Spaces:
Sleeping
Sleeping
File size: 1,390 Bytes
9f8c4dc |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 |
---
title: Intelligent Web-Search RAG
emoji: 🌐
colorFrom: blue
colorTo: yellow
sdk: docker
app_port: 7860
pinned: false
---
🌐 Intelligent Web-Search RAG Chatbot
An advanced Retrieval-Augmented Generation (RAG) chatbot that combines the reasoning power of Llama 3.2 with real-time web search capabilities. Unlike standard AI, this agent can browse the live internet to answer questions about current events, world leaders, and breaking news.
🚀 Key Features
Real-Time Web Intelligence: Uses a custom-built DuckDuckGo search tool to fetch current facts beyond the model's training cutoff.
Conversational Memory: Remembers previous interactions using ConversationBufferMemory for a natural, multi-turn chat experience.
Intelligent Reasoning: Powered by the ReAct (Reason + Act) framework, allowing the AI to decide when to search the web and when to answer from memory.
Optimized for Performance: Implements smart result truncation to prevent context overflow and handles API rate limits gracefully.
Deployment Ready: Fully configured for local hosting or cloud deployment (Hugging Face Spaces, Docker).
🛠️ Tech Stack
Core Logic: LangChain (Agents, Tools, and Memory)
LLM: Meta Llama 3.2 3B Instruct via Hugging Face Endpoint
Search Engine: DuckDuckGo Search API (ddgs)
Web Framework: Flask (Python)
Environment: Python 3.10+ / Anaconda |