GrimoirePlot

GrimoirePlot is a live dashboard of plotly-compatible plots of remote data

Adding to your current project

uv add grimoireplot

Or install from source:

# git clone the repo
cd grimoireplot
uv sync --extra dev

Installation as a tool

uv tool install grimoireplot  # not yet on pypi, will setup ci/cd on github later on

Quick Start

1. Start the Server

grimoireplot serve --host localhost --port 8080

Then open your browser at http://localhost:8080 to see the dashboard.

2. Push Sample Plots (Test the Server)

In another terminal, push some sample plots to verify everything works:

grimoireplot push-samples --host localhost --port 8080

CLI Reference

grimoireplot serve

Start the GrimoirePlot dashboard server.

grimoireplot serve [--host HOST] [--port PORT]

grimoireplot push-samples

Push sample plots to test the server.

grimoireplot push-samples [--host HOST] [--port PORT] [--secret SECRET] [--grimoire-name NAME]

grimoireplot live-test

Test live plot updates by continuously adding datapoints to a line plot.

grimoireplot live-test [--host HOST] [--port PORT] [--secret SECRET] [--grimoire-name NAME] [--interval SECONDS] [--max-points N]

Programmatic Usage

Sending Plots from Python

GrimoirePlot organizes plots in a hierarchy: Grimoire → Chapter → Plot

Synchronous API

import plotly.graph_objects as go
from grimoireplot.client import push_plot_sync

# Create a Plotly figure
fig = go.Figure()
fig.add_trace(go.Scatter(x=[1, 2, 3, 4], y=[10, 11, 12, 13], mode='lines+markers'))
fig.update_layout(title='My Plot')

# Push to the server
response = push_plot_sync(
    grimoire_name="my_experiment",
    chapter_name="training_metrics",
    plot_name="loss_curve",
    fig=fig,
    grimoire_server="http://localhost:8080",
    grimoire_secret="your-secret",
)

Asynchronous API

import asyncio
import plotly.graph_objects as go
from grimoireplot.client import push_plot

async def main():
    fig = go.Figure()
    fig.add_trace(go.Bar(x=['A', 'B', 'C'], y=[20, 14, 23]))
    fig.update_layout(title='Async Plot')

    response = await push_plot(
        grimoire_name="my_experiment",
        chapter_name="results",
        plot_name="bar_chart",
        fig=fig,
        grimoire_server="http://localhost:8080",
        grimoire_secret="your-secret",
    )

asyncio.run(main())

Integration Example: Training Loop

import plotly.graph_objects as go
from grimoireplot.client import push_plot_sync

losses = []

for epoch in range(100):
    loss = train_one_epoch()  # Your training code
    losses.append(loss)

    # Update the plot every 10 epochs
    if epoch % 10 == 0:
        fig = go.Figure()
        fig.add_trace(go.Scatter(y=losses, mode='lines', name='Training Loss'))
        fig.update_layout(title=f'Training Progress (Epoch {epoch})',
                          xaxis_title='Epoch', yaxis_title='Loss')

        push_plot_sync(
            grimoire_name="experiment_001",
            chapter_name="training",
            plot_name="loss",
            fig=fig,
        )

Configuration

GrimoirePlot can be configured via environment variables:

You can also use a .env file in your project directory.

Testing

# you need to install with --extra dev
GRIMOIRE_TEST=true uv run pytest

Concepts

• Grimoire: A collection of related visualizations (e.g., an experiment)
• Chapter: A group of plots within a grimoire (e.g., training metrics, evaluation results)
• Plot: A single Plotly figure

Acknowledgments

GrimoirePlot is inspired by visdom