Add Hugging Face README config

README.md

@@ -1,52 +1,11 @@
-# Glacial Lake Outburst Flood (GLOF) Agentic System
-
-An advanced, multimodal AI agent designed to monitor, forecast, and explain Glacial Lake Outburst Flood (GLOF) risks in real-time. This system fuses Deep Learning computer vision, Graph Neural Networks (GNN), XGBoost, and an Agentic Retrieval-Augmented Generation (RAG) backend to act as a complete early warning and analysis system.
-
-## Features
-*   **Agentic RAG Backend**: Processes natural language queries to search scientific literature, traverse a semantic Knowledge Graph, or trigger live predictive models.
-*   **Dual-Inference Forecasting**: 
-    *   **XGBoost**: Predicts GLOF susceptibility based on historical static geomorphology and live 90-day ERA5 weather data.
-    *   **Graph Neural Network (GConvGRU)**: A spatiotemporal physics-informed model that simulates surface runoff and snowmelt cascading through the downstream river network.
-*   **Live Geospatial UI**: A web-based frontend integrating Mapbox and Google Earth Engine (GEE) to visualize glacial lakes, river hierarchies, building infrastructure, and real-time inference results.
-*   **Hybrid Knowledge Graph**: Combines dense vector searching (BAAI/bge-m3) with deterministic SPARQL/Graph traversal (RDFLib) to ensure physically accurate answers without LLM hallucination.
-
-## Setup & Installation
-
-### Prerequisites
-*   [Miniconda](https://docs.anaconda.com/free/miniconda/) or Anaconda installed.
-*   Google Earth Engine account/credentials.
-*   Hugging Face API Token.
-
-### 1. Create the Environment
-Due to complex C++ dependencies for PyTorch Geometric on Windows, use the provided Conda environment file:
-```bash
-conda env create -f environment.yml
-conda activate glof_agent
-```
-
-### 2. Environment Variables
-You must set your Hugging Face API token in your environment (or create a `.env` file):
-```bash
-setx HF_TOKEN "your_hf_token_here"
-```
-
-### 3. Run the Application
-You can start both the backend FastAPI server and the frontend web interface simply by running:
-```bash
-run_application.bat
-```
-*(Alternatively, you can run `python chatbot/chat_server.py` and serve the `website/` folder manually).*
-
-## Repository Structure
-*   `chatbot/`: The core Agentic RAG logic, routing, Earth Engine fetching, and dual-inference scripts.
-*   `models/`: Pre-trained weights for the GNN (`.pth`), Lake Segmentation U-Net (`.pth`), and XGBoost (`.joblib`).
-*   `gis_data/`: Geopackages (`.gpkg`) containing the static river networks, glacial lake boundaries, and building footprints.
-*   `rag/`: Vector indices and the semantic Knowledge Graph (`.ttl`).
-*   `website/`: The HTML/JS/CSS code for the interactive Mapbox frontend.
-
-## Git Large File Storage (LFS)
-This repository uses Git LFS to host heavy model weights and `.gpkg` files. If you clone this repository, make sure you have Git LFS installed to pull the actual files rather than their pointer references:
-```bash
-git lfs install
-git lfs pull
-```
+---
+title: GLOF Backend API
+emoji: 🌊
+colorFrom: blue
+colorTo: indigo
+sdk: docker
+pinned: false
+---
+
+# GLOF Backend API
+This is the Docker backend for the GLOF Agentic RAG System.