+
+{% endblock %}
+
+
+{% block footer %}
+ {{ super() }}
+{% endblock %}
\ No newline at end of file
diff --git a/docs/content/overview/.pages.yml b/docs/content/overview/.pages.yml
index f2902be4..ab4d78f8 100644
--- a/docs/content/overview/.pages.yml
+++ b/docs/content/overview/.pages.yml
@@ -1,8 +1,6 @@
---
nav:
- Overview: index.md
- - Mental Model: model
- - How it Works: how
- - Getting Started: quick-start
- - Security: security
- - Contributing: contributing
+ - Mental model: model
+ - How it works: how
+ - Security: security
\ No newline at end of file
diff --git a/docs/content/overview/index.md b/docs/content/overview/index.md
index 9048267e..04587f36 100644
--- a/docs/content/overview/index.md
+++ b/docs/content/overview/index.md
@@ -1,30 +1,26 @@
# Overview
--8<-- "wip.md"
---8<-- "news-feed.md"
---8<-- "disclaimer/third-party-model.md"
---8<-- "disclaimer/prompt-engineering-template.md"
---8<-- "aws-genai-llm-chatbot/mention.md"
-## Introduction
+Galileo Generative AI Reference Sample is a reference implementation of a deployable 3-tier retrieval augmented generative (RAG) application that aims to meet the needs of developers as they seek to rapidly experiment with, deploy, and launch GenAI powered products and services that utilise RAG. It is implemented in a number of languages, using AWS Cloud Development Kit ([CDK](https://aws.amazon.com/cdk/)) to define the infrastructure it deploys.
-Galileo is a reference implementation of a deployable 3-tier retrieval augmented generative (RAG) application that aims to meet the needs of developers as they seek to rapidly experiment with, deploy, and launch GenAI powered products and services that utilise RAG. It is implemented in a number of languages, using CDK to define the infrastructure it deploys. It can be optionally deployed with a sample dataset of US Supreme Court decision documents, providing content users can interact with immediately.
+(Optional) To provide your users interaction with sample data, you can deploy Galileo with a sample dataset of US Supreme Court decision documents.
## User Experience
-The screenshot below shows the default web interface, showing an active chat session, session management features (rename, delete, etc.), as well as the chat developer settings panel (collapsed).
+Once you deploy the reference sample (Galileo), the default web interface, displays an active chat session, session management features (rename, delete, etc.), as well as the chat developer settings panel (collapsed).
## Architecture
-This simplified architecture diagram shows the key service components utilised to implemented each of the three tiers described described in the [mental model](./model/index.md).
+This architecture diagram shows the key service components used to implement each of the three tiers described described in the [mental model](./model/index.md).
## Tooling
-**Meeting development teams where they are... between notebooks and application code.**
+**Meeting development teams where they are - between notebooks and application code.**
New LLMs are being released frequently, which is why Galileo focuses on model agnostic integration and supporting developers to try out the latest LLMs against the full application with zero code.
diff --git a/docs/content/overview/model/index.md b/docs/content/overview/model/index.md
index 364e80a1..2ccaf495 100644
--- a/docs/content/overview/model/index.md
+++ b/docs/content/overview/model/index.md
@@ -1,14 +1,14 @@
-# Mental Model
+# Mental model
-Taking the above into account, the mental model for this approach can be envisaged as a multi-tier architecture:
+The mental model for the Galileo Generative AI Reference Sample is designed as a multi-tier architecture:
-1. **Data Tier**: This tier is responsible for data ingestion, storage, and transformation (ETL), and provide API for querying data and performing operations against the data.
-2. **AI Tier**: This tier involves the core machine learning model development, training, tuning, and deployment (ML Ops), and all inference.
-3. **Presentation Tier**: This tier manages user interaction, result presentation, and integration with other systems. Users can input queries or requests and receive responses generated by the AI model. The presentation tier can consist of web interfaces, mobile apps, or any other user-facing interface. Client-side Chat Agent and similar prompt session management is core this to layer.
