curious-me

Small library to ask questions, get reviews and citations of papers. Currently papers are fetched from arxiv but it can be possible to use other sources as well.

Installation

1. Create virtual environment and activate it

python -m venv venv

In Windows

venv\Scripts\activate

In GNU/Linux

source venv/bin/activate

2. Install package If using pip

pip install curious-me

If cloned the repo

cd curious-me/curious_me
pip install .

4. Run the application, first open python then run the following commands

from curious_me import Curious
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
            model_name="grok-2-1212",
            temperature=0.1,
            base_url="https://api.x.ai/v1",
            api_key=<your api key>,
        )

topics = ['GPT', 'LLM', 'Decoder', 'Artificial generative Intelligence']
curious = Curious(topics=topics, llm=llm) # or Curious(topics=topics, llm=llm, skip_search=True) if you # want to search papers and build vector store again
curious.ask("What are recent advances in GPT?")
curious.get_review("RAG")
curious.get_citation(claim="Leaky ReLU is better than ReLU")

Note:

Steps for using own pdfs

1. Clone the repo.
2. Add your pdfs in the folder curious-me/curious_me/research_papers
3. Run the following commands

from curious_me import Curious
from langchain_openai import ChatOpenAI
llm = ChatOpenAI(
            model_name="grok-2-1212",
            temperature=0.1,
            base_url="https://api.x.ai/v1",
            api_key=<your api key>,
        )
topics = [] # your topics
curious = Curious(topics=topics, llm=llm, skip_search=True, rebuild_vec_store=True)
curious.ask("Questions?")