Skip to content

style: apply /style-guide pass to models/runs#2724

Open
johndmulhausen wants to merge 5 commits into
mainfrom
style-guide/models-runs-20260604-113608
Open

style: apply /style-guide pass to models/runs#2724
johndmulhausen wants to merge 5 commits into
mainfrom
style-guide/models-runs-20260604-113608

Conversation

@johndmulhausen

@johndmulhausen johndmulhausen commented Jun 4, 2026

Copy link
Copy Markdown
Contributor

Summary

This PR applies the /style-guide skill (Google Developer Style Guide + CoreWeave conventions) to 20 files in models/runs. The run was automated; no technical content was intentionally changed.

Files edited

  • models/runs/alert.mdx
  • models/runs/color-code-runs.mdx
  • models/runs/compare-runs.mdx
  • models/runs/customize-run-display.mdx
  • models/runs/delete-runs.mdx
  • models/runs/filter-runs.mdx
  • models/runs/forking.mdx
  • models/runs/grouping.mdx
  • models/runs/infrastructure-alerts.mdx
  • models/runs/initialize-run.mdx
  • models/runs/manage-runs.mdx
  • models/runs/resuming.mdx
  • models/runs/rewind.mdx
  • models/runs/run-colors.mdx
  • models/runs/run-identifiers.mdx
  • models/runs/run-states.mdx
  • models/runs/search-runs.mdx
  • models/runs/stop-runs.mdx
  • models/runs/tags.mdx
  • models/runs/view-logged-runs.mdx

Recommendations for technical review

Prerequisites

  • Add a Prerequisites section to alert.mdx listing required wandb SDK version, W&B account, and Slack workspace permissions needed to authorize the W&B Slack app.
  • delete-runs.mdx: document required permissions for deleting runs (project admin, member, team-level role) in the W&B App procedure, and consider linking to API authentication setup for the Python tab.
  • infrastructure-alerts.mdx: define what constitutes a "supported CKS cluster" (region, node type, GPU model, Mission Control enablement) and document the enablement path (automatic, admin toggle, CoreWeave-side configuration, or Preview gating).
  • initialize-run.mdx: link install/upgrade instructions for the v0.19.10+ requirement called out in the Requirements note.
  • manage-runs.mdx: clarify which role/permission grants the "access to the run at its current and new locations" required to move runs across teams.
  • resuming.mdx: verify the relative anchor link ./#custom-run-ids (line 102) resolves to the intended section.
  • rewind.mdx: add guidance on how to check or upgrade the installed SDK version for the >=0.17.1 requirement.
  • run-identifiers.mdx: consider stating that readers must have the SDK installed and an active W&B account before the SDK examples.
  • view-logged-runs.mdx: note any minimum CLI version or install/setup link for wandb beta leet; clarify whether code logging is opt-in (e.g., wandb.Run.log_code()); link to per-run Git enablement instructions.

Verification steps

  • alert.mdx: add a minimal, always-fires example for end-to-end verification of run.alert(), and add troubleshooting guidance for "I enabled alerts but nothing arrives."
  • color-code-runs.mdx: consider whether a brief description of the workspace immediately after enabling key-based colors would help readers confirm success.
  • customize-run-display.mdx: add confirmation sentences after the Add columns procedure (steps 3 and 5) and after the Sort runs procedure.
  • delete-runs.mdx: add verification guidance after the UI procedure and a verification command after the Python loop.
  • initialize-run.mdx: add sample output or a workspace screenshot for the multi-run/concurrent example.
  • infrastructure-alerts.mdx: add a short "what you'll see" subsection (annotated run plot, alert types, workspace location).
  • manage-runs.mdx: add a verification cue for successful run moves across teams/projects.
  • resuming.mdx: add guidance for confirming a run actually resumed (UI cue or stdout).
  • rewind.mdx: consider adding a verification step after each rewind code example.
  • run-identifiers.mdx: add expected-outcome text after each procedure (custom run ID/name, rename a run, rename display name).
  • run-states.mdx: add a short sample output block to the Python SDK example.
  • search-runs.mdx: add an expected-outcome sentence after a successful search and after disabling RegEx.
  • stop-runs.mdx: state the resulting run state in the Python SDK tab so it parallels the W&B App tab.
  • view-logged-runs.mdx: add an "On success, LEET opens with…" sentence; add an outcome sentence at the end of the W&B App procedure.

