Add report building

[nv-alicheng]

What does this PR do?

Adds functionality to build a report like legacy MetricsReporter from the MetricsAggregator service.

Type of change

• Bug fix
• New feature
• Documentation update
• Refactor/cleanup

Related issues

Testing

• Tests added/updated
• All tests pass locally
• Manual testing completed

Checklist

• Code follows project style
• Pre-commit hooks pass
• Documentation updated (if needed)

[github-actions]

MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅

[gemini-code-assist]

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request introduces a new, streamlined reporting system for benchmark metrics, designed to replace the older, more complex MetricsReporter. The new system simplifies metric aggregation and processing by directly utilizing data from a KVStore, avoiding the overhead of SQL databases and UUID tracking. It provides a clear, structured way to generate benchmark reports, offering both human-readable output and JSON serialization, thereby enhancing the overall efficiency and clarity of the metrics infrastructure.

Highlights

• New Report Module: Introduced a new report module for benchmark result summarization, display, and JSON serialization, replacing the legacy MetricsReporter.
• Simplified Metrics Processing: Implemented a compute_summary function for efficient rollup statistics, including percentiles and histograms, directly from metric values, eliminating the need for complex SQL queries and UUID tracking.
• Structured Report Generation: Added a Report dataclass to store summarized benchmark results, including calculated QPS and TPS, and provided methods for human-readable display and JSON output.
• KVStore Integration: Developed a build_report function that constructs Report objects by reading snapshots from the BasicKVStoreReader, ensuring compatibility with both live and final metrics.
• Comprehensive Testing: Included unit tests for the compute_summary function, the Report class, and the build_report functionality to ensure correctness and reliability.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request introduces a new, simplified metrics reporting system. It includes a compute_summary function for statistical rollups, a Report dataclass to hold benchmark results, and a build_report function to construct reports from KVStore data. The changes also include comprehensive unit tests and design documentation. Feedback suggests improving robustness by using zip(..., strict=True) in compute_summary and _display_metric to prevent potential length mismatches, and ensuring consistency in the Report dataclass by having the tps property return 0.0 instead of None when the total output sequence length is zero.

[nvzhihanj]

Review Council — Multi-AI Code Review

Reviewed by: Claude | Depth: standard

Found 2 issues across 2 files:

• 0 high
• 2 medium

#	File	Line	Severity	Category	Summary
1	report_builder.py	40	medium	data-integrity	Float default in _counter risks precision loss for nanosecond timestamps
2	report.py	119	medium	data-integrity	arr.std() uses population std (ddof=0) — sample std (ddof=1) more appropriate

🤖 Generated with Claude Code

[nv-alicheng]

Review Council — Multi-AI Code Review

Reviewed by: Claude (Codex unavailable) | Depth: standard

Found 3 issues across 2 files:

• 2 medium
• 1 low

#	File	Line	Severity	Category	Summary
1	report_builder.py	55	medium	performance	_summarize copies all raw values into numpy for every series metric. At 50k+ QPS over 60s, that's
2	report.py	127	medium	performance	dataclasses.asdict(self) deep-copies every field, then msgspec.json.encode traverses it again. C
3	report.py	222	low	data-integrity	All histogram buckets display as [lo, hi), but np.histogram's last bin is actually [lo, hi] (i

🤖 Generated with Claude Code