Skip to content

Latest commit

 

History

History
37 lines (23 loc) · 1.58 KB

README.md

File metadata and controls

37 lines (23 loc) · 1.58 KB

🌱 Climate Mate

Climate Mate is a RAG application that utilizes Google AI (Gemini) to respond to user questions on the topic of climate change and generate embedding vectors for use in semantic search. The software relies on data from trusted sources such as NASA and the Intergovernmental Panel on Climate Change (IPCC).

Try it

Navigate to http://www.climate-mate.org and try asking questions on climate change topic. There are some prebuilt quick questions for you to try too.

Workflows

Upload document

upload document workflow

Query

query workflow

Exposed endpoints

[POST] http://www.climate-mate.org/v1/upload
This is protected endpoint. The one used to upload and index the file, where the request body is multipart form with two fields - file (bin) and summary (string)

[GET] http://www.climate-mate.org/v1/query
Query endpoint is used to return an answer to the user's question.
Arguments:

  • required q - the user question. For example: ?q="what is climate change?"
  • optional searchby - strategy used search in indexed documents. Supports two options: top (default) and wide. Here top picks the top N pages by score, whereas wide takes top N/(number of files) from each file to form a final list.

[GET] http://www.climate-mate.org/v1/search
Semantic search in vectore store by user input.
Arguments:

  • required q - same as in query
  • optional searchby - same as in query
  • optional n - number of pages to return (default - 10).