Welcome to Ragable!

Ragable is an AI library that helps you build multi-turn Chatbots with ease, the library provides several easy-to-use classes to help you build "agent style" workflows. The Agent will first analyze the user's input and then determine what function to invoke (known as Runnable's).

Note: This is a BETA version and may not be suitable for production just yet. A stable PIP package and tag will be published soon in the next few weeks. Please consider subscribing to my newsletter @ https://kevincoder.co.za to get notified of updates.

Each Runnable takes the following arguments:

• Name: a unique identifier that describes this runnable.
• Instruction: Instruct the LLM on when it should invoke this runnable.
• Func : Any python function you want to run if the model chooses to execute this Runnable. The function should return text as it's output. You can also pass a vector store to this function, when doing so, the Runnable will query the vector store for similar documents and pass on to the LLM for further reasoning.
• AskLLM: By default the LLM will simply execute the function and return it's output to the user. If you prefer to pass on the functions output to the LLM for further reasoning, set this to True.
• Params: Any local data you want to make available to the Runnable function. This can be used to pass a user object or session information. Params are not sent to the LLM.

from ragable.agent import get_openai_agent
from ragable.runnable import Runnable
from ragable.adapters.qdrant import QdrantAdapter
from ragable.embedders import StandardEmbedder
import os

# See examples.py for more details
    agent = get_openai_agent()
    qdrant = QdrantAdapter("ragable_documents")
    embedder = StandardEmbedder(qdrant)
    embedder.train_from_document("./testdata/bulbasaur.txt")

    bulbasaur_knowledge = Runnable(
        Name="Information about bulbasaur",
        Instruction="When the human asks about bulbasaur",
        Func=qdrant
    )

    agent.add_tasks([
        legendary_pokemon,
        php_strings,
        bulbasaur_knowledge
    ])

    questions = [
        "What is a legendary pokemon?",
        "How to perform a string replace in PHP?",
        "How to find a string in another string in PHP?",
        "Which Pokemon are the evolved forms of bulbasaur?"
    ]

    for q in questions:
        response = agent.invoke(q)
        print(response)

How to install?

You can GIT clone this repository and run:

pip install requirements.txt

Thereafter, you can then import any of the Ragable classes into your Python projects.

Note: If you plan on using a vector store with Ragable, we currently support Qdrant. You can use their docker image to setup Qdrant as follows:

docker run -dit --name raggable-qdrant -p 6333:6333 -p 6334:6334 \
-v $(pwd)/qdrant_storage:/qdrant/storage:z \
qdrant/qdrant