PlanetCrash/Chaos_and_Authaumatronics_LLM_Training_Dataset

Chaos & Authaumatronics — LLM training dataset, revision 2

Author of the book and of this dataset: Parker Emmerson (Yaohushuason), Independent Researcher, ORCID 0009-0007-1288-3292, Parkeremmerson@icloud.com

This dataset is the whole book Chaos & Authaumatronics (subtitle on the contents page: "Analogous Semiotic Notations for Topology, Game Theory and Chaos Theory"; the fifth book of the author's collected works, 784 printed pages, about 60 papers in four sections, three Google Colab notebooks and two Mathematica notebooks) in the forms a training run learns from. The rule of the build: every piece of content in the book's source files is in the training data, in the form a training run actually learns from — inside the answers of the question-and-answer records, inside the pretraining records, inside the vision records, and in the notebook files.

What is in the book, and where it comes from

Source	What it is	Where it goes
58 LaTeX source files (the archive)	the papers as written: prose, headings, labels, formulas, tables, listings, bibliographies, comments	reproduced verbatim as answers (latex_source) and as pretraining records; copies under sources/latex/
the printed book (PDF, 784 pages)	typeset papers, cover and contents pages, three Colab prints, two Mathematica prints	the 74 pages whose text the sources lack are given from the text layer (pdf_text); the notebook prints are rebuilt as .ipynb files and given cell by cell (notebook_*); the pictures are given as vision records
46 picture files of the archive	figures of the papers (40 printed, 6 not printed)	vision records with the text inside each picture transcribed
the vector drawings of pages 470, 487 and 488	the ListPlot curve, axes, ticks, panel frames and slider tracks of the Mathematica outputs	plot-number records, numbers exactly as stored

Papers

Section of the book	Paper	Date	LaTeX source	Printed pages
front matter	Chaos & Authaumatronics (cover and contents)	—	text layer of the PDF	1–4
Section 1: Quasi-Quanta Language Package	Quasi-Quanta Language Package	July 2023	text layer of the PDF	5
Section 1: Quasi-Quanta Language Package	Vector Wave through Calculus Syngergizes Quasi Quanta to Transcendental Numbers Synchronistically from Infinity Meanings	June 2023	yes	6–43
Section 1: Quasi-Quanta Language Package	Combinations of Quasi Quanta Expressions	June 2023	yes	44–52
Section 1: Quasi-Quanta Language Package	Quasi Quanta Logic	June 2023	yes	53–56
Section 1: Quasi-Quanta Language Package	Reverse Engineering Imaginary Gauge Artefacts of Sharp Quasi-Quanta Logic Algebras	June 2023	yes	57–66
Section 1: Quasi-Quanta Language Package	Reverse Engineering Imaginary Gauge Artefacts of Sharp Quasi-Quanta Logic Algebras (second, longer version of the source)	June 2023	yes	not printed (archive only)
Section 1: Quasi-Quanta Language Package	Non-linear Solve Methods (A Generalization)	July 2023	yes	67–76
Section 1: Quasi-Quanta Language Package	Quasi-Quanta Symbolic Numeric Energy Algebra	July 2023	yes	77–84
Section 1: Quasi-Quanta Language Package	Encephalon Logic	July 2023	yes	85–93
Section 2: Extensions of Quasi-Quanta Linguistics	Infinity Logic Ray Calculus with Quasi-Quanta Algebra Limits	July 2023	yes	94–111
Section 2: Extensions of Quasi-Quanta Linguistics	Light Ray Morphisms of the Fractal Antenna	July 2023	yes	112–117
Section 2: Extensions of Quasi-Quanta Linguistics	Maximizing Maximum Time of a Dynamical System Through Optimal Radius of Acceleration Calculation	October 2023	yes	118–121
Section 2: Extensions of Quasi-Quanta Linguistics	Vector Calculus of Notated Infinitones	July 2023	yes	122–137
Section 2: Extensions of Quasi-Quanta Linguistics	Thickness Algebra	November 2023	yes	138–148
Section 2: Extensions of Quasi-Quanta Linguistics	Abstract Emotives	August 2023	yes	149–155, 384–390
Section 2: Extensions of Quasi-Quanta Linguistics	Dimensional Gateway	June 2023	yes	156–160
Section 2: Extensions of Quasi-Quanta Linguistics	Navigating Data Structures	July 2023	yes	161–165
Section 2: Extensions of Quasi-Quanta Linguistics	Type-meme Genetic Calculus	July 2023	yes	166–174
Section 2: Extensions of Quasi-Quanta Linguistics	Lorentz Fossilization	August 2023	yes	175–177
Section 2: Extensions of Quasi-Quanta Linguistics	True Number Fields	September 2023	yes	178–180
Section 2: Extensions of Quasi-Quanta Linguistics	Twitchy Limit Notation	October 2023	yes	181–185
Section 2: Extensions of Quasi-Quanta Linguistics	Set Theory and Energy Numbers	October 2023	text layer of the PDF	186–190
Section 2: Extensions of Quasi-Quanta Linguistics	Quasi Quanta Logic 2	October 2023	text layer of the PDF	191–196
Section 2: Extensions of Quasi-Quanta Linguistics	Vectroid Entangler 1	September 2023	yes	197–205
Section 2: Extensions of Quasi-Quanta Linguistics	Vectroid Entangler 2	September 2023	yes	206–216
Section 2: Extensions of Quasi-Quanta Linguistics	TRRRANN	September 2023	yes	217–233
Section 2: Extensions of Quasi-Quanta Linguistics	Table of Logic to Arithmetic Mappings	September 2023	yes	234–235
Section 2: Extensions of Quasi-Quanta Linguistics	Axiom of Choice	September 2023	yes	236
Section 2: Extensions of Quasi-Quanta Linