https://github.com/jv456/gemmatician

A generative AI Streamlit app that solves math and logic problems from natural language, powered by Google Gamma and Langchain agents.
https://github.com/jv456/gemmatician

google-gemma2 langchain streamlit

Last synced: about 2 months ago
JSON representation

A generative AI Streamlit app that solves math and logic problems from natural language, powered by Google Gamma and Langchain agents.

• Host: GitHub
• URL: https://github.com/jv456/gemmatician
• Owner: JV456
• Created: 2025-07-19T12:49:57.000Z (11 months ago)
• Default Branch: main
• Last Pushed: 2025-07-19T15:38:21.000Z (11 months ago)
• Last Synced: 2025-07-19T17:12:00.936Z (11 months ago)
• Topics: google-gemma2, langchain, streamlit
• Language: Python
• Homepage:
• Size: 5.86 KB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Gemmatician - MathsGPT

Welcome to my **Gemmatician - MathsGPT** – an end-to-end generative AI application powered by Google Gamma! This app is built using Streamlit and leverages the latest AI models and agent toolchains to provide dynamic step-by-step solutions for math and logical reasoning problems.

---

## ✨ Features

- **Chatbot Interface:** Ask math questions in natural language and get step-by-step explanations.

- **Supports Logical Reasoning:** Not just calculations–get thoughtful answers to logical and multi-step problems.

- **Integrated Wikipedia Search:** Fetches relevant information for queries requiring background knowledge.

- **Persistent Chat History:** Conversations are maintained for session continuity using Streamlit's session state.

- **Agent-Based Toolchain:** Combines multiple tools (calculator, Wikipedia, logic/reasoning) with Google Gamma's LLM for robust results.

---

## 🛠️ How It Works

### Architecture

- **Streamlit** provides the frontend and stateful chat interface.

- **Google Gamma (via Grok API):** The backbone LLM for all generative and reasoning tasks.

- **Tools & Agents (Langchain):**

  - **Wikipedia Tool:** Handles search and fetch queries.

  - **Calculator Tool:** Parses and solves mathematical expressions.

  - **Reasoning Tool:** Runs custom prompts with the LLM to solve multi-step or logic-based queries.

### Flow

1. **User inputs a math or logic question.**

2. **Agent orchestrates** the required tools based on the task:

    - For direct math: Calculator Tool

    - For context/knowledge: Wikipedia Tool

    - For logic/reasoning: Reasoning Tool (LLM chain with a prompt template)

3. **Step-by-step response**, often with pointwise explanations, is displayed in the chat interface.

4. **Session history** is maintained, and every new question/answer is appended to the chat.

---

## 📝 Example Usage and Output





---