+1. **Data Tier**: This tier is responsible for data ingestion, storage, and transformation (ETL), and provide an API for querying data and performing operations against the data.
+2. **AI Tier**: This tier involves the core machine learning model development, training, tuning, and deployment (ML Ops), and all inferences.
+3. **Presentation Tier**: This tier manages user interaction, result presentation, and integration with other systems. Users can input queries or requests and receive responses generated by the AI model. The presentation tier can consist of web interfaces, mobile apps, or any other user-facing interface. The Client-side Chat Agent and similar prompt session management is core this to layer.
4. **Networking & Identity Layer**: Shared layer for networking and identity resources, such as VPC and Cognito.
-5. **Tooling Layer**: This supporting tier has the supporting tooling necessary for developers to build Generative AI applications. The tooling tier consists of supporting interfaces for developers (Notebooks, IDE tools, etc) and specific interfaces for generative ai (Prompt engineering etc).
+5. **Tooling Layer**: This tier has the supporting tooling necessary for developers to build Generative AI applications. The tooling tier consists of supporting interfaces for developers (Notebooks, IDE tools, etc) and specific interfaces for generative AI (Prompt engineering etc).
-The high-level components shown below illustrates the adapted three tier architecture we used to form the mental model when designing the architecture for this sample.
+The high-level components shown illustrates the adapted three-tier architecture we used to form the mental model when designing the architecture for this sample.
@@ -16,16 +16,16 @@ The high-level components shown below illustrates the adapted three tier archite
This tier is responsible for data ingestion, storage, and transformation (ETL), including the Vector Store for embeddings.
-Data Processing and Transformation: Depending on the application, there might be additional components for data processing and transformation. For example, in natural language processing tasks, text preprocessing, entity extraction, or semantic analysis modules may be required to enhance the AI processing and improve the accuracy of the generated responses.
+Data processing and transformation: Depending on the application, there might be additional components for data processing and transformation. For example, in natural language processing tasks, text preprocessing, entity extraction, or semantic analysis modules may be required to enhance the AI processing and improve the accuracy of the generated responses.
## AI Tier
This tier involves the core machine learning model development, training, tuning, and deployment stages, along with all inference related services, and a control plane to choreograph more complex use cases.
-This tier is where the generative AI model resides. It performs the main processing and generates responses based on the user inputs received from the presentation tier. The AI model can be a large language model, a stable diffusion image generation model, or a deep learning model specifically designed for a particular task, such as image generation or music composition. The AI processing tier involves complex algorithms,machine learning models, and compute infrastructure to perform the AI processing efficiently.
+This tier is where the generative AI model resides. It performs the main processing and generates responses based on the user inputs received from the presentation tier. The AI model can be a large language model, a stable diffusion image generation model, or a deep learning model specifically designed for a particular task, such as image generation or music composition. The AI processing tier involves complex algorithms, machine learning models, and compute infrastructure to perform the AI processing efficiently.
-Inference Engine: The inference engine is responsible for running the trained AI model to generate responses in real-time. It takes user inputs, processes them using the AI model, and produces the corresponding output. This component can be optimized for high-performance inference to handle concurrent user requests efficiently.
+Inference Engine: The inference engine is responsible for running the trained AI model to generate responses in real-time. It takes user inputs, processes them using the AI model, and produces the corresponding output. You can optimize this component for high-performance inference to handle concurrent user requests efficiently.
## Presentation Tier (Application)
-The presentation tier remains similar to the traditional three-tier architecture. It handles the user interface and interaction with the system. Users can input queries or requests and receive responses generated by the AI model. The presentation tier can consist of web interfaces, mobile apps, or any other user-facing interface.
+The presentation tier is similar to a traditional three-tier architecture. It handles the user interface and interaction with the system. Users can input queries or requests and receive responses generated by the AI model. The presentation tier can consist of web interfaces, mobile apps, or any other user-facing interface.
diff --git a/docs/content/troubleshooting/index.md b/docs/content/troubleshooting/index.md
new file mode 100644
index 00000000..1b99d292
--- /dev/null
+++ b/docs/content/troubleshooting/index.md
@@ -0,0 +1,31 @@
+# Troubleshooting
+
+Guide to common development concerns and issues.
