> ## Documentation Index
> Fetch the complete documentation index at: https://mirage-eval.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Getting started

> Set up Mirage and verify the demo path without downloading a model.

## Requirements

* Python and pip
* The runtime packages in `requirements.txt`; install `requirements-dev.txt` to run tests
* For real evaluations, local dataset files and access to the selected model weights

The model wrappers choose CUDA when it is available and otherwise fall back to CPU. Real model inference can require substantial memory and can be slow on CPU.

## Install the environment

From the repository root, create and activate a virtual environment, then install the pinned dependencies:

```bash theme={null}
python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -r requirements.txt -r requirements-dev.txt
```

## Verify the demo path

Run the focused smoke test:

```bash theme={null}
pytest -q tests/test_demo_smoke.py
```

This test creates one temporary COCO-format image and annotation, then exercises the demo's data-loading, prediction, metric, and JSON-output flow with a stub model. It does not download model weights or dataset files.

## Inspect command options

```bash theme={null}
python demo.py --help
```

Before running a real evaluation, prepare the dataset files described in [Datasets](/datasets) and choose a supported [metric and model](/metrics-and-models).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.