/company-climate-rag-llamaindex

Eco-impact assessment tool using the Llama-Index to process Wikipedia data for query-specific, AI-powered information retrieval and analysis

Primary LanguageJavaScript

Company Climate RAG LlamaIndex

Introduction

In the wake of urgent climate action, we envisioned a tool that could quantify corporate sustainability efforts and inspire real change. Imagine if businesses could navigate the complexities of sustainability like they check their credit scores.

Company Climate RAG LlamaIndex

WebPage_1

WebPage_2

What it does?

Utilizing the cutting-edge Llama-Index, our platform provides a robust and dynamic assessment of a company's environmental impact. Firstly, we extract unstructured data from Wikipedia from LlamaIndex data loaders. Next, we performing chunking/segmentation to chunk the long-context data and store the data in MongoDB vectorstore. Finally, given a query we performing LlamaIndex recursive retrieval + document agent architecture to route the query to the most relevant document agent. Finally, we pass the relevant chunks to Llama2 from Together AI for efficient question answering.

Tools

  • MongoDB Atlas Vector Search
  • Llama2-chat (7B)
  • Llama-index
  • TogetherAI

Contributors

@DanielDaCosta @ianwu13 @sauravjoshi23

Reference