swizz - Major update – Swizz v0.1.3

🚀 Swizz v0.1.3 – Major update

Turn your experiment logs into beautiful, animated insights—now with Manim animations and seamless W&B integration!

✨ What's New?

• New Plots
  • Added more plots

• Manim Animations

  • Easily animate plots to create polished, engaging visual stories
  • Ideal for presentations, teaching, or showcasing results dynamically

• Weights & Biases (W&B) Integration

  • Pull experiment run data directly into swizz with a single command
  • Instantly plot your run history and metrics without manual exports

🔗 Full Changelog: https://github.com/lars-quaedvlieg/swizz/compare/v0.1.1...v0.1.3

📦 Installation / Upgrade

Install or upgrade swizz easily via pip:

pip install --upgrade swizz

👐 Contribute

Feedback, bug reports, and PRs welcome! Help us keep making swizz better.

- Python
Published by lars-quaedvlieg 9 months ago

swizz - Initial release – Swizz v0.1.1

🎉 Initial release – Swizz v0.1.0

Welcome to the first release of Swizz – a lightweight, publication-ready plotting library tailored for academic machine learning papers. This version includes the first functional building blocks for tables, plots, and figure layouts.

---

✨ Features

• Consistent and minimal styling for paper figures
• Plug-and-play API with curated function interfaces
• Early support for:
  • LaTeX tables with multi-level grouping
  • Many different kinds of plots (check them out in the gallery)
  • Multi-panel layout composition via PlotBlock and render_layout
• Extensive documentation with usage galleries

---

📚 Usage Examples

📊 Table: Grouped LaTeX Table with Multicolumn Headers

Render grouped tables with hierarchical columns and LaTeX formatting.
See full docs with image→

```python from swizz import table import pandas as pd

df = pd.DataFrame(...) # your long-form data

latexcode = table( "groupedmulticollatex", df=df, rowindex="Model", colindex=["Split", "Budget"], groupby="Task", valuecolumn="score", highlight="max" )

print(latex_code) ```

---

📈 Plot: Dual Histogram with Error Bars

Compare distributions from multiple models, with mean + standard error bars.
See full docs with image→

```python from swizz import plot import numpy as np

datamain = [...] # list of arrays, one per seed databaseline1 = [...] # baseline 1 data_baseline2 = [...] # baseline 2

fig, ax = plot( "dualhistogramwitherrorbars", datamain=datamain, databaselinelist=[databaseline1, databaseline2], baselinelabels=["Baseline 1", "Baseline 2"] ) ```

---

🧩 Layout: Compose Figure Panels for Papers

Use layout primitives like PlotBlock to build figure panels from Swizz components.
See full docs with image→

```python from swizz.layouts import render_layout from swizz.layouts.blocks import PlotBlock, Row, Col

plot1 = PlotBlock("multiplestdlines", kwargs={...}) plot2 = PlotBlock("multiplestdlines", kwargs={...}) plot3 = PlotBlock("generalbarplot", kwargs={...})

layout = { Row([ Col([plot1, plot2]), Col([plot3]), ]) }

fig = render_layout(layout) ```

---

📌 Notes

This is an early-stage release. Expect minimal API changes as features stabilize.
Contributions, bug reports, and suggestions are very welcome!

📘 View documentation →

- Python
Published by lars-quaedvlieg 10 months ago