URI=<MLFLOW_TRACKING_URI>
+    $ $Env:MLFLOW_TRACKING_USERNAME=<MLFLOW_TRACKING_USERNAME>
+    $ $Env:MLFLOW_TRACKING_PASSWORD=<MLFLOW_TRACKING_PASSWORD>
+    $ python -c "import mlflow; mlflow.artifacts.download_artifact(artifact_uri='runs:/$PRED_MODEL_UUID/', dst_path='models/$PRED_MODEL_UUID')"
     ```
 
 Executing the commands above will download the artifacts related to the
@@ -203,7 +164,7 @@ Run the FastAPI server using [Gunicorn](https://gunicorn.org)
     ```bash
     $ conda activate {{cookiecutter.repo_name}}
     $ cd src
-    $ gunicorn {{cookiecutter.src_package_name}}_fastapi.main:APP -b 0.0.0.0:8080 -w 4 -k uvicorn.workers.UvicornWorker
+    $ gunicorn {{cookiecutter.src_package_name}}_fastapi.main:APP -b 0.0.0.0:8080 -w 2 -k uvicorn.workers.UvicornWorker -t 90
     ```
 
     See
@@ -264,7 +225,7 @@ the environment variables.
 `src/{{cookiecutter.src_package_name}}_fastapi/config.py`:
 ```python
 ...
-class Settings(pydantic.BaseSettings):
+class Settings(pydantic_settings.BaseSettings):
 
     API_NAME: str = "{{cookiecutter.src_package_name}}_fastapi"
     API_V1_STR: str = "/api/v1"
@@ -280,8 +241,29 @@ class Settings(pydantic.BaseSettings):
 FastAPI automatically generates interactive API documentation for
 easy viewing of all the routers/endpoints you have made available for
 the server. You can view the documentation through
-`<API_SERVER_URL>:<PORT>/docs`. In our case here, it is viewable through
-[`localhost:8080/docs`](http://localhost:8080/docs). It will look like
+`<API_SERVER_URL>:<PORT>/docs`. 
+
+It's optional, but let's view the labour of our hard work:
+
+> The following steps assumes you have installed coder on your machine.
+> Else, follow the installation steps [here](https://coder.com/docs/v2/latest/install#install-coder)
+> and login via `coder login <CODER_URL>`
+
+=== "Linux/macOS"
+
+    ```bash
+    $ coder port-forward <WORKSPACE_NAME> --tcp 8080:8080
+    ```
+
+=== "Windows PowerShell"
+
+    ```powershell
+
+    $ coder port-forward <WORKSPACE_NAME> --tcp 8080:8080
+    ```
+
+And with that, our document site for our server is viewable through
+[`localhost:8080/docs`](http://localhost:8080/docs) and will look as
 such:
 
 ![FastAPI - OpenAPI Docs](assets/screenshots/fastapi-openapi-docs.png)

{{cookiecutter.repo_name}}/aisg-context/guide-site/docs/guide-for-user/10-cicd.md

@@ -171,6 +171,7 @@ Let's look at the job defined for the `test`stage first:
       before_script:
         - conda env create -f {{cookiecutter.repo_name}}-conda-env.yaml
         - source activate {{cookiecutter.repo_name}}
+        - pip install -r dev-requirements.txt
       script:
         - pylint src --fail-under=7.0 --ignore=tests --disable=W1202
         - pytest src/tests

{{cookiecutter.repo_name}}/{{cookiecutter.repo_name}}-conda-env.yaml

@@ -4,43 +4,23 @@ channels:
   - pytorch
   - conda-forge
 dependencies:
+  - torch=2.1.2
+  - torchvision=0.16.2
+  - cpuonly=2.0
   - python=3.11.7
   - pip=23.3.2
   - pip:
-    - torch==2.1.2
-    - mlflow==2.9.2
-    - boto3==1.34.17
+    - mlflow-skinny==2.9.2
     - hydra-core==1.3.2
     - hydra-optuna-sweeper==1.2.0
-    - torchvision==0.16.2
     - python-json-logger==2.0.7
     - fastapi==0.109.0
     - uvicorn[standard]==0.25.0
     - python-multipart==0.0.6
-    #- python-json-logger==2.0.7
-    #- pytest==7.4.0
-    #- pandas==2.0.3
-    #- torch==2.0.1
-    #- torchdata==0.6.1
-    #- torchvision==0.15.2
-    #- hydra-core==1.3.2
-    #- hydra-optuna-sweeper==1.2.0
-    #- pre-commit==3.3.3
-    #- black==23.3.0
-    #- scipy==1.11.1
-    #- Pillow==10.0.0
-    #- mlflow==2.6.0
-    #- boto3==1.28.2
-    #- scikit-learn==1.3.0
-    #- pylint==2.17.4
-    #- pydocstyle==6.3.0
-    #- pygments==2.15.1
-    #- semgrep==1.33.2
-    #- ipykernel==6.25.1
-    #- fastapi==0.103.0
-    #- uvicorn[standard]==0.23.2
-    #- gunicorn==21.2.0
-    #- pydantic==1.10.12
-    #- python-multipart==0.0.6
-    #- jsonlines==3.1.0
-    #- sphinx==4.3.2
+    - jsonlines==4.0.0
+    - pandas==2.1.4
+    - gunicorn==21.2.0
+    - pydantic==2.5.3
+    - pydantic-settings==2.1.0
+    - ipykernel==6.25.0
+    - sphinx==7.2.5