Skip to content

Docs/readme structure - #116

Merged
MHindermann merged 3 commits into
mainfrom
docs/readme-structure
Sep 16, 2026
Merged

MHindermann merged 3 commits into
mainfrom
docs/readme-structure

Conversation

@MHindermann

Copy link
Copy Markdown
Member

clean up README structure

MHindermann and others added 3 commits September 15, 2026 23:12
Replace the numbered sections with task-oriented ones, lead with the
results dashboard, and move reference material into collapsed blocks.
The benchmarking rationale is back at the top, and the intro follows the
project abstract instead of narrowing the scope to historical documents.

- fix the generate_date_report.py example, which passed a path where the
  script requires --date
- add the missing HUGGINGFACE_API_KEY and the huggingface provider
- restore the fork step for contributors
- say ground truths throughout, not reference annotations
- use bold labels rather than headings inside <details>, so the GitHub
  outline holds no entries that lead into collapsed content

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Replace the textbook definitions of F1, precision, recall and error rate with
what a reader has to check before combining two numbers: whether it is a
measurement, a count or a setting; which direction is better; and whether it
aggregates at all. Each point names the case in the suite that shows it --
mean_iou beside iou_threshold, library_cards rebuilding micro F1 from summed
counts, book_advert_xml recording fuzzy on 0-100 where every other benchmark
uses 0-1.

Drop the limitations section. Its one point that reads as description rather
than caveat -- that a blank is a missing measurement and not a zero -- moves
into the cost and runtime notes, and "Methodology and limitations" becomes
"Methodology".

Also fold the changelog's Fixed entries into the single Changed bullet and add
the missing v0.5.5 release link.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The section argues for benchmarking in humanities research; its title promised a
definition first. The nav entry already read "Why benchmark?", so the heading and
the link that points at it now agree.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
@MHindermann
MHindermann merged commit 76cf2d3 into main Sep 16, 2026
2 checks passed
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.

1 participant