[DOCS-1620] Improve accessibility of images #1429
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…lease notes - Converted 6 HTML <img /> tags to Markdown format in release notes - Enhanced alt text descriptions based on surrounding context - Limited alt text to approximately 20 words for better accessibility - Fixed typo in 0.33.0 release notes (composotion -> composition) Files updated: - ref/release-notes/releases/0.63.0.md - ref/release-notes/releases/archived/0.39.0.md - ref/release-notes/releases/archived/0.34.0.md - ref/release-notes/releases/archived/0.33.0.md - ref/release-notes/releases/archived/0.38.0.md - ref/release-notes/releases/archived/0.36.0.md
…roved alt text - Convert local images to Hugo img shortcode format - Convert external images to Markdown format - Improved alt text for tutorial image with descriptive context - Maintained all image attributes in conversions Files updated: - _index.md: 2 local images converted to Hugo shortcodes - guides/models/app/settings-page/user-settings.md: 1 external image converted to Markdown - tutorials/weave_models_registry.md: 1 local image converted to Hugo shortcode with improved alt text
…ility - Updated 50+ Hugo img shortcodes with empty or missing alt attributes - Added contextual alt text descriptions limited to ~20 words each - Improved accessibility for screen readers across tutorials, guides, and integrations - Covers key areas: tutorials (sweeps, artifacts, experiments), integrations (HuggingFace, PyTorch, TensorFlow, OpenAI, etc.), and core features Categories updated: - Tutorial images: sweeps, artifacts, experiments, integration tutorials - Integration guides: HuggingFace, PyTorch, TensorFlow, OpenAI, Cohere, TorchTune, YOLOX, etc. - Architecture diagrams: W&B platform, Models, Core components - Dashboard visualizations: training metrics, hyperparameter optimization, model comparisons Still in progress - more images remaining to update.
…nd integrations - Updated 30+ additional Hugo img shortcodes with descriptive alt text - Enhanced accessibility for registry, reports, launch, hosting, and SSO features - Added contextual descriptions for scientific visualizations and configuration interfaces Key areas improved: - Registry features: collection creation, visibility settings, artifact types - Launch system: queue configuration, job management, AWS/GCP setup - Reports: creation workflow, collaboration features, visualization - Hosting/SSO: authentication setup, team management, access controls - Integration visualizations: scikit-learn metrics, TensorBoard sync, model comparisons Progress: ~80+ Hugo shortcodes improved total Remaining: Continue with remaining empty alt attributes
- Updated 25+ Hugo img shortcodes across registry and model management features - Improved accessibility for lineage visualization, tagging, search, and model lifecycle management - Added descriptive alt text for user interface demonstrations and workflow visualizations Key areas covered: - Registry lineage: artifact dependencies, node expansion, detailed views - Tag management: collection tagging, search functionality, organization tools - Model Registry: model versions, linking workflows, lifecycle management - Access control: user permissions, team settings, role management - Search and discovery: tag-based filtering, query interfaces Progress: ~105+ Hugo shortcodes improved total Focus: Critical user workflows and interface demonstrations
Deploying docs with
|
| Latest commit: |
0eb92e3
|
| Status: | ✅ Deploy successful! |
| Preview URL: | https://7876deb3.docodile.pages.dev |
| Branch Preview URL: | https://docs-1620.docodile.pages.dev |
mdlinville
commented
Jul 1, 2025
Address review feedback by simplifying verbose alt text descriptions: - Remove detailed explanations after 'showing', 'displaying', etc. - Keep essential identification without excessive details - Standardize to shorter, more focused descriptions - Examples: * 'TorchTune training dashboard showing model fine-tuning progress with loss curves and system metrics' → 'TorchTune training dashboard' * 'Benefits of using W&B: unified dashboard, lightweight integration, accessibility, security, and industry trust' → 'Benefits of using W&B' * 'W&B platform architecture diagram showing Models, Weave, and Core components for AI development' → 'W&B platform architecture diagram' Applied systematically across ~40 files including tutorials, integrations, and core documentation.
