Skip to content

Latest commit

 

History

History
108 lines (77 loc) · 4.85 KB

README.md

File metadata and controls

108 lines (77 loc) · 4.85 KB
Raw

Chat Copilot Memory Pipeline

!IMPORTANT This sample is for educational purposes only and is not recommended for production deployments.

IMPORTANT: The pipeline will call Azure OpenAI/OpenAI which will use tokens that you may be billed for.

Introduction

Memory

One of the exciting features of the Chat Copilot App is its ability to store contextual information to memories and retrieve relevant information from memories to provide more meaningful answers to users through out the conversations.

Memories can be generated from conversations as well as imported from external sources, such as documents. Importing documents enables Chat Copilot to have up-to-date knowledge of specific contexts, such as enterprise and personal data.

Memory pipeline in Chat Copilot

Chat copilot integrates Kernel Memory as the memory solution provider. The memory pipeline is designed to be run as an asynchronous service. If you are expecting to import big documents that can require minutes to process or planning to carry long conversations with the bot, then you can deploy the memory pipeline as a separate service along with the chat copilot webapi.

Configuration

(Optional) Before you get started, make sure you have the following requirements in place:

Webapi

Please refer to the webapi README.

Memorypipeline

The memorypipeline is only needed when KernelMemory:DataIngestion:OrchestrationType is set to Distributed in ../webapi/appsettings.json.

  • Content Storage: storage solution to save the original contents. Available options:
    • AzureBlobs
    • SimpleFileStorage: stores data on your local file system.
  • Message Queue: asynchronous service to service communication. Available options:
    • AzureQueue
    • RabbitMQ
    • SimpleQueues: stores messages on your local file system.
  • Vector database: storage solution for high-dimensional vectors, aka embeddings. Available options:
    • AzureAISearch
    • Qdrant
    • SimpleVectorDb
      • TextFile: stores vectors on your local file system.
      • Volatile: stores vectors in RAM.

        Note that do not configure the memory pipeline to use Volatile. Use volatile in the webapi only when its KernelMemory:DataIngestion:OrchestrationType is set to InProcess.

AzureBlobs & AzureQueue

Note: Make sure to use the same resource for both the webapi and memorypipeline.

  1. Create a storage account.
  2. Find the connection string under Access keys on the portal.
  3. Run the following to set up the authentication to the resources: 
    dotnet user-secrets set KernelMemory:Services:AzureBlobs:Auth ConnectionString
dotnet user-secrets set KernelMemory:Services:AzureBlobs:ConnectionString [your secret]
dotnet user-secrets set KernelMemory:Services:AzureQueue:Auth ConnectionString
dotnet user-secrets set KernelMemory:Services:AzureQueue:ConnectionString [your secret]
Azure Cognitive Search

Note: Make sure to use the same resource for both the webapi and memorypipeline.

  1. Create a search service.
  2. Find the Url under Overview and the key under Keys on the portal.
  3. Run the following to set up the authentication to the resources: 
    dotnet user-secrets set KernelMemory:Services:AzureAISearch:Endpoint [your secret]
dotnet user-secrets set KernelMemory:Services:AzureAISearch:APIKey [your secret]
RabbitMQ

Note: Make sure to use the same queue for both webapi and memorypipeline

Run the following:

docker run -it --rm --name rabbitmq \
  -e RABBITMQ_DEFAULT_USER=user \
  -e RABBITMQ_DEFAULT_PASS=password \
  -p 5672:5672 \
  rabbitmq:3
Qdrant

Note: Make sure to use the same vector storage for both webapi and memorypipeline

docker run -it --rm --name qdrant \
-p 6333:6333 \
qdrant/qdrant

To stop the container, in another terminal window run

docker container stop [name]
docker container rm [name]