+
+!!! info "Work in progress"
+
+ This page is work in progress. More content coming soon.
+
+---
+
+## Errors
+
+### `no space left on device` (build)
+
+This is caused by Docker being out of virtual space. The easiest fix is to run `docker system prune -f` and/or increasing the virtual disk size to free up space.
+
+### `error two`
+
+...fix
+
+---
+
+## FAQs
+
+### Question one?
+
+Answer one...
+
+### Question two?
+
+Answer two...
diff --git a/docs/content/user-guide/index.md b/docs/content/user-guide/index.md
new file mode 100644
index 00000000..d4815c34
--- /dev/null
+++ b/docs/content/user-guide/index.md
@@ -0,0 +1,5 @@
+# User guide
+
+!!! info "Work in progress"
+
+ This page is work in progress. More content coming soon.
diff --git a/docs/content/whats-new/index.md b/docs/content/whats-new/index.md
new file mode 100644
index 00000000..d1911904
--- /dev/null
+++ b/docs/content/whats-new/index.md
@@ -0,0 +1,6 @@
+# What's New
+
+--8<-- "news-feed.md"
+--8<-- "disclaimer/third-party-model.md"
+--8<-- "disclaimer/prompt-engineering-template.md"
+--8<-- "aws-genai-llm-chatbot/mention.md"
diff --git a/docs/mkdocs.yml b/docs/mkdocs.yml
index 2998e22f..bf833468 100644
--- a/docs/mkdocs.yml
+++ b/docs/mkdocs.yml
@@ -3,8 +3,8 @@ site_description: Galileo Generative AI Reference Sample
repo_name: aws-genai-conversational-rag-reference
repo_url: https://github.com/aws-samples/aws-genai-conversational-rag-reference/
site_url: https://aws-samples.github.io/aws-genai-conversational-rag-reference/
-copyright: Copyright © 2023 Amazon Web Services, Inc.
-use_directory_urls: true
+copyright: Copyright © 2024 Amazon Web Services, Inc.
+use_directory_urls: false
docs_dir: content
extra_css:
- assets/stylesheets/extra.css
@@ -26,9 +26,11 @@ theme:
logo: fontawesome/brands/aws
repo: fontawesome/brands/github
favicon: assets/images/favicon.png
+ custom_dir: content/overrides
palette:
scheme: default
features:
+ - content.code.copy
- toc.follow
- navigation.indexes
- navigation.instant
@@ -46,6 +48,7 @@ markdown_extensions:
- admonition
- tables
- footnotes
+ - md_in_html
- pymdownx.details
- pymdownx.highlight
- pymdownx.inlinehilite
@@ -53,6 +56,15 @@ markdown_extensions:
- pymdownx.caret
- pymdownx.keys
- pymdownx.mark
+ - pymdownx.snippets:
+ base_path: content/_
+ - pymdownx.tabbed:
+ alternate_style: true
+ - pymdownx.emoji:
+ emoji_index: !!python/name:materialx.emoji.twemoji
+ emoji_generator: !!python/name:materialx.emoji.to_svg
+ - pymdownx.arithmatex:
+ generic: true
- pymdownx.tilde
- toc:
toc_depth: 4
@@ -62,28 +74,11 @@ markdown_extensions:
provider: github
user: aws
repo: pdk
- - pymdownx.snippets:
- base_path:
- - !relative $config_dir
- - !relative $docs_dir
- - !relative $docs_dir/_
- auto_append:
- - _/abbreviations.md
- - pymdownx.tabbed:
- alternate_style: true
- - pymdownx.emoji:
- emoji_index: !!python/name:material.extensions.emoji.twemoji
- emoji_generator: !!python/name:material.extensions.emoji.to_svg
- - pymdownx.arithmatex:
- generic: true
plugins:
- awesome-pages:
filename: .pages.yml
- git-revision-date
- search
- - redirects:
- redirect_maps:
- "index.md": "overview/index.md"
- glightbox
extra_javascript: