Add reproducible Epoch AI frontier training compute (@fig-ai-compute-growth) and single-run pretraining energy (@fig-frontier-training-energy) figures to Volume II, Chapter 15, paired with builder-reported anchors at declared measurement boundaries, deterministic CSV extraction and manifest receipts, and structured provenance enforcement across book CSV series and figure cells.
METHOD.md marks each applicable clause met, partial, or open, including the
departures: no expert committee, no human indexing trial, and model-assisted
decisions the standard assumes people make.
- Two audits applied the all-and-some test to 127 specializes edges; the 27
both failed are retyped or removed, and specializes now exports as
skos:broader while part-of stays associative under Z39.19 8.3.3.2
- Compiler enforces SKOS S27; tests validate the export
- Concepts record added_in and computed warrant counts
- Release adds an OWL relations vocabulary, DCAT metadata, CITATION.cff,
scope notes, and redirect pages for identifiers
- Agreement reports Hooper and Rolling consistency
Subject agreement 78.7% (kappa 0.78) under the first rules and 83.7%
(kappa 0.83) after revising them on the same sample, which the record
marks as no longer held out.
The framework compiler now accepts a groups tier and nested readable codes
(L4.4), resolves chapter pointers to titles and clean published URLs, and
rejects scope notes that cite a row or area that does not exist. The field
compiler enforces a controlled relation vocabulary, rejects duplicate concept
pairs, and keeps retired concept identities pointing at a replacement.
Each build writes public downloads (CSV, JSON, SKOS Turtle, GraphML, BibTeX)
to site/explore/downloads. `binder concepts coverage` scores recorded
checklists and fails when a judgment cites a concept that no longer exists.
Chapter URLs now use the flattened published form (/volN/<chapter>/...).
Bug fixes
- Build lock is keyed on books/ so builds started from different
directories block each other (fixes two tests broken by 89b828efa3).
- ConfigManager resolves the repo root when binder runs from books/ or
binder/, so bib, newsletter, doctor and render paths are right.
- `build <word> --all` treats intro/scaling/agentic/physical as chapter
names (it built every volume), and Ctrl-C stops a --all loop.
- doctor checked chapter names that no longer exist and failed on the
generated _quarto.yml; info defaulted to a deleted books/contents.
- `headings apply --path <typo>` no longer rewrites the whole book;
`bib clean` stops on unreadable chapters instead of dropping entries.
- Discovery falls back to a line scan on invalid YAML instead of crashing.
- Volume III/IV were labeled Volume II in layout and release output.
Output
- Unexpected errors print the command, exception type, and raising
file:line (new cli/core/errors.py), with the traceback on stderr.
- Chapter-not-found suggests the closest names and the list command.
- Silent except blocks in bib lookups, info, headings, layout,
maintenance, preview and debug now say what went wrong.
Cleanup
- Shared qmd_files_under/display_path, file-rewrite loop, doctor result
helper, preview server, clean helpers, volume display names.
- Removed legacy import fallbacks, dead config attributes, legacy backup
restore logic, and unused imports.
Bug fixes
- Single-file --path now scans in nine scopes that treated a file as a
directory and passed having checked nothing.
- Citations: vol3/vol4 chapters were skipped since the bib unification;
every volume now falls back to the shared references.bib, and a missing
bib is reported. Line numbers match the exact key.
- Table caption scopes resume correctly after fences; pdf-entry-numbering
reads the index files it exists to check; python-dollar-collision no
longer stops at a one-line $$...$$; rendered-python-leak line offsets;
float-flow handles :::: openers; pdf-verify no longer passes a failed
edge-overflow row; checklist labels vol3/vol4 correctly.
- --check-scope reports check_scope_retired instead of crashing.
Output
- Checks that cannot run (missing PDF, EPUB, pdfplumber, PyMuPDF, aspell,
unparsable YAML, a crashing checker) report an error with the command
that fixes it instead of passing.
- Book-relative paths everywhere; index/registry findings carry file:line.
- Issues without a suggestion take the fix from check_catalog.yaml.
- Failures end with the command to rerun each failing scope; a clean run
prints one summary line (suppressed by --quiet).
Cleanup
- One helper each for the index, LEGO/registry, and repo registry
wrappers; merged audit-check runners and shared violation conversion.
- Removed dead helpers, aliases, constants, and unused imports.
Check counts on the books are unchanged for every pre-commit invocation.
Parallel and debug builds removed their worktrees only on normal exit or
Ctrl-C. SIGTERM (timeouts, agent harnesses) and SIGHUP (closed terminal)
killed the process before cleanup, and nothing ever swept the leftovers,
so detached worktrees piled up in $TMPDIR/binder-workspaces (28 GB today).
- run_jobs and the debug bisection now turn SIGTERM/SIGHUP into
KeyboardInterrupt, so their existing cleanup runs.
- run_jobs first removes runs whose binder process has exited (the pid is
in the run directory name) and prunes their worktree registrations.
- --keep-workspaces runs are marked and skipped by the automatic sweep.
