Skip to main content

​
Installation

Follow these steps to install Quest and set up the required dependencies.

​
Prerequisites

Before installing Quest, ensure you have the following installed:
  • Python 3.8+ - Quest requires Python 3.8 or higher
  • Git - For cloning the repository
  • Ollama - For running local language models (we’ll install this in the next section)

​
Installation steps

1

Clone the repository

Clone the Quest repository from GitHub:
git clone https://github.com/udit-rawat/Quest.git
cd Quest
This downloads the complete Quest codebase including the Flask app, RAG engine, and pre-built HNSW index.
2

Install Python dependencies

Install all required Python packages using pip:
pip install -r requirements.txt
This installs the following key dependencies:
  • faiss-cpu - Fast similarity search library
  • sentence-transformers - For encoding queries and documents
  • flask - Web framework for the API
  • torch - PyTorch for ML models
  • transformers - Hugging Face transformers library
  • requests - For Ollama API communication
  • numpy and pandas - Data processing libraries
The installation may take several minutes as it downloads pre-trained models and compiles native extensions.
3

Unzip the solutions dataset

Extract the pre-generated solutions dataset:
unzip leetcode_solutions.zip -d src/DSAAssistant/components/
This extracts 1800+ JSON solution files to the components directory. Each file contains:
  • Problem title and description
  • Complete solution code
  • Difficulty level (Easy, Medium, Hard)
  • Related topics (Arrays, Dynamic Programming, etc.)
  • Companies that ask the problem
The solutions were generated using qwen2.5-coder:1.5b and manually enriched with metadata.
4

Verify the installation

Check that all required files are in place:
# Verify HNSW index exists
ls src/DSAAssistant/components/leetcode_hnsw2.index

# Verify metadata file exists
ls src/DSAAssistant/components/leetcode_metadata2.pkl

# Verify solutions were extracted
ls src/DSAAssistant/components/leetcode_solutions/ | head -5
You should see the index file, metadata pickle file, and solution JSON files.
If any files are missing, check that you extracted the zip file to the correct directory.

​
Verify Python imports

Test that the core components can be imported: 
from src.DSAAssistant.components.retriever2 import LeetCodeRetriever
from rag_engine3 import RAGEngine

# Initialize the retriever
retriever = LeetCodeRetriever()
print(f"Loaded {len(retriever.solutions)} solutions")

# Should print: Loaded 1800+ solutions
If you see import errors, ensure you’re running Python from the Quest root directory and all dependencies were installed correctly.

​
Directory structure

After installation, your Quest directory should look like this: 
Quest/
├── app.py                          # Flask application entry point
├── rag_engine3.py                  # RAG engine implementation
├── requirements.txt                # Python dependencies
├── README.md                       # Project documentation
├── src/
│   └── DSAAssistant/
│       └── components/
│           ├── retriever2.py       # LeetCode retriever
│           ├── memory_buffer.py    # Conversation history
│           ├── prompt_temp.py      # Prompt templates
│           ├── leetcode_hnsw2.index      # FAISS HNSW index
│           ├── leetcode_metadata2.pkl    # Solution metadata
│           └── leetcode_solutions/       # JSON solution files
├── templates/                      # HTML templates for Flask
└── tests/                          # Test files

​
Troubleshooting

This means FAISS wasn’t installed correctly. Try:
pip install faiss-cpu --force-reinstall
If you have a GPU and want to use FAISS-GPU:
pip uninstall faiss-cpu
pip install faiss-gpu
The HNSW index file is missing. Make sure you:
  1. Cloned the complete repository (not just downloaded specific files)
  2. Didn’t delete or move files from src/DSAAssistant/components/
The index file should be included in the repository.
This usually means the pickle file was corrupted or created with a different Python version. Try:
  1. Re-clone the repository to get a fresh copy
  2. Ensure you’re using Python 3.8 or higher
  3. Check that numpy versions match (the pickle was created with numpy >= 1.24.0)
Quest uses PyTorch for the sentence transformer models. If you see CUDA errors:
  • Quest works fine with CPU-only PyTorch
  • To use CPU explicitly: pip install torch --index-url https://download.pytorch.org/whl/cpu
  • For GPU support, install the appropriate CUDA version of PyTorch from pytorch.org

​
Next steps

Now that Quest is installed, you’ll need to set up Ollama and pull the required language models.

Quickstart guide

Set up Ollama, pull models, and run your first query

Build docs developers (and LLMs) love

Get started for freeTalk to us