Project templates

Kairntech offers a set of project templates (project archive file .spa) to simplify the creation and configuration of your project. Once you’ve imported the project template, all you need to do is import your own documents:

👉 You can download project templates here.

How to proceed?

Question answering

A question-answering project template contains:

  • A set of pre-configured vectorizers (sentence transformer embedding models) for semantic search (language-dependent)
  • A set of pre-configured Large Language Models (LLMs) to generate the response (GPT3.5, GPT4, Llama2 70B, Mistral 7B, Mixtral 8*7B…)

How to create your project with this template?

  • Create a project using “import project”.
    • select the template file (.spa) you have downloaded above
    • name the new project and create it
  • Open your project by clicking on the ticker / project title
  • Import documents
  • A job is started to import, index and vectorize all documents in your project
  • Go to the Q&A menu and ask question
  • Experiment different vectorizers for semantic search and LLM to generate answer
  • For frequently asked questions:
    • create a JSON file with these questions using Kairntech JSON format (sample here)
    • import the question file
  • If you want to modify or create new components, go to the processing menu

Text summarization

A Text summarization template contains:

  • A set of pre-configured Large Language Models (GPT3.5, GPT4, Llama2 70B, Mistral 7B…) to generate Text summarization

How to create your project with a template?

  • Create a project using “import project”.
    • select the template file (.spa) you have downloaded above
    • name the new project and create it
  • Open your project by clicking on the ticker / project title
  • Import documents or go directly to the Annotation test menu
  • Test the different pre-configured text summarization models in the text page as below:
  • If you want to modify or create new Text summarizaion models, go to the Processing menu