- New `binder clean workspaces` removes leftovers on demand, kept runs
included.
The 2026-10-12 HTML numbering change put a top-level number-sections: false
in 55 front-matter files, including each volume's PDF entry file. In a book
PDF that file's metadata is book-wide, so every chapter went unnumbered and
every figure and section reference printed as 0.x. The setting now sits under
format: html:, and binder check structure --scope pdf-entry-numbering fails
if an entry file turns numbering off book-wide again.
Verified with full PDF builds of all four volumes: figures print as 1.1 and
up, appendices are lettered, and HTML appendix pages stay unnumbered.
A \ref to a callout defined in another chapter prints raw in per-chapter
HTML. Each one now names its target by @sec- (vol2 5, vol3 99, vol4 13).
In vol3 the recurring lighthouse asides keep one pointer to
@sec-vol3-intro-lighthouses per chapter and drop the rest. The vol2
conclusion margin pointer is split so each cites one section.
All four books are clean after the plain-language pass; the check keeps the
word from returning. It flags and never fixes, because the plain replacement
(list, log, table, record, account) depends on what the record holds.
check structure --scope number-offset requires a chapter's number-offset
to equal its NN_ folder number minus one. Two renumberings left stale
offsets that printed the wrong chapter number, and nothing caught it.
All four volumes now pass `check figures --scope print-scale`, so the check
moves from opt-in to enforced. It stays default=False in the figures group
(it executes chapter cells) and runs through two hooks instead, which CI
also runs:
- book-check-figure-print-scale: the changed chapter files only
- book-check-figure-print-scale-all: all volumes, when the shared style,
the mlsysim plotters, or the check itself changes
`check figures --scope print-scale` now accepts the staged chapter files as
positional arguments, like other pre-commit scopes.
check structure --scope concept-maps now validates schema v2 maps: claims
cover exactly the owned concepts, one owner per concept per volume, and
every preview or touch names its real owner (previews point later).
concept-order lists concepts used before their owning chapter (opt-in).
info concept-map renders a volume's ownership map from its contracts.
The structure hook also runs when a *_concepts.yml file changes.
`bib add` refuses existing keys, so an entry edited after its gateway
seal (sentence-case title, @misc -> @inproceedings, venue, pages) had no
binder path back to a verified manifest record; records were re-written
by hand. `bib reseal --key K --note "..."` re-resolves the source the
entry was sealed from (or a --doi/--arxiv/--url given), compares title,
author surnames in order, year, DOI, and pages with the edited entry,
and rewrites the manifest record only if they all agree. It never edits
references.bib.
`fetch_url` now reads the page head and records Highwire citation_*
meta tags (title, authors, year, pages, venue, DOI) as evidence, so
publisher pages such as PMLR are checked on authors and pages rather
than on caller overrides. Author comparison folds LaTeX accent escapes
(M\k{a}dry), Unicode diacritics, and name particles.
Only bibs/ conflicted (adjacent additions on both sides). references.bib
keeps both sides; the Waymo data hub title takes dev's sentence case.
manifest.json is the union, dev winning on shared keys; the offline
ledger check passes.
Body figures are scaled to the text block, so printed text size is the
font size times text width over cropped image width. The style module
fixes the font sizes; nothing measured the other half, and vol2 canvases
drifted from 0.62x to 1.04x with no check failing.
The scope executes each chapter's python cells in a subprocess, measures
every figure inside a #fig- div, and flags print-scale (outside the
range), figure-font (a literal size below the floor), raw-figure (a
figure made without setup_plot/set_book_style), and figure-exec. Limits
are read from mlsysim.viz.style, not defined in the check.
default=False: vol2 and vol3 pass; vol1 (96 findings) and vol4 (72) still
use 8-10 in canvases and await their migration passes.
Branch commits skipped hooks; the full suite at the merge tip found these.
- Bib: drop duplicate epochai2026benchmarks from the merge; rename
villalobos2022will to villalobos2024will to match its year; re-record
ledger hashes for 24 entries whose content is unchanged (tidy reformat)
- Checks: scope cross-chapter-callout-ref to vol1 (vol2-4 carry 113 legacy
refs); add Whetstone to heading proper nouns; lego-dead-code skips
render output
- Units: computed sizes in decimal units, hardware capacities through
fmt_memory_capacity (ch6, ch11, ch14, ch16)
- Prose: abbreviation first-use expansions after content moves, footnote
marker placement, two footnote fixes, margin caption, bracketed source
citations, index main, orphan and duplicate equation labels, dead
LEGO exports
- Auto-fixers: Black on hidden cells, whitespace, blank lines, tables,
bibtex tidy, canonical spellings
The margin-pointers scope required "developed / measured /
operationalized / enforced in" on every forward pointer, which rejected
the ruled Vol I wording for appendix targets. Appendix targets now also
accept "collected for reference in" (accepted, not required, so vol2-4
pointers are unaffected; all four volumes pass). The four remaining
Vol I appendix pointers that said "operationalized in" (ch11, ch12,
ch16 x2) now use the ruled wording.