Technical accuracy

  • alert.mdx: confirm wait_duration=300 units and that 300s matches the "no more than once every 5 minutes" claim; document the full set of AlertLevel values and per-level delivery/formatting behavior; verify Slack mention syntax <@U1234ABCD> and clarify behavior in email; confirm the wandb.ai/teams/[YOUR-TEAM] path and the alert types configurable on team settings; confirm Dedicated Cloud coverage for the Multi-tenant Cloud caveat.
  • color-code-runs.mdx: verify the Latest description phrasing versus Max/Min; confirm UI casing of Latest; verify the Key dropdown displays "Accuracy" vs. "acc"; confirm the exact Buckets UI label; verify the screenshot's correct bucket placement for accuracy 0.8232 (Bucket 5 vs. Bucket 6); confirm new alt-text matches the screenshots.
  • compare-runs.mdx (line 108): the parenthetical defining "underperforms" currently duplicates "outperforms" — should read "(is directionally worse than) the baseline".
  • customize-run-display.mdx: confirm both alternative paths for pinning/unpinning columns are valid; confirm whether the Note on lines 68-70 duplicates the section intro; confirm the 1,000-run display limit in the commented-out note on lines 7-9; clarify whether Show latest appears only for aggregated/metric columns.
  • delete-runs.mdx: verify the "starting a new run with a deleted run's ID fails with an error" claim across UI/API and Cloud/Self-Managed; verify the Warning callout against the actual UI flow (Registry-specific confirm prompt); verify the Mermaid flowchart's Registry warning branch and "Production aliases affected" detail; confirm GORILLA_DATA_RETENTION_PERIOD is current and documented; confirm "W&B" as the actor in line 17; consider precision for "background jobs catch up"; verify the "Public API" reference in description/keywords; reconcile "self-managed deployments" vs. "W&B Self-Managed" usage.
  • filter-runs.mdx: verify the literal UI string for **Show only my works** (likely "Show only my work" or "Show only my runs"); confirm whether steps 3 and 4 of the Tags example are distinct or merged in error; normalize **New Filter** vs. **New filter** casing; confirm operator casing in the table (is/is not vs. IN/NOT IN); reconcile state = "finished" vs. State = finished value rendering; confirm whether the close affordance is a literal x glyph.
  • forking.mdx: confirm the Warning's "active development and in private preview" status; verify the wandb SDK 0.16.5+ minimum; confirm the live demo link still resolves and is appropriate.
  • grouping.mdx: verify the "Organize runs into groups" step ordering (lines 55-63); clarify whether step 4 of "Remove runs from a group" removes runs or deletes the group; clarify whether "select a Group" means a literal label or a specific group name; reconcile "Move runs between groups" with "Move to group"; verify wandb.Run.group and wandb.Run.job_type as the current public SDK property names; confirm that removing all runs auto-deletes the group; clarify whether grouping is mutable via SDK after wandb.init().
  • infrastructure-alerts.mdx: confirm 0.20.1 as the minimum SDK version; confirm 0.73.0 as the minimum W&B Server version and the exact SERVER_FLAG_ENABLE_CORE_WEAVE_OBSERVABILITY env var name (including CORE_WEAVE vs. COREWEAVE casing); verify the three runtime behaviors on lines 25-27 are accurate and complete.
  • initialize-run.mdx: confirm wandb.setup() is a context manager that binds run correctly; fix the evaluate_loss_accuracy return annotation -> (float, float) to valid typing syntax (tuple[float, float]); reconcile concurrent-example variable names (run_a/run_b1 vs. tracking_run); confirm the 6-space indentation in the inner with wandb.init() block is intentional.
  • manage-runs.mdx: verify runs are truly removed from the original project (no tombstone/redirect); confirm wandb artifact get/put and Api.artifact are still recommended for artifact migration; confirm whether moving a run preserves run history, system metrics, logged media, and tags; decide whether the title should be pluralized; name the destination team/project dropdown explicitly; verify the Move to project button label.
  • resuming.mdx: the first resume procedure's code is missing the id="[RUN-ID]" argument expected by the prose; verify the "~5-minute crash detection window" on line 122; confirm whether W&B's agent specifically expects 128 + signum exit codes; verify that wandb.agent() runs the training function in a thread; confirm sweep agents drain the run queue before generating new combinations; normalize the Markdown separator row mixing --- and --.
  • rewind.mdx: enumerate the "known performance limitations with rewind"; confirm the private preview status and support@wandb.com contact are current; confirm whether entity and project in wandb.init() must match the original run for resume_from; confirm tense changes "produces" → "produced" and "created" → "creates" still describe runtime behavior accurately.
  • run-colors.mdx: confirm that hovering over the Name column header is required to surface the action menu (vs. visible by default); verify the randomize gating depends only on sort/filter/search/group modifications and not also on minimum run count.
  • run-identifiers.mdx: verify the entity="[ENTITY]" placeholder in Create a custom run ID (the original used entity="<project>", which appears to be a typo); confirm the same parameter labeling in Create a custom run name; confirm the /images/runs/unique-run-id.png image is accurate for both run ID and run name sections (or split into two images); confirm api.run(path = "...") spacing matches project Python style.
  • run-states.mdx: confirm Pending is exposed exactly as Pending (capitalized) via run.state from the Python SDK.
  • search-runs.mdx: confirm the inactive RegEx toggle has a non-color visual cue (dimmed icon, unfilled state) so the new "appears inactive" wording matches actual UI.
  • stop-runs.mdx: verify the casing of state labels ("from running to Killed") in the W&B App and align with /models/runs/run-states.
  • tags.mdx: confirm whether wandb.Run.tags is a tuple, a list, or both, and align the prose with the example that uses .append(); confirm the + icon and x affordance still match the current UI.
  • view-logged-runs.mdx: verify the nested-config search workaround (whether dot-notation or another pattern works); confirm Command and Python executable are meaningfully distinct fields; verify all new alt-text matches the screenshots; confirm marketing-style claims about Tracked Hours, Runtime, and Git state are still accurate.

