Intermediate

W&B Reports

Create interactive, shareable documents that combine live experiment visualizations with narrative context for team communication.

What Are Reports?

W&B Reports are interactive documents that pull live data from your experiments. Unlike static notebooks or slides, Reports update automatically when new runs complete, making them ideal for ongoing project documentation.

Creating Reports Programmatically

Python - Create a report with the API
import wandb
import wandb.apis.reports as wr

report = wr.Report(
    project="my-project",
    title="Model Comparison: ResNet vs EfficientNet",
    description="Comparing architectures on ImageNet subset"
)

# Add text blocks
report.blocks = [
    wr.H1("Experiment Overview"),
    wr.P("This report compares ResNet50 and EfficientNet-B0 "
         "trained on a subset of ImageNet."),

    wr.H2("Training Curves"),
    wr.PanelGrid(
        panels=[
            wr.LinePlot(x="epoch", y=["train_loss", "val_loss"],
                       title="Loss Curves"),
            wr.LinePlot(x="epoch", y=["train_acc", "val_acc"],
                       title="Accuracy Curves"),
        ],
        runsets=[wr.Runset(project="my-project")]
    ),

    wr.H2("Key Findings"),
    wr.UnorderedList([
        "EfficientNet-B0 achieves 2.3% higher accuracy",
        "ResNet50 trains 1.5x faster per epoch",
        "Both models converge by epoch 30",
    ]),
]

report.save()
print(f"Report URL: {report.url}")

Report Block Types

BlockPurposeAPI Class
TextMarkdown headings, paragraphs, listswr.H1, wr.P, wr.UnorderedList
Panel GridCharts and visualizations from runswr.PanelGrid
Line PlotTime-series metricswr.LinePlot
Bar PlotComparative metricswr.BarPlot
Scatter PlotCorrelation analysiswr.ScatterPlot
Code BlockCode snippets with syntax highlightingwr.CodeBlock
LaTeXMathematical equationswr.LaTeX

Report Use Cases

  1. Weekly team updates

    Create a template report that auto-updates with the latest experiment results each week.

  2. Model comparison documents

    Side-by-side visualizations of different architectures, hyperparameters, or datasets.

  3. Experiment postmortems

    Document what worked, what didn't, and why - with live links to the actual runs.

  4. Stakeholder presentations

    Share polished reports with non-technical stakeholders who need to see results without the dashboard.

✅
Pro tip: Reports can be shared via URL with anyone - even people without a W&B account. Use the "Share" button to generate a public link for stakeholders.

Ready to Go Deeper?

Live instructor-led courses from our partners. Affiliate disclosure.

🧠

AI & ML Courses - 30% Off

Live instructor-led AI, machine learning, data science, and cloud courses for working professionals. Use code Limited30 at checkout.

Edureka
📊

DataCamp - AI & Data Science

Hands-on Python, machine learning, and AI courses with interactive exercises and real projects.

DataCamp
🎓

edX - Top AI Courses

University-level AI courses from MIT, Harvard, Stanford. Earn certificates that employers recognize.

edX