Address additional review feedback by making alt text even more concise:
- Remove redundant descriptive phrases ('showing', 'displaying', 'featuring')
- Eliminate unnecessary UI details ('bottom right corner', 'top left')
- Simplify technical descriptions to essential identification
- Follow accessibility best practices from A11Y Project and Vision Australia
Examples:
* 'Overview page of a run logged to W&B. Bottom right corner of UI shows...' → 'Run overview with summary metrics'
* 'OpenAI fine-tuning auto-scan feature showing automatic detection and tracking...' → 'OpenAI auto-scan feature'
* 'Scikit-learn ROC curve visualization showing classifier performance with area under curve metrics' → 'Scikit-learn ROC curve'
Updated 35+ files including integrations, tutorials, registry, hosting, and core docs.
Alt text now focuses on essential context without verbose descriptions.
Applied feedback to make alt text much more concise by: - Removing verbose phrases like 'demonstration of', 'showing', 'displaying' - Focusing on essential identification rather than descriptive actions - Shortening to 3-5 words when possible while maintaining clarity Updated sections: - Reports: embedding workflows, cloning, editing features, gallery examples - Model Registry: access controls, creation workflows, organization features Examples: - 'Step-by-step demonstration of creating a Docker queue' → 'Docker queue creation' - 'Registry collection view showing selected tags' → 'Registry collection with selected tags' - 'Demonstration of linking a model version to the Model Registry' → 'Linking model version to registry'
Continued applying concise alt text approach across core user features: Integration examples: - Scikit-learn visualizations: 'Scikit-learn confusion matrix', 'Scikit-learn elbow plot' - Azure OpenAI: 'Azure OpenAI fine-tuning metrics' - Prodigy: 'Prodigy annotation table' - DeepChem: 'DeepChem molecular analysis' Core tracking features: - Experiments: 'Experiments dashboard' - Tables: 'Semantic segmentation predictions table' - Jupyter: 'Jupyter W&B widget' Project workspace: - Workspace panels: 'Project workspace with panels', 'Adding workspace section' - Runs management: 'Runs table', 'Filtering runs by validation loss' - Artifact panels: 'Artifact overview panel', 'Artifact lineage visualization' Artifacts workflows: - 'Single vs distributed artifact workflows' - 'Incremental artifact versioning' Total: ~30 more images improved with 3-5 word concise descriptions
Applied concise alt text approach to key user-facing functionality: Core UI & Tracking: - Runs tracking: 'Single run workspace', 'Training run with logged metrics', 'Run ID location' - Workspace settings: 'Workspace settings gear icon', 'Workspace layout options', 'Line plot settings' - Panel management: 'Section settings menu', 'Panel edit icon', 'Panel settings modal' - Charts: 'Customized bar plot' Artifact TTL Management: - 'Setting TTL permissions', 'Editing TTL policy', 'Setting default TTL policy' - 'Removing TTL policy', 'TTL collection view' Support KB Articles: - Offline sync: 'Offline mode terminal output', 'Cloud sync terminal output' - Reports: 'Adding multiple report authors', 'Adding image to report', 'Embedding reports' - UI: 'Anonymous mode interface', 'Deleting chart preset' - Sweeps: 'Estimated runs column', 'Sweep run estimation' Total: ~25 more images improved with clear, concise descriptions focused on essential identification
mdlinville
commented
Jul 1, 2025
- Apply user feedback to remove verbose descriptions - Remove redundant phrases like 'showing', 'interface', 'example of' - Simplify to essential identification: 'Weave UI interface showing...' → 'Weave UI' - Keep alt text focused and concise per accessibility best practices - Improved ~25 more alt text descriptions across tutorials, integrations, and core features
- Launch queue observability: job status, queued time, GPU metrics, error logs - Add job to queue: navigation, configuration, templates - Parameter importance: feature analysis, correlation plots - Custom charts: PR/ROC curves, line plots, histograms, scatter plots - Line plot panels: visualization options, legend settings, zoom functionality - Code panels: comparer, Jupyter session history - Run management: stop controls, workspace tabs, artifacts, logs, files - Project management: starring, deletion workflows - Tables and distributed training visualizations - Launch job inputs and schema forms Applied concise alt text methodology: 'Job status timeline', 'GPU usage metrics', 'Custom line plot' ~40 more critical workflow images improved
…anels, and settings - SSO authentication: Added concise alt text for Okta, Entra (Azure AD), and AWS Cognito setup workflows (~22 images) - Slack alerts: Improved alt text for enterprise Slack integration setup - Query panels: Enhanced alt text for Weave data visualization features (~18 images) - Embedding projector: Added descriptive alt text for 2D/3D visualization features - Team/email settings: Improved alt text for enterprise team management workflows Applied accessibility best practices with 3-5 word descriptions focusing on essential identification for screen readers.
