Getting Started

This is a guide to help you get started with the hexamind library.

Installation

Using pip

You should be able to install the library using pip. But you must have the hxm_rag repository cloned on your local machine. Because this library is private and not yet deploy on private registry.

git clone https://github.com/HexamindOrganisation/hexamind.git

Then you can install the library using pip.

pip install -e path/to/hexamind

Usage

The library serves many purpose and can be used not only for RAG solutions but also when you need to parse some document and just use a Llm agent.

Parsing a docx file

Here is a simple example of how to parse a docx file using the WordReader class.

from hxm_rag.model import WordReader

# Create an instance of the WordReader class

word_reader = WordReader("path/to/your/docx/file")

# Create a nested dictionnary structure of the document

doc_structure = word_reader.get_document_structure()

this sample should give you a nested dictionnary structure that must look like this:

  {
        'type' : 'container' or 'block',
        'children' : [list of nested dictionaries],
        'content' : None or string,
        'level' : int,
        'parent' : parent dictionary
    }

See readers for more information about the readers available in the library and the return format.

Converting a nested dictionnary into a Document object

To get more information about the document click here

You can then convert the nested dictionnary into the hxm_rag Document object.

from hexamind import Document

doc = Document(doc_structure)

Using the LLm agent

You can use select the proper agent in many ways, feel free to explore the llm module to see the different model supported by the library.

from hexamind.llm.adapters import MistralClientAdapter
from hexamind import LlmAgent
from mistralai import MistralClient

# Create an instance of the MistralClient class

mistral_client = MistralClientAdapter(MistralClient('your_mistral_api_key'))

# Create an instance of the LlmAgent class

llm_agent = LlmAgent(mistral_client)

A factory method is also available to instanciate the client automatically. This can be coupled with the Initializer class that can initialize every component needed in your solution.