Skip to content

🎨 Web Playground

Interactive Gradio interface for experimenting with Rankify models.

Quick Start

from rankify.ui import launch_playground

# Launch interactive playground
launch_playground(port=7860)

Then open http://localhost:7860 in your browser.

Features

🔍 Search Tab

Test different retrievers interactively: - Enter your query - Select retriever (BM25, BGE, ColBERT, etc.) - View ranked results - Copy generated code

📊 Rerank Tab

Test rerankers on your documents: - Enter query and documents - Select reranker method - See reranked results with scores - Export code

🤖 RAG Tab

Full RAG pipeline testing: - Enter question - Configure retriever + reranker + RAG method - Generate answers with context - View source documents

💻 Code Generator

Generate code for any configuration: - Select task type - Choose models - Get ready-to-use Python code

Screenshot

┌──────────────────────────────────────────────────────────────┐
│  🚀 Rankify Playground                                       │
├──────────────────────────────────────────────────────────────┤
│  [🔍 Search] [📊 Rerank] [🤖 RAG] [💻 Code Generator]        │
├──────────────────────────────────────────────────────────────┤
│                                                              │
│  Query: [What is machine learning?_____________]             │
│                                                              │
│  Retriever: [bge ▼]     N Docs: [50 ──●──── 100]            │
│                                                              │
│  [🔍 Search]                                                 │
│                                                              │
│  ─────────────────────────────────────────────────────────   │
│  Results:                                                    │
│  1. Introduction to Machine Learning                         │
│     Machine learning is a subset of AI that enables...       │
│                                                              │
│  2. Types of ML Algorithms                                   │
│     There are three main types: supervised, unsupervised...  │
│                                                              │
│  ─────────────────────────────────────────────────────────   │
│  Code:                                                       │
│  ```python                                                   │
│  from rankify import pipeline                                │
│  search = pipeline("search", retriever="bge", n_docs=50)     │
│  results = search("What is machine learning?")               │
│  ```                                                         │
└──────────────────────────────────────────────────────────────┘

Configuration

from rankify.ui import launch_playground

launch_playground(
    port=7860,           # Port to run on
    share=True,          # Create public Gradio link
    server_name="0.0.0.0",  # Bind address
)

Use Cases

  1. Quick Testing - Try models before writing code
  2. Demo to Stakeholders - Show capabilities visually
  3. Model Comparison - Test different configurations
  4. Code Generation - Generate starter code
  5. Debugging - Understand model behavior

Requirements

Install Gradio: 

pip install gradio

Customization

Create a custom playground:

from rankify.ui.playground import create_playground_app

# Get the Gradio app
app = create_playground_app()

# Add custom components
with app:
    gr.Markdown("## My Custom Section")
    # ... add more

# Launch
app.launch()