Missing content

  • alert.mdx: add a reference table for run.alert() parameters (title, text, level, wait_duration); explain AlertLevel enum members and per-level behavior; clarify whether @ mentions work in email or only in Slack.
  • color-code-runs.mdx: consider whether the Example: key-based colors with a loss metric section should move after the explanatory subsections for better IA.
  • compare-runs.mdx: resolve the stale {/* TODO screenshot */} author marker on line 213; consider whether steps 2 and 3 of the example workflow (lines 144-211) need confirmation sentences; verify the PinRunsCondensed snippet provides any "why" framing needed; clarify standard W&B terms (run selector, runs table, workspace, run set) inline or via cross-reference; consolidate or confirm the repeated "lines for the baseline run appear bolder" sentence (lines 40, 84, 90); review 2-space vs. 4-space nested list indentation on lines 209-210.
  • customize-run-display.mdx: clarify whether the Hide columns "View all hidden columns" line belongs there, in Add columns, or in its own subsection.
  • delete-runs.mdx: document bulk-deletion limits, behavior for sweep/group runs and runs with active alerts, whether run.delete() is reversible, and behavior in team/org-locked projects; note any support-side recovery path for the "Cannot be recovered" flowchart node.
  • filter-runs.mdx: add Prerequisites/permissions section; document default filter persistence across sessions/projects/views; fix the broken anchor link #filter-runs-with-tags (target generates #example-filter-runs-with-tags); decide whether to restore or remove the commented-out sections (lines 93-132); fix the "AND OR conditions" frontmatter keyword.
  • forking.mdx: align the example script comment with the loop's actual range; harmonize the fork_step = int("[NUM]") vs. fork_step = 200 placeholder pattern; consider whether the <Tip> after the second code block duplicates the preceding paragraph.
  • grouping.mdx: add prose explaining what to replace [ENTITY] and [PROJECT] with before each code block; decide whether to restore or remove the commented-out WANDB_RUN_GROUP/wandb.util.generate_id() block.
  • infrastructure-alerts.mdx: link to where/how to set env vars on the W&B app pod for the Self-Managed instructions.
  • initialize-run.mdx: fill in the empty "Example use case" cell for return_previous in the reinit table; add a header to the first column of the reinit table; consider a brief note on reinit interaction with wandb.finish() in notebooks; consider renaming the Example: Concurrent processes heading to a noun phrase.
  • manage-runs.mdx: decide whether to restore or remove the large commented-out "Move runs to a group" and "Manage runs in bulk" sections; document what happens to references from Reports, Sweeps, or Artifacts when a run moves; document whether the move is reversible.
  • resuming.mdx: fill in the empty "Use case" cell for the "never" row in the resume-behavior table; show how the preemption signal is delivered in non-SLURM environments; normalize placeholder style (bracketed vs. literal paths).
  • rewind.mdx: review the commented-out block at lines 27-30 referring to "Run archiving"; define what an "archived resumption" is in concrete terms.
  • run-colors.mdx: remove or relocate the stale {/* … */} comment on line 35 that duplicates the visible Note.
  • run-identifiers.mdx: fix the broken anchor #autogenerated-run-ids (line 11) and #overview-tab (line 82); clarify which page/view (Workspace tab vs. Runs tab) the "action menu at the top of the list of runs" refers to; decide whether to restore or remove the commented-out <Tabs> block describing renaming runs from a report.
  • run-states.mdx: consider adding a brief note on state transitions (e.g., Crashed → Finished after resync, Killed vs. Failed).
  • search-runs.mdx: consider linking "Runs table" and "Workspace" in the intro to their concept pages.
  • stop-runs.mdx: consider noting whether Ctrl+D is the only SDK stop signal across platforms (macOS terminals, Jupyter notebooks, IDEs).
  • tags.mdx: document SDK/Public API tag removal for parity with the "Add tags" coverage.
  • view-logged-runs.mdx: normalize UI label casing (Title Case vs. sentence case) across Overview fields; consider whether the in-page anchor links in the intro sentence (line 27) add value.

