docs: update dedup docs for WorkflowRunResult (PR #1275)

[lbliii]

Description

Updates v26.04 fern documentation to reflect the new WorkflowRunResult return type introduced in #1275. All deduplication workflow run() code examples now capture the return value and show available metadata keys. Adds API reference documentation for WorkflowRunResult and WorkflowBase to the pipeline reference page. Replaces the 26.02 release notes with a 26.04 skeleton including the Workflow Results API entry and breaking changes.

Usage

from nemo_curator.stages.deduplication.exact.workflow import ExactDeduplicationWorkflow

exact_workflow = ExactDeduplicationWorkflow(...)
result = exact_workflow.run()
print(result.metadata)  # {"total_time": 42.1, "num_duplicates": 1500, ...}

Checklist

• I am familiar with the Contributing Guide.
• New or Existing tests cover these changes.
• The documentation is up to date with these changes.

Run the following to install Fern:

npm install -g fern-api

Run the following to generate a local preview:

fern docs dev

[greptile-apps]

Greptile Summary

This PR updates the v26.04 Fern documentation to reflect the new WorkflowRunResult return type introduced in PR #1275, capturing result in all deduplication workflow run() call examples and adding inline metadata key comments. The release notes and breaking-change entries are well-written, and the previous review concerns about WorkflowBase inheritance and missing metadata keys in semdedup.mdx have been addressed.

One minor inconsistency remains: pipeline.mdx was only reformatted (plain link → list item) but the PR description states that WorkflowRunResult and WorkflowBase API reference docs were added to that page — those additions are absent from the diff.

Confidence Score: 5/5

Safe to merge; all findings are P2 documentation improvements with no impact on runtime behaviour.

All prior P1 concerns (WorkflowBase inheritance claim, missing id_generator_path, incomplete semdedup metadata comments) have been resolved. The one remaining inline comment (incomplete TextSemanticDeduplicationWorkflow metadata list in index.mdx) is a minor P2 inconsistency between the overview page and the more detailed semdedup.mdx page.

fern/versions/v26.04/pages/curate-text/process-data/deduplication/index.mdx — the TextSemanticDeduplicationWorkflow metadata comment at line 118 is incomplete relative to semdedup.mdx and the source code.

Important Files Changed

Filename	Overview
fern/versions/v26.04/pages/curate-text/process-data/deduplication/index.mdx	Adds WorkflowRunResult capture and metadata comments to all workflow examples; the TextSemanticDeduplicationWorkflow comment is still incomplete compared to semdedup.mdx and the source code.
fern/versions/v26.04/pages/curate-text/process-data/deduplication/exact.mdx	Captures WorkflowRunResult for all run() calls and adds accurate metadata comments including the previously-missing id_generator_path.
fern/versions/v26.04/pages/curate-text/process-data/deduplication/fuzzy.mdx	Captures WorkflowRunResult for all run() calls and adds accurate metadata comments (all keys verified against fuzzy/workflow.py).
fern/versions/v26.04/pages/curate-text/process-data/deduplication/semdedup.mdx	Updates TextSemanticDeduplicationWorkflow and SemanticDeduplicationWorkflow result capture; all seven metadata keys verified against semantic.py and semantic/workflow.py.
fern/versions/v26.04/pages/about/release-notes/index.mdx	Adds Workflow Results API section and breaking-change entries; correctly qualifies that TextSemanticDeduplicationWorkflow implements the interface without formal WorkflowBase inheritance.
fern/versions/v26.04/pages/api-reference/pipeline.mdx	Source Code section reformatted to a list item; no WorkflowRunResult or WorkflowBase API reference added despite the PR description claiming this.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["workflow.run()"] --> B["WorkflowRunResult"]
    B --> C["result.metadata dict"]
    B --> D["result.pipeline_tasks dict"]

    C --> E1["ExactDeduplicationWorkflow\ntotal_time, num_duplicates,\nidentification_time, id_generator_path"]
    C --> E2["FuzzyDeduplicationWorkflow\ntotal_time, num_duplicates, minhash_time,\nlsh_time, connected_components_pipeline_time,\nid_generator_path"]
    C --> E3["TextSemanticDeduplicationWorkflow\ntotal_time, num_duplicates, num_duplicates_removed,\nembedding_time, identification_time,\nremoval_time, final_output_path"]
    C --> E4["SemanticDeduplicationWorkflow\ntotal_time, num_duplicates,\nkmeans_time, pairwise_time"]
    C --> E5["TextDuplicatesRemovalWorkflow\ntotal_time, num_duplicates_removed"]

Loading

_{Reviews (3): Last reviewed commit: "docs: remove internal WorkflowRunResult/..." | Re-trigger Greptile}

[praateekmahajan]

Do we need this change? @ayushdg / @sarahyurick / @VibhuJawa seems too internal

[sarahyurick]

Yeah maybe it can be added after the WorkflowBase section and just be:

## WorkflowRunResult

A dataclass returned by all deduplication workflow run() methods. Contains workflow outputs, pipeline task mappings, and metadata such as timing and duplicate counts.

See the API documentation (link) for more details.

without the full list of methods etc.

[praateekmahajan]

Rest of the PR LGTM except that one comment

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.