odh-34800: doc updates for model catalog performance data #1039
Conversation
WalkthroughReordered and expanded four AsciiDoc modules to make the Model Catalog primary: added UI descriptions (categories, search, labeled filters), Performance Insights for validated models, updated discovery/registration/deployment procedures, clarified deployment naming constraints, and adjusted cross-references and conditional blocks. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
participant User
participant UI as Model Catalog UI
participant Catalog as Model Catalog Service
participant Registry as Model Registry
participant Deployer as Deployment Service
Note over UI: Discover & evaluate models
User->>UI: Open catalog, choose category (All / Org / Validated / Community)
UI->>Catalog: Query models (search, filters, labels)
Catalog-->>UI: Return model list (metadata, badges)
alt View model details
User->>UI: Open model details
UI->>Catalog: Fetch metadata & benchmarks
Catalog-->>UI: Return details + Performance Insights
UI->>User: Display details & Performance Insights tab
end
alt Deploy model
User->>UI: Click Deploy
UI->>Registry: (optional) register or query model resource info
Registry-->>UI: Return resource name constraints
UI->>Deployer: Start deployment (project, deployment name, options)
Deployer-->>UI: Return deployment status
UI->>User: Show deployment result
end
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes
Pre-merge checks and finishing touches✅ Passed checks (3 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (1)
modules/viewing-models-in-the-catalog.adoc (1)
38-56: Performance Insights section is comprehensive but complex.The Performance Insights section (lines 38–56) provides detailed performance benchmark guidance for Red Hat AI validated models, covering workload types, latency metrics, percentiles, sliders, and hardware types. While thorough, the heavily nested bullet structure with four separate filter options and their sub-steps may be challenging to follow. Consider simplifying the organization or adding a brief introductory note to help users navigate the multiple filtering options.
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (4)
modules/deploying-a-model-from-the-model-catalog.adoc(2 hunks)modules/overview-of-model-registries.adoc(1 hunks)modules/registering-a-model-from-the-model-catalog.adoc(1 hunks)modules/viewing-models-in-the-catalog.adoc(2 hunks)
🔇 Additional comments (5)
modules/deploying-a-model-from-the-model-catalog.adoc (2)
35-37: Aligned with catalog-first workflow.The added Model Catalog overview and search/filter descriptions are consistent with the parallel guidance in other catalog modules and properly emphasize the UI-driven discovery approach.
46-57: Helpful clarification on deployment naming.The NOTE block clearly distinguishes between the Model deployment name (inference service name) and Resource name (OpenShift label), with detailed constraints on resource naming conventions. This guidance should help users avoid configuration errors.
modules/overview-of-model-registries.adoc (2)
4-4: Well-executed restructuring for catalog-first narrative.The reordering of sections to place the Model Catalog before the Model Registry, combined with the expanded abstract and updated cross-references, successfully establishes the catalog as the primary entry point for data scientists. The Model Registry is appropriately positioned as a supporting governance and metadata management component. This structure aligns well with the overall PR objective.
Also applies to: 13-13, 29-29
15-19: Catalog description effectively communicates breadth and quality.Line 17 clearly explains the benchmarking value (third-party models benchmarked for performance/quality) and hardware configuration comparison capability. Line 19's list of provider examples (Red Hat, IBM, Meta, Nvidia, Mistral AI, Google) helps users understand the diversity of available models.
modules/registering-a-model-from-the-model-catalog.adoc (1)
15-17: Consistent with parallel guidance in other modules.The Model Catalog overview and search/filter descriptions properly align with the catalog-first approach established across the documentation suite.
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (1)
modules/viewing-models-in-the-catalog.adoc (1)
52-52: Clarify available percentile options.Line 52 references "a percentile value, for example,
P90" but doesn't list which percentile values are available (e.g., Mean, P90, P95, P99). Without explicit options, users may be uncertain what values they can select.Consider replacing this with a clearer list similar to the metrics options above:
-You can also select a percentile value, for example, `P90`. +You can select a percentile value from the list: `Mean`, `P90`, `P95`, or `P99`.
|
except for couple places calling in "data scientists" explicitly as it also applies to "AI engineers" or Platform engineers the rest is LGTM |
3 participants
HTML rendered preview:
Summary by CodeRabbit