Skip to content

refactor: rename eval_set to dataset across codebase#814

Merged
christso merged 2 commits intomainfrom
812-rename-eval-set-to-dataset
Mar 28, 2026
Merged

refactor: rename eval_set to dataset across codebase#814
christso merged 2 commits intomainfrom
812-rename-eval-set-to-dataset

Conversation

@christso
Copy link
Copy Markdown
Collaborator

Summary

Closes #812

  • Rename eval_set field to dataset in core types (EvalTest, EvaluationResult), Zod schema, YAML/JSONL parsers, orchestrator, and OTel exporter
  • Update all CLI commands: artifact-writer, junit-writer, manifest, serve endpoint (/categories/datasets), pipeline (input/run/bench/grade), trace (show/stats/utils)
  • Rename Studio UI: CategorySummaryDatasetSummary, CategoriesResponseDatasetsResponse, route /category//dataset/, heading "Categories" → "Datasets", CategorySidebarDatasetSidebar
  • Regenerate eval-schema.json and routeTree.gen.ts
  • Maintain backward compatibility: JSONL parser/manifest accept both eval_set and dataset, pipeline readers use dataset ?? eval_set, trace stats accepts --group-by eval-set as deprecated alias

Risk

High — breaking API change (renamed wire format fields, API endpoint, route URLs). Backward compat maintained for reading old data.

Test plan

  • bun --filter @agentv/core typecheck — clean
  • bun --filter @agentv/core test + bun --filter agentv test — 353 tests pass, 0 fail
  • bun --filter @agentv/studio build — builds clean
  • biome check . — lint clean
  • Pre-push hooks (build, typecheck, lint, test, validate) — all pass
  • Manual: load old JSONL result files with eval_set field in Studio

🤖 Generated with Claude Code

Rename every occurrence of eval_set to dataset — core types, wire format,
API endpoints, Studio UI labels, routes, and component names — to align
with industry conventions (Braintrust, LangSmith, DeepEval all use "dataset").

Backward compatibility maintained:
- JSONL parser/manifest reader accept both eval_set and dataset fields
- Zod schema accepts both field names (eval_set as deprecated alias)
- Pipeline bench/grade read manifest.dataset ?? manifest.eval_set
- Trace stats CLI accepts --group-by eval-set as deprecated alias

Studio UI changes:
- "Categories" → "Datasets" in headings and labels
- Route /runs/:runId/category/:category → /runs/:runId/dataset/:dataset
- CategorySidebar → DatasetSidebar component rename
- API endpoint /api/runs/:filename/categories → /datasets

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@cloudflare-workers-and-pages
Copy link
Copy Markdown

cloudflare-workers-and-pages bot commented Mar 28, 2026

Deploying agentv with  Cloudflare Pages  Cloudflare Pages

Latest commit: b622b3f
Status:⚡️  Build in progress...

View logs

Rename "eval_set" to "dataset" in all 42 example baseline JSONL files
under examples/features/ and examples/showcase/ so new users see the
current field name.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@christso christso merged commit 5f8f61c into main Mar 28, 2026
1 of 2 checks passed
@christso christso deleted the 812-rename-eval-set-to-dataset branch March 28, 2026 13:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

refactor: rename eval_set to dataset across codebase

1 participant