Welcome to the Weaviate Javascript recipes repository!

This repo covers end-to-end examples on the various features and integrations with Weaviate for Javascript Developers!

⚠️ Before getting started, you need to follow the installation Setup Instructions detailed in the setup section of this document. You will need the setup completed to successfully run the recipes.

Setup Instructions 🚀

1. Install npm packages

Clone this repository, and install dependencies

npm install

2. Choose where to run Weaviate

2.1 Run in Weaviate Cloud Service

Head to WCS, where you can easily create a free sandbox cluster. Take note of your cluster url and apiKey

2.2 Run locally using Docker

Considering you already have docker installed, you can run:

docker compose up -d

IMPORTANT: make sure to define the environment variables before running Docker

3. Define environment variables

get your OPENAI key here

get your COHERE key here

cp .env_example .env

If you are using docker, you can keep it like:

WEAVIATE_SCHEME_URL=http
WEAVIATE_URL=localhost:8080
OPENAI_API_KEY=<your openai apikey>
COHERE_API_KEY=<your cohere apikey>

if you are using WCS, you can keep it like:

WEAVIATE_SCHEME_URL=https
WEAVIATE_URL=<yourcluster.weaviate.network>
WEAVIATE_API_KEY=<your_apikey>
OPENAI_API_KEY=<your openai apikey>
COHERE_API_KEY=<your cohere apikey>
HUGGING_FACE_API_KEY=<your huggingface apikey>
PALM_API_KEY=<your palm apikey>

4. Run a Recipe!

npm run <recipe>

current available recipes:

npm run generative-search/openai
npm run generative-search/cohere
npm run similarity-search/text2vec/openai
npm run conditional-filters/contains-all-any
npm run multi-tenancy/example
npm run classification/zeroshot
npm run classification/knn
npm run data-with-vectors
more coming soon!

Here is an outline of the concepts this repository covers:

Similarity Search 🔎

Similarity Search shows how to run nearText, nearObject and nearVector queries in Weaviate. It is divided by the different providers:

Cohere - npm run similarity-search/text2vec/cohere
HuggingFace - npm run similarity-search/text2vec/hugging-face
OpenAI - npm run similarity-search/text2vec/openai
PaLM - npm run similarity-search/text2vec/palm
Transformers - coming soon ⏳

Hybrid Search ⚖️

Hybrid Search allows you to combine keyword and vector search. The notebook covers how to run a hybrid search query, search on a specific property, add in a where filter, and how to search with an embedding. It is divided by the different providers:

Cohere - coming soon ⏳
Contextionary - coming soon ⏳
HuggingFace - coming soon ⏳
OpenAI - coming soon ⏳
PaLM - coming soon ⏳
Transformers - coming soon ⏳

Generative Search ⌨️

Generative Search allows you to improve your search results by piping them through LLM models. It is divided by the different providers:

Cohere - npm run generative-search/cohere
OpenAI - npm run generative-search/openai
PaLM - npm run generative-search/palm

Classification 🗂️

Classification allows you to classify data objects by predicting cross-references based on the semantic meaning of the data objects.

Knn - npm run classification/knn
Zero-Shot - npm run classification/zeroshot

Integrations 🤝

Integrations with Weaviate

LlamaIndex
- Episode 1: Data Loaders - coming soon ⏳
- Episode 2: Indexes - coming soon ⏳
- Simple Query Engine - coming soon ⏳
- Sub Question Query Engine - coming soon ⏳
Llama 2 Demo
- Quick demo on using the Llama 2 model using Replicate and LlamaIndex - coming soon ⏳
GPTCache
- GPTCache notebook - coming soon ⏳

Ranking 🏅

Ranking in Weaviate

Cohere Rerank - coming soon ⏳

Feedback ❓

Please note this is an ongoing project, and updates will be made frequently. If you have a feature you would like to see, please drop it in the Weaviate Forum.

dudanogueira / recipes-js