How to review

  • Each file's changes are style edits only. Compare side-by-side and flag any that change technical meaning.
  • Approve and merge to accept the edits, or close to reject them.

@johndmulhausen johndmulhausen requested a review from a team as a code owner June 4, 2026 15:37
@mintlify

mintlify Bot commented Jun 4, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

Preview deployment for your docs. Learn more about Mintlify Previews.

Project Status Preview Updated (UTC)
wandb 🟢 Ready View Preview Jun 4, 2026, 3:45 PM

@github-actions

github-actions Bot commented Jun 4, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

📚 Mintlify Preview Links

🔗 View Full Preview

📝 Changed (21 total)

📄 Pages (21)

File Preview
models/runs.mdx Runs
models/runs/alert.mdx Alert
models/runs/color-code-runs.mdx Color Code Runs
models/runs/compare-runs.mdx Compare Runs
models/runs/customize-run-display.mdx Customize Run Display
models/runs/delete-runs.mdx Delete Runs
models/runs/filter-runs.mdx Filter Runs
models/runs/forking.mdx Forking
models/runs/grouping.mdx Grouping
models/runs/infrastructure-alerts.mdx Infrastructure Alerts
... and 11 more files

🤖 Generated automatically when Mintlify deployment succeeds
📍 Deployment: 32ce07b at 2026-06-10 17:04:59 UTC

@github-actions

github-actions Bot commented Jun 4, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

🔗 Link Checker Results

✅ All links are valid!

No broken links were detected.

Checked against: https://wb-21fd5541-style-guide-models-runs-20260604-113608.mintlify.app

@johndmulhausen johndmulhausen force-pushed the style-guide/models-runs-20260604-113608 branch from 890eaac to 765929f Compare June 5, 2026 17:19
@johndmulhausen johndmulhausen force-pushed the style-guide/models-runs-20260604-113608 branch from 765929f to 1df3e6e Compare June 5, 2026 17:28
@johndmulhausen johndmulhausen requested a review from Copilot June 5, 2026 18:45
Copilot AI reviewed Jun 5, 2026
View reviewed changes

Copilot AI left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

This PR applies an automated style-guide pass across the models/runs documentation set, primarily standardizing wording, placeholders, and adding frontmatter keywords to improve search/discoverability.

Changes:

  • Added/updated frontmatter metadata (keywords) and standardized placeholder formats (for example [ENTITY], [PROJECT], [RUN-ID]).
  • Refined headings, lists, and procedural steps for consistency with the style guide.
  • Introduced some new/updated cross-references and URL templates within the runs documentation.

Reviewed changes

Copilot reviewed 20 out of 20 changed files in this pull request and generated 14 comments.

