Chat with your dataset.

CSV Data Agent with DeepSeek-R1

Overview

This project demonstrates how to interact with CSV data using a DeepSeek-R1-Distill-Qwen-32B. It integrates a streaming LLM from Hugging Face to process natural language queries on a pandas DataFrame. The project leverages langchain, pandas, and huggingface_hub to enable seamless data interaction.

Features

• Uses a Hugging Face-hosted LLM (deepseek-ai/DeepSeek-R1-Distill-Qwen-32B) for processing queries.
• Parses structured output from the LLM.
• Integrates langchain to create a pandas DataFrame agent.
• Allows querying CSV data using natural language prompts.
• Supports structured parsing for responses.

Installation

1. Clone this repository:

git clone https://github.com/DataByteSun/CSV-Data-Agent-with-DeepSeek-R1.git

2. Create and activate a virtual environment (optional but recommended):

python -m venv venv
source venv/bin/activate   # On Windows use: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Set up your environment variables:

• Create a .env file in the project directory and add your Hugging Face API key(Get Free key from <a href="https://huggingface.co/" target="_blank">Huggung Face</a>):

  HUGGINGFACE_API_KEY=your_huggingface_api_key
  

• Load environment variables:

  from dotenv import load_dotenv
  load_dotenv()
  

Usage

1. Load and preprocess your CSV data:

import pandas as pd
df = pd.read_csv("your_data.csv")

2. Instantiate the LLM and create an agent:

from huggingface_hub import InferenceClient

client = InferenceClient(api_key="your_huggingface_api_key")
agent = create_pandas_dataframe_agent(llm=client, df=df, verbose=True)

3. Query the agent using natural language:

result = agent.invoke("how many rows are there?")

Expected Output (Screenshots)

• Running agent.invoke("how many rows are there?") may yield an output like: ![image](https://github.com/user-attachments/assets/9a