> ## Documentation Index
> Fetch the complete documentation index at: https://docs.haiqu.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Experiments

> Find, describe and compare your experiments, with their circuits, jobs and metrics

export const Shot = ({name, alt}) => <Frame>
    <img className="block dark:hidden" src={`/images/dashboard/${name}.png`} alt={alt} />
    <img className="hidden dark:block" src={`/images/dashboard/${name}-dark.png`} alt={alt} />
  </Frame>;

An experiment groups everything for one piece of work. Calling `haiqu.init("name")` in the SDK creates an experiment, and every circuit and job you create after that belongs to it. See [Experiment tracking](/core_features/exp_tracking).

## My Experiments

Open **My Experiments** in the sidebar to see all your experiments.

<Shot name="experiments" alt="My Experiments list" />

* **Search** by name, and filter by **Visibility** (Public, Shared, Private) or **Tags**. The filters are saved in the page URL, so you can bookmark or share a filtered view.
* Click a column header to sort by **Created**, **Last activity**, **Circuits** or **Jobs**.
* The icon at the end of each row shows who can see the experiment: public, shared or private. Click it to open the [share dialog](/dashboard/sharing).
* Use **Export to CSV** to download the list. To delete experiments, select their rows and click **Delete (N)**.

<Warning>
  Deleting an experiment also permanently deletes all its circuits, jobs, comments and artifacts.
</Warning>

## The experiment page

Click any experiment to open it. The page has a header, tags, a description, artifacts and metrics, then the experiment's circuits and jobs.

<Shot name="experiment" alt="Experiment page header, tags, description and artifacts" />

### Header and tags

* Click the title to rename the experiment.
* The **Private** / **Shared** / **Public** button opens the [share dialog](/dashboard/sharing).
* **Lab Notebook** opens the notebook this experiment was run from in Haiqu Lab, when one is linked.
* Click **+** next to **Tags** to add a tag. Use tags to filter **My Experiments**.

### Description

The description is a rich-text editor that saves automatically, so you can document what the experiment is for and what you found. Type `/` to open the block menu. You can add:

* headings, lists and tables
* multi-column layouts
* **block** and **inline equations** in LaTeX: type `$$E = mc^2$$` on an empty line, or `$\epsilon_{2q}$` inside a sentence
* images up to 2 MB

To comment on part of the description, select text and click **Add comment**. See [Comments](/dashboard/sharing#comments).

### Artifacts and Logged Metrics

What you log from the SDK appears here. How it is shown depends on what you log:

| You log | It appears as |
| - | - |
| The same numeric metric several times | A **line chart** card. Click it to enlarge |
| A pandas DataFrame | A **Table data** card. Click it to view, or use **Export to CSV** |
| A matplotlib figure or an image | An **image** card. Click it to see it full size |
| A single value (number, text) | A row under **Logged Metrics** |

<Shot name="experiment-artifacts" alt="Artifacts: an image, a line chart and a table" />

<Shot name="experiment-table-artifact" alt="Table artifact opened in a dialog" />

* Click **+** next to **Artifacts** to upload a file of up to 2 MB.
* Use the **…** menu on a card to **Rename** or **Delete** it. On table cards, **Insert into description** adds the table to the description.
* Click a metric name under **Logged Metrics** to rename it. Hover over a metric to delete it.

### Experiment Circuits

This table lists every circuit in the experiment: circuits you logged, plus transpiled and compressed versions. Each derived circuit appears under the circuit it came from.

<Shot name="experiment-circuits" alt="Experiment Circuits table and Quantum volume chart" />

* The columns show **Source**, target **Device**, **Estimated Fidelity**, **Active Qubits**, **Depth** and **Gates** (2-qubit and total). For depth and gate counts, lower is better: shorter circuits pick up less noise.
* **Estimated Fidelity** (0–100%, higher is better) is an estimate made before any run; after a run, the closest measured counterpart is the job's **QPU Quality** (see [Jobs](/dashboard/jobs#metrics-in-the-table)). Its meaning depends on the source:
  * **Transpiled**: expected fidelity on the target device, from the device's noise characteristics.
  * **Compressed**: how closely the compressed circuit's state matches the original's.
  * **Data loading**: how closely the prepared state matches your data.
  * Circuits you logged yourself show 100%: there is no device or approximation to compare against yet.
* Filter by name, **Source** or **Device**. Click a circuit name or description to edit it.
* Use the row actions to show the jobs that used this circuit, or open its [details page](/dashboard/circuits).

**Quantum volume** plots the selected circuits by 2-qubit depth and active qubits, colored by estimated fidelity. Pick a device in **Select device** to overlay its quantum volume (the shaded box) and an approximate volumetric frontier. Circuits inside the box or under the frontier are likely to run well on that device; the further a circuit sits beyond the frontier, the more its results degrade. Use the checkboxes in the table to choose which circuits are plotted.

<Shot name="experiment-quantum-volume" alt="Quantum volume chart" />

### Experiment Jobs

This list shows the experiment's jobs. It has the same columns and actions as [All Jobs](/dashboard/jobs). **Job Comparison** plots the selected jobs by QPU quality against **CPU Time** or **QPU Cost**.

<Shot name="experiment-jobs" alt="Experiment Jobs list" />


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