From 9f1408516226364e361b007b7a82adc077b7144a Mon Sep 17 00:00:00 2001
From: "Nat Kershaw (MSFT)" <nakersha@microsoft.com>
Date: Tue, 13 Aug 2024 21:26:49 -0700
Subject: [PATCH] Update install instructions for onnxruntime-genai separate
 packages (#21646)

Staged:
https://natke.github.io/onnxruntime/docs/genai/howto/install.html
---
 docs/genai/howto/install.md | 54 ++++++++++++++++++++++---------------
 1 file changed, 32 insertions(+), 22 deletions(-)

diff --git a/docs/genai/howto/install.md b/docs/genai/howto/install.md
index c77a9e089cc0b..86f969c8ccf32 100644
--- a/docs/genai/howto/install.md
+++ b/docs/genai/howto/install.md
@@ -13,69 +13,79 @@ nav_order: 1
 * TOC placeholder
 {:toc}
 
-## Pre-requisites
 
-### CUDA
-
-If you are installing the CUDA variant of onnxruntime-genai, the CUDA toolkit must be installed.
-
-The CUDA toolkit can be downloaded from the [CUDA Toolkit Archive](https://developer.nvidia.com/cuda-toolkit-archive).
-
-Ensure that the `CUDA_PATH` environment variable is set to the location of your CUDA installation.
+## Python package installation
 
-## Python packages
-
-Note: only one of these packages should be installed in your application.
+Note: only one of these sets of packages (CPU, DirectML, CUDA) should be installed in your environment.
 
 ### CPU
 
 ```bash
 pip install numpy
-pip install onnxruntime-genai --pre
+pip install onnxruntime-genai
 ```
 
 ### DirectML
 
-Append `-directml` for the library that is optimized for DirectML on Windows
-
 ```bash
 pip install numpy
-pip install onnxruntime-genai-directml --pre
+pip install onnxruntime-genai-directml
 ```
 
 ### CUDA
 
-Append `-cuda` for the library that is optimized for CUDA environments
+If you are installing the CUDA variant of onnxruntime-genai, the CUDA toolkit must be installed.
+
+The CUDA toolkit can be downloaded from the [CUDA Toolkit Archive](https://developer.nvidia.com/cuda-toolkit-archive).
+
+Ensure that the `CUDA_PATH` environment variable is set to the location of your CUDA installation.
 
 #### CUDA 11
 
 ```bash
 pip install numpy
-pip install onnxruntime-genai-cuda --pre --index-url=https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-genai/pypi/simple/
+pip install onnxruntime-genai-cuda --index-url=https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-genai/pypi/simple/
 ```
 
 #### CUDA 12
 
 ```bash
 pip install numpy
-pip install onnxruntime-genai-cuda --pre --index-url=https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/
+pip install onnxruntime-genai-cuda --index-url=https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-12/pypi/simple/
 ```
 
-## Nuget packages
 
-Note: only one of these packages should be installed in your application.
+## Nuget package installation
+
+Note: install only one of these packages (CPU, DirectML, CUDA) in your project.
+
+### Pre-requisites
+
+#### ONNX Runtime dependency
+
+ONNX Runtime generate() versions 0.3.0 and earlier came bundled with the core ONNX Runtime binaries. From version 0.4.0 onwards, the packages are separated to allow a more flexible developer experience.
+
+Version 0.4.0-rc1 depends on the ONNX Runtime version 1.19.0 RC. To install 0.4.0-rc1, add the following nuget source *before* installing the ONNX Runtime generate() nuget package.
+
+```
+dotnet nuget add source https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/ORT-Nightly/nuget/v3/index.json --name ORT-Nightly
+```
+
+### CPU
 
 ```bash
 dotnet add package Microsoft.ML.OnnxRuntimeGenAI --prerelease
 ```
 
-For the package that has been optimized for CUDA:
+### CUDA 
+
+Note: only CUDA 11 is supported for versions 0.3.0 and earlier, and only CUDA 12 is supported for versions 0.4.0 and later.
 
 ```bash
 dotnet add package Microsoft.ML.OnnxRuntimeGenAI.Cuda --prerelease
 ```
 
-For the package that has been optimized for DirectML:
+### DirectML
 
 ```bash
 dotnet add package Microsoft.ML.OnnxRuntimeGenAI.DirectML --prerelease