<<<<<<< HEAD

This is a sample app that combines Elasticsearch, Langchain and a number of different LLMs to create a chatbot experience with ELSER with your own private data.

Requires at least 8.11.0 of Elasticsearch.

Download the project from Github and extract the chatbot-rag-app folder.

curl https://codeload.github.com/elastic/elasticsearch-labs/tar.gz/main | \
tar -xz --strip=2 elasticsearch-labs-main/example-apps/chatbot-rag-app

There are a number of ways to install Elasticsearch. Cloud is best for most use-cases. Visit the Install Elasticsearch for more information.

This app requires the following environment variables to be set to connect to Elasticsearch hosted on Elastic Cloud:

export ELASTIC_CLOUD_ID=...
export ELASTIC_API_KEY=...

You can add these to a .env file for convenience. See the env.example file for a .env file template.

You can also connect to a self-hosted Elasticsearch instance. To do so, you will need to set the following environment variables:

export ELASTICSEARCH_URL=...

By default, the app will use the workplace-app-docs index and the chat history index will be workplace-app-docs-chat-history. If you want to change these, you can set the following environment variables:

ES_INDEX=workplace-app-docs
ES_INDEX_CHAT_HISTORY=workplace-app-docs-chat-history

We support three LLM providers: Azure, OpenAI and Bedrock.

To use one of them, you need to set the LLM_TYPE environment variable:

export LLM_TYPE=azure

To use OpenAI LLM, you will need to provide the OpenAI key via OPENAI_API_KEY environment variable:

export LLM_TYPE=openai
export OPENAI_API_KEY=...

You can get your OpenAI key from the OpenAI dashboard.

If you are using Azure LLM, you will need to set the following environment variables:

export LLM_TYPE=azure
export OPENAI_VERSION=... # e.g. 2023-05-15
export OPENAI_BASE_URL=...
export OPENAI_API_KEY=...
export OPENAI_ENGINE=... # deployment name in Azure

To use Bedrock LLM you need to set the following environment variables in order to AWS.

export LLM_TYPE=bedrock
export AWS_ACCESS_KEY=...
export AWS_SECRET_KEY=...
export AWS_REGION=... # e.g. us-east-1
export AWS_MODEL_ID=... # Default is anthropic.claude-v2

Optionally, you can connect to AWS via the config file in ~/.aws/config described here: https://boto3.amazonaws.com/v1/documentation/api/latest/guide/credentials.html#configuring-credentials

[default]
aws_access_key_id=...
aws_secret_access_key=...
region=...

To use Vertex AI you need to set the following environment variables. More infos here.

export LLM_TYPE=vertex
export VERTEX_PROJECT_ID=<gcp-project-id>
export VERTEX_REGION=<gcp-region> # Default is us-central1
export GOOGLE_APPLICATION_CREDENTIALS=<path-json-service-account>

Once you have indexed data into the Elasticsearch index, there are two ways to run the app: via Docker or locally. Docker is advised for testing & production use. Locally is advised for development.

Build the Docker image and run it with the following environment variables.

docker build -f Dockerfile -t chatbot-rag-app .

Make sure you have a .env file with all your variables, then run:

docker run --rm --env-file .env chatbot-rag-app flask create-index

See "Ingest data" section under Running Locally for more details about the flask create-index command.

You will need to set the appropriate environment variables in your .env file. See the env.example file for instructions.

docker run --rm -p 4000:4000 --env-file .env -d chatbot-rag-app

Note that if you are using an LLM that requires an external credentials file (such as Vertex AI), you will need to make this file accessible to the container in the run command above. For this you can use a bind mount, or you can also edit the Dockerfile to copy the credentials file to the container image at build time.

With the environment variables set, you can run the following commands to start the server and frontend.

• Python 3.8+
• Node 14+

For Python we recommend using a virtual environment.

ℹ️ Here's a good primer on virtual environments from Real Python.

# Create a virtual environment
python -m venv .venv

# Activate the virtual environment
source .venv/bin/activate

# Install Python dependencies
pip install -r api/requirements.txt

# Install Node dependencies
cd frontend && yarn && cd ..

You can index the sample data from the provided .json files in the data folder:

flask create-index

By default, this will index the data into the workplace-app-docs index. You can change this by setting the ES_INDEX environment variable.

The ingesting logic is stored in data/index-data.py. This is a simple script that uses Langchain to index data into Elasticsearch, using the JSONLoader and CharacterTextSplitter to split the large documents into passages. Modify this script to index your own data.

Langchain offers many different ways to index data, if you cant just load it via JSONLoader. See the Langchain documentation

Remember to keep the ES_INDEX environment variable set to the index you want to index into and to query from.

# Launch API app
flask run

# In a separate terminal launch frontend app
cd frontend && yarn start

build a lLarge language model (LLM) chatbot that uses a pattern known as Retrieval-Augmented Generation (RAG)

origin/main