Nov 12 Yesterday 14:20 Fix viewer filters and content-scope wording for v3.0.0 Patch-level metadata/docs fix for v3.0.0. No task, reference, edition, population, archive or definitions change.
- `manifest.json`: set `viewer_filters` to exclude `content-scopes-530` and the `webfetch` `dynamic` track, so a Webstone republish from this revision reproduces the current viewer configurations (verified: Webstone `DatasetRelease.build` output equals the committed `data/*.jsonl` byte for byte).
- `README.md`: describe `--score-scope` (both `main` and `full` scored from the same saved responses since Webstone PR #276) and the role of `--content-scope`; rename `browser-harness-eval run` to `webstone run`.
Checks: `make check` (37 tests OK) and `tools/rebuild_publication.py --check` -> 530 tasks, 65 archive members. Archive sha256 and all Webstone data-lock file hashes are unchanged. 10:11 Release v3.0.0: 530-task edition without duplicated examples (#9 ) - Release v3.0.0: 530-task edition without duplicated examples (3ce9462cbc108560a8bf130f74600f3d1e0fb9e6)
- Qualify the dynamic-example exclusion and selection timing (b3f6b5e0ad6f8b57dc8d22fd05103d59698e97ee) 06:49 Add paired main/full content requirements for 530 frozen tasks (#6 ) - Add paired main/full content requirements for 530 frozen tasks (9ddbdad2f6fd90921e31ba2db4902cbdd96772ec)
- Preserve v2.2.0 metadata and validate annotation cohorts separately (5f15e1e4cf6477b81395fee89ffeecb01fd5cdfb)
- Bind annotation publication to its uploaded source revision (0bb8b410a3836c605f5cf3828e0a30faa41a006c)
- fix: integrate v2.2.0 and remove duplicate annotation viewer (048b9962c11ee8518374133203275c306bedb888)
- fix: bind scope annotations to merged Webstone and register publication inputs (289f9011155d322adc68676e94839ac21448e530)
- fix: publish selectable paired content scope annotations (3feb3ff7d8cc8695edbd0a62979dd09e2231469f)
- docs: explain the frozen input default content scope (6ae0fc167a4f3c6228703e7ab75c574bd1accb34)
- Restore original suite order and link merged Webstone PR #268 (5bf15bc6200990c7ea852b54ee22ead190b9ab9f) 04:24 Rename Browser Eval references to Webstone The GitHub repository lexmount/browser-eval is now lexmount/webstone. Prose and links only; manifest.json, snapshots.json and definitions/ are unchanged, so pinned revisions and hashes stay valid. Nov 11 2 days ago 07:08 Release v2.2.0: corrected references for the 550-task edition (#7 ) - Release v2.2.0: corrected references for the 550-task edition (8ae4400191ff0743a66652ce8640b635c1035b56) 04:08 docs: document LB-WF citation shorthand Nov 10 3 days ago 15:56 Add reviewed primary-content task scope revision (#3 ) - Add reviewed primary-content task scope revision (3a1f5a96fc766194a6bcf74bb0ec8a7af5f66444)
- Address scope revision review without changing frozen editions (c068f05b669ec0c671c1e83ac12bdbdaea9f2a7b)
- Bind publication to the uploaded source commit (7ce1ddd0d0a43255ce2ba381220ac17db0aa3ee2)
- Finalize the 530-task contract and verify the original 64 case changes (7dfdbfb3ef4072345178729dcba19d28482dc1c9)
- Publish the verified archive for the final 530-task contract (fbe33cb47179286d6ccaf1dccaa7b6e0d0e8ad47)
- Verify revised article source answers and fix citation spacing (ee1663f03769082bacaf656d8928a61b69afd572)
- Rebuild publication from verified article source revision (411574aab9aa1ea4aafed1cb5244958829655df3)
- Bind recovered full-answer controls without changing task rules (8016e3c90ac206f5c94342689b2d4962753a89ab)
- Rebuild publication with recovered source controls (8dbe96cf27ef4d4696d66a45bbfde9570765224b)
- Complete source controls for the remaining four reviewed cases (e1… Nov 2 11 days ago 03:54 Document dataset versions and select the expanded 550-task edition (#2 ) - Document dataset versions and select the expanded 550-task edition (a6950ac452951dbe8a1c6de7e710599e98c4c4da) Oct 30 14 days ago 07:14 Align current Web Fetch viewer with seven benchmark categories Oct 29 15 days ago 14:11 Use canonical dataset IDs in load examples and references 14:06 Use canonical LexBench dataset names 09:58 Link reviewed Browser Eval implementation without changing task inputs 09:34 Link reviewed Browser Eval implementation without changing task inputs 08:56 Link reviewed Browser Eval implementation without changing task inputs 05:04 Document benchmark scope and reproducible Browser Eval usage 03:42 Improve task hierarchy, languages, sources and usage documentation Oct 28 16 days ago 14:01 Bind reproducible archive hash in dataset manifest 13:27 Preserve current definitions from 5a5b9367118dcc25f126ee16ef4de2d7c666f190 13:26 Preserve historical definitions from d0e78e71cf5145217cb1f2b71420a7c4194d80dd 13:23 initial commit 