Retitle the chapter ML Lifecycle and rename its folder, files, assets,
lab, slides, audit data, and instructor links, with redirects for the
old URLs. Section IDs keep the sec-ml-workflow prefix. Update the ch2
hand-off and the conclusion to name the lifecycle.
Rename the chapter folder, assets, lab, slides, audit data, and staffml
links from 02_ml_systems to 02_system_paradigms, with redirects for the
old URLs. Revise the introduction, notation, conventions, and about
pages, adding the frontmatter stack figures. Add park2024offloading.
Every vol4 math symbol is a registered macro: 2,342 vol4 registry entries
(mostly physical domain); the applier converted 18,945 sites; per-chapter
editors finished captions, fmt_math strings and 288 reviewed flags; 134
same-letter collisions were resolved (back-matter appendices judged per
derivation), with figures and fig-alt following. The notation chapter is
self-contained. Vol4 joins STRICT_CONVERTED_VOLUMES and
COLLISION_CHECKED_VOLUMES. Baseline 21,666 -> 2,721 (no converted volume
left with a raw symbol).
Registry, book-wide and rendering-neutral where it touches vol1-3:
- states, actions, observations and similar entries that baked in a running
index (s_t, a_t, o_t, bar-alpha_k, f_k, G_k) hold the bare symbol; 370 uses
now write the index, so o_{t+1} and s_{0:T} are possible;
- four entries registered twice merge (StateAtTurn into TrajectoryState,
ActionObservation into TrajectoryObservation, ActionProposal into
TrajectoryAction, TokenLossMask into PositionLossMask); the old names stay
as aliases;
- LaTeX fixes: tau_{max}, mu_{min} braced; T_C for the Bouligand cone; the
centroidal momenta no longer carry "in R^3"; v_d pairs with v_q.
Vol1-3 files touched by these changes verify identical. Vol4 PDF (1,071 pp)
and HTML build clean; HTML has 0 MathJax errors on every page except two
pseudocode blocks in appendix_ml that predate this change (\Statex is not
supported by pseudocode.js); vol3 PDF rebuilt clean.
Every vol3 math symbol is a registered macro: 893 vol3 registry entries; the
applier converted the chapters (re-applied after the line-break fix);
per-chapter editors finished captions, fmt_math strings and the 128 ruled
review flags (restored dropped indices, grouped indices on labeled symbols,
SPRT and canary symbols, policy-superscript returns Q^pi, V^pi, A^pi); 70
same-letter collisions were resolved, the math appendix judged per
derivation. Chapters without front matter now carry one for notation-local.
The notation chapter is self-contained. Vol3 joins STRICT_CONVERTED_VOLUMES
and COLLISION_CHECKED_VOLUMES; validate checks back-matter appendices per
### derivation. Baseline 29,899 -> 21,666.
Registry: StateExposureLevel renders S (one level per index);
ServiceTimeVariationCoefficient renders c_s (also vol2 ch08, Kingman pair
with c_a); definitions widened where reviewers found them too narrow.
Vol3 PDF (913 pp) and HTML build clean, 0 MathJax errors on all 26 math
pages; vol2 PDF rebuilt clean.
Every vol2 math symbol is a registered macro: the applier converted 7,600
sites; per-chapter editors finished captions, fmt_math strings and the
reviewers' ruled fixes, proving each span with KaTeX; 43 same-letter
collisions inside sections were resolved (recolor, switch, new entry) with
figures, fig-alt and prose following. The notation chapter is
self-contained: rows only for symbols the chapters use, \Glyph for any other
symbol a note mentions. Vol2 joins STRICT_CONVERTED_VOLUMES and
COLLISION_CHECKED_VOLUMES; the glossary, a catalog like the notation
chapter, is exempt from the per-section rule. Baseline 37,038 -> 29,899.
Registry: CloudOffloadEnergy renders E_transmit (also vol1 ch02),
ResponseTimeStandardDeviation sigma_t, definitions widened where reviewers
found them too narrow, new entries merged. Vol1 ch10: brace the latency
reward exponent. Vol2 PDF and HTML build clean (0 MathJax errors on all 36
pages); vol1 PDF rebuilt clean.
Volume I: the 60 places where one letter meant two things inside one top-level
section are resolved (author ruling: per section). 51 entries take a
distinguishing glyph confined to vol1 (upright SI micro, Kolda's A/B/C for
tensor decomposition with the mode order fixed, x_feat, p_val, ...), 10 sites
switch to an existing entry, 4 entries are new, and a few labels move into
words or figures (fig-perceptron's activation is now f). A general quantity
and its special case may share a letter (GEMM column N and square N; listed in
vocabulary.yaml same_glyph_ok). vol1 joins COLLISION_CHECKED_VOLUMES; the
check is clean. Three entries no chapter uses any more are retired;
\AcceleratorBatchTime gets a notation row.
Registry: the ~700 entries the vol2 conversion needs are registered (no
chapter uses them yet).
Vijay Janapa Reddi