Show a summary per file
File Description
models/runs/alert.mdx Style/wording updates and added keywords for scriptable alerts documentation.
models/runs/color-code-runs.mdx Reorganized/clarified key-based colors instructions; added keywords.
models/runs/compare-runs.mdx Wording and accessibility text adjustments for baseline/pinned run comparison docs.
models/runs/customize-run-display.mdx Expanded procedural clarity for runs table customization; added keywords.
models/runs/delete-runs.mdx Clarified deletion behavior/retention wording; added keywords.
models/runs/filter-runs.mdx Reworded filter instructions/operators table; added keywords.
models/runs/forking.mdx Style updates and expanded explanations; added keywords.
models/runs/grouping.mdx Wording/step formatting updates for grouping/job type; added keywords.
models/runs/infrastructure-alerts.mdx Improved prerequisite wording and formatting; added keywords.
models/runs/initialize-run.mdx Expanded guidance for reinit / multiple runs; added keywords.
models/runs/manage-runs.mdx Clarified run move workflow and outcomes; added keywords.
models/runs/resuming.mdx Updated wording and placeholders for resume behaviors; added keywords.
models/runs/rewind.mdx Clarified rewind limitations/workflow; added keywords.
models/runs/run-colors.mdx Style updates for run color customization/randomize/reset guidance; added keywords.
models/runs/run-identifiers.mdx Updated placeholders and expanded explanations for run IDs/names; added keywords.
models/runs/run-states.mdx Wording/formatting updates for run state definitions and viewing states; added keywords.
models/runs/search-runs.mdx Clarified search/regex toggle procedure; added keywords.
models/runs/stop-runs.mdx Expanded stop-run guidance and added keywords.
models/runs/tags.mdx Clarified tagging workflows across SDK/API/UI and added keywords.
models/runs/view-logged-runs.mdx Expanded run-inspection guidance and improved alt text; added keywords.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Comment thread models/runs/run-identifiers.mdx Outdated
Comment thread models/runs/run-identifiers.mdx Outdated
Comment thread models/runs/run-identifiers.mdx Outdated
Comment thread models/runs/rewind.mdx Outdated
Comment thread models/runs/resuming.mdx Outdated
Comment thread models/runs/filter-runs.mdx Outdated
Comment thread models/runs/compare-runs.mdx Outdated
Comment thread models/runs/stop-runs.mdx Outdated
Comment thread models/runs/forking.mdx Outdated
Comment thread models/runs/run-identifiers.mdx Outdated
@johndmulhausen @claude
fix: address Copilot review feedback in models/runs
73bba3f 
Fix broken in-page anchors, code samples, and wording flagged in PR review:

- run-identifiers: drop dead #autogenerated-run-ids link; point Overview link
  to view-logged-runs#overview; add /runs/ to the run URL template; remove
  spaces around the kwarg = in api.run()
- forking: retarget #unique-run-identifiers link to #run-id
- rewind: retarget Overview tab link to view-logged-runs#overview
- resuming: add id="[RUN-ID]" to the resume="must" example; retarget custom
  run ID link to run-identifiers#create-a-custom-run-id
- initialize-run: call wandb.setup() instead of using it as a context manager;
  fix nested with-block indentation
- filter-runs: fix anchor to #example-filter-runs-with-tags; "works" -> "work"
- compare-runs: underperforms is "directionally worse than" the baseline
- stop-runs: capitalize Running to match run-states

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@johndmulhausen @claude
fix: correct broken run-identifiers anchors on runs landing page
e6fd29f 
models/runs.mdx linked to #unique-run-identifiers (x2) and #autogenerated-run-ids, neither of which exists on the run-identifiers page. Point all three to #run-id (the ## Run ID heading). The #run-name link was already valid and is unchanged.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@johndmulhausen @claude
Merge origin/main into style-guide/models-runs-20260604-113608
f764512 
Resolve conflict in models/runs/compare-runs.mdx: keep main's pin-limit feature (20/19), the new 'Compare runs across projects' section, and the heading restructure, layered with the style-pass edits (keywords frontmatter, section lead-ins, the 'directionally worse than' fix, and sequential list numbering).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
@johndmulhausen @claude
style: remove auto-added keywords frontmatter from models/runs
32ce07b 
The automated /style-guide pass added keywords: frontmatter to each page. Per reviewer preference (auto-generated keywords are noise), back them out across all 20 models/runs pages.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

Copilot code review Copilot Copilot left review comments

At least 1 approving review is required to merge this pull request.

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@johndmulhausen