…flows - Run management: Enhanced alt text for grouping, filtering, workspaces, alerts (~15 images) - Data visualization: Improved plots (line, scatter, bar, histogram, PR/ROC curves, confusion matrix) (~12 images) - Tables: Added descriptive alt text for table logging and visualization features (~4 images) - Sweeps: Enhanced alt text for hyperparameter optimization workflows (~5 images) - Custom charts: Improved alt text for chart creation and configuration (~6 images) - Model registry: Added alt text for access controls and promotion workflows (~2 images) Applied concise 3-5 word descriptions focusing on essential functionality for accessibility.
- Model registry: Enhanced alt text for model consumption, notifications, concepts, linking, model cards (~8 images) - Reports: Improved alt text for report creation and collaboration features (~2 images) - Line plots: Added descriptive alt text for plot styling and sampling features (~5 images) - Media logging: Enhanced alt text for bounding boxes and molecule visualizations (~2 images) - Embedding projector: Fixed duplicate and improved MNIST digits projection alt text Final result: 300+ images improved with concise, accessibility-focused alt text following screen reader best practices.
…d descriptions Based on user feedback preferring 'Create an empty collection' over verbose descriptions: - 'AWS CloudFormation template example for advanced Launch queue configuration with resource specifications' → 'AWS CloudFormation template' - 'Interactive walkthrough of creating a new collection within a registry with metadata configuration' → 'Create a new collection' - 'Line plot panel with numerous semi linear lines depicting system metrics logged by a multi node process' → 'Multi-node system metrics' - 'Console logs from multiple nodes with the x_label filter applied' → 'Multi-node console logs' - 'Line plot with color coded runs based on a specific metric' → 'Color coded runs plot' - 'Query panel with three W&B runs and the values they logged at step 295 using the step slider feature' → 'Step slider feature' - 'Collection card with information about model architecture, intended use, performance information and more' → 'Collection card' - 'Draw insights from large hyperparameter tuning experiments with interactive dashboards' → 'Hyperparameter tuning insights' Applied principle: Focus on core action/purpose, remove verbose descriptive language.
- Replace verbose descriptions with concise 2-4 word phrases - Remove unnecessary 'User typing...', 'showing...', 'displaying...' patterns - Focus on core functionality rather than descriptive elaboration - Examples: 'User typing text into registry search bar...' → 'Searching within a Registry' - Apply feedback systematically across ~25 files
Following user feedback pattern: 'Adding tags to registry collection for improved organization and searchability' → 'Adding tags to a Registry collection' Applied to ~15 more files: - Remove 'for [purpose]', 'performing [action]', explanatory clauses - Simplify 'X vs Y' patterns to 'X comparison' - Remove redundant 'visualization', 'Image of' prefixes - Focus on core functionality: 'GAN discriminator gradients', 'Registry lineage', 'Launch UI' - Make alt text 2-4 words maximum where possible
Applied user feedback pattern systematically to ~15 more files: - Remove explanatory clauses: 'Add a new custom chart, then edit the query' → 'Custom chart creation' - Simplify descriptions: 'See granular differences: the left model detects...' → 'Model comparison' - Remove verbose explanations: 'If regex doesn't provide...' → 'Tag filtering' - Focus on core action: 'Interactive mask viewing in the W&B UI' → 'Interactive mask viewing' - 2-4 word maximum where possible, essential functionality only
johndmulhausen
approved these changes
Jul 2, 2025
ngrayluna
approved these changes
Jul 2, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
DOCS-1620 Improve accessibility of images
I worked with Claude to:
Though in fact I had it author the alt text for each group of conversions as it went.
Ready for review.