Finetuning updates to the QnA RAG Demo

Makefile

@@ -0,0 +1,92 @@
+TOPDIR	  := $(shell git rev-parse --show-toplevel)
+BASE_VER  ?= v1.2.0
+BUILD_ID  ?= $(shell git describe --always --dirty)
+VERSION	  := $(BASE_VER)-$(BUILD_ID)
+
+
+export REPO    := ezkf
+REGISTRY       ?= lr1-bd-harbor-registry.mip.storage.hpecorp.net/develop
+EZKF_REGISTRY  ?= $(REGISTRY)/$(REPO)
+
+docker-build:
+	@echo "Building JupyterLab images..."
+	$(foreach target, base jupyter, \
+		docker build \
+		  --build-arg="BASE_IMG=$(EZKF_REGISTRY)/base:$(VERSION)" \
+			-t $(EZKF_REGISTRY)/$(target):$(VERSION) \
+			-f $(TOPDIR)/dockerfiles/notebooks/$(target)/Dockerfile \
+			$(TOPDIR)/dockerfiles/notebooks/$(target); \
+	)
+
+	@echo "Building the images for the Question-Answering demo..."
+	$(foreach target, app llm transformer vectorstore, \
+		docker build \
+			-t $(EZKF_REGISTRY)/qna-$(target):$(VERSION) \
+			-f $(TOPDIR)/demos/rag-demos/question-answering/dockerfiles/$(target)/Dockerfile \
+			$(TOPDIR)/demos/rag-demos/question-answering/dockerfiles/$(target); \
+	)
+
+	@echo "Building the images for the Question-Answering GPU demo..."
+	$(foreach target, app transformer vectorstore triton-inference-server, \
+		docker build \
+			-t $(EZKF_REGISTRY)/qna-$(target)-gpu:$(VERSION) \
+			-f $(TOPDIR)/demos/rag-demos/question-answering-gpu/dockerfiles/$(target)/Dockerfile \
+			$(TOPDIR)/demos/rag-demos/question-answering-gpu/dockerfiles/$(target); \
+	)
+
+	@echo "Building the images for the Fraud Detection demo..."
+	docker build \
+		-t $(EZKF_REGISTRY)/fraud-detection-app:$(VERSION) \
+		-f $(TOPDIR)/demos/fraud-detection/dockerfiles/app/Dockerfile \
+		$(TOPDIR)/demos/fraud-detection/dockerfiles/app
+
+docker-push:
+	@echo "Pushing JupyterLab images..."
+	$(foreach target, base jupyter, \
+		docker push $(EZKF_REGISTRY)/$(target):$(VERSION); \
+	)
+
+	@echo "Pushing the images for the Question-Answering demo..."
+	$(foreach target, app llm transformer vectorstore, \
+		docker push $(EZKF_REGISTRY)/qna-$(target):$(VERSION); \
+	)
+
+	@echo "Pushing the images for the Question-Answering GPU demo..."
+	$(foreach target, app transformer vectorstore triton-inference-server, \
+		docker push $(EZKF_REGISTRY)/qna-$(target)-gpu:$(VERSION); \
+	)
+
+	@echo "Pushing the images for the Fraud Detection demo..."
+	docker push $(EZKF_REGISTRY)/fraud-detection-app:$(VERSION)
+
+################################################################################
+# Pipeline API                                                                 #
+################################################################################
+
+version:
+	@echo $(VERSION)
+
+deliverables:
+	@echo
+
+dependencies:
+	@echo
+
+test:
+	@echo
+
+.PHONY: images
+images:
+	@echo $(EZKF_REGISTRY)/base:$(VERSION)
+	@echo $(EZKF_REGISTRY)/jupyter:$(VERSION)
+	@echo $(EZKF_REGISTRY)/qna-app:$(VERSION)
+	@echo $(EZKF_REGISTRY)/qna-llm:$(VERSION)
+	@echo $(EZKF_REGISTRY)/qna-transformer:$(VERSION)
+	@echo $(EZKF_REGISTRY)/qna-vectorstore:$(VERSION)
+	@echo $(EZKF_REGISTRY)/qna-app-gpu:$(VERSION)
+	@echo $(EZKF_REGISTRY)/qna-transformer-gpu:$(VERSION)
+	@echo $(EZKF_REGISTRY)/qna-vectorstore-gpu:$(VERSION)
+	@echo $(EZKF_REGISTRY)/qna-triton-inference-server-gpu:$(VERSION)
+	@echo $(EZKF_REGISTRY)/fraud-detection-app:$(VERSION)
+
+release: docker-build docker-push

README.md

@@ -13,6 +13,15 @@ capabilities.
 
 ![ezua-tutorials](images/ezua-tutorials.jpg)
 
+## Branching Model
+
+This repository adheres to the versioning scheme of the EzUA platform. The `develop` branch serves
+as the hub for active development. Release branches are derived from the develop branch, aligning
+closely with EzUA versions. For instance, `release-x.y.z` is designed to seamlessly woth with EzUA
+version `x.y.z`. It's worth noting that when you clone this repository, the default branch aligns
+with the most recent EzUA version. Therefore, if you're working with the latest EzUA iteration, you
+can clone it and proceed without the need to switch branches.
+
 ## Repository Structure
 
 This repository is organized into two main directories:
@@ -34,8 +43,7 @@ find specialized guides that show you how to leverage EzUA's frameworks and tool
 The current list of framework-specific tutorials include:
 
 - [FEAST feature store](tutorials/feast/): Ride sharing tutorial
-- [Kubeflow Pipelines](tutorials/kubeflow-pipelines/): Financial time series tutorial
-- [RAY](tutorials/ray): News recommendation tutorial
+- [Superset](tutorials/superset/): Data connection and visualization
 
 ## Getting Started
 

demos/bike-sharing/README.md

@@ -32,10 +32,11 @@ For this tutorial, ensure you have:
 To complete the tutorial follow the steps below:
 
 1. Login to your EzUA cluster, using your credentials.
-1. Create a new Notebook server using the `jupyter-data-science` image. Request at least 4Gi of memory for the Notebook
-   server.
+1. Create a new Notebook server using the `jupyter-data-science` image. Request at least 4Gi of
+   memory for the Notebook server.
 1. Connect to the Notebook server, launch a new terminal window, and clone the repository locally.
-1. Navigate to the tutorial's directory (`ezua-tutorials/tutorials/mlflow`)
+   See the troubleshooting section if this step fails.
+1. Navigate to the tutorial's directory (`ezua-tutorials/demos/bike-sharing`)
 1. Create your virtual environment:
     - Deactivate the base conda environment:
         ```
@@ -65,6 +66,15 @@ To complete the tutorial follow the steps below:
 1. Launch the two Notebooks in order and execute the code cells. Make sure to select the `bike-sharing` environment
    kernel for each Notebook.
 
+## Troubleshooting
+
+If you are behind a proxy you will have to set a few environment variables to be able to clone the
+`ezua-tutorials` repository locally and install the dependencies via `pip`. To this end, launch a
+terminal window and before cloning the repository, run the following commands:
+
+- `export http_proxy=<your http proxy URL>`
+- `export https_proxy=<your https proxy URL>`
+
 ## How it Works
 
 MLflow is an open-source platform designed to manage the end-to-end machine learning lifecycle. It encompasses tools for

demos/bike-sharing/environment.yaml

@@ -1,7 +1,6 @@
 name: bike-sharing
 channels:
   - conda-forge
-  - defaults
 dependencies:
   - python=3.8
   - pip

demos/fraud-detection/application/helm/Chart.yaml

@@ -1,6 +1,6 @@
 name: fraud-detection-app
-version: 0.1.0
+version: 0.2.1
 apiVersion: v2
-appVersion: v0.1.0
+appVersion: v0.2.1
 description: Install the Fraud Detection application on EzAF
 type: application

demos/fraud-detection/application/helm/values.yaml

@@ -8,7 +8,7 @@ image:
   repository: dpoulopoulos/fraud-detection-app
   pullPolicy: Always
   # Overrides the image tag whose default is the chart appVersion.
-  tag: "v0.1.0"
+  tag: "v0.2.1"
 
 imagePullSecrets: []
 nameOverride: ""
@@ -75,10 +75,10 @@ ingress:
 resources:
   limits:
     cpu: 100m
-    memory: 128Mi
+    memory: 1Gi
   requests:
     cpu: 100m
-    memory: 128Mi
+    memory: 1Gi
 
 autoscaling:
   enabled: false