Informational Retriever Task

A document retrieval system with multiple ranking algorithms and evaluation capabilities.

Structure

• playground.ipynb - Main pipeline and experimentation notebook
• rankers/ - Various ranking algorithms (BM25, TF-IDF, embedding-based)
• indexes/ - Indexing implementations for different approaches
• chroma_db/ - Vector database storage (not in remote repository, created in playground.ipynb)
• evaluator.py - Evaluation metrics and testing
• document_loader.py - Document processing utilities
• results_*.json - Evaluation results from different models

Usage

Run the main pipeline in playground.ipynb to experiment with different retrieval approaches and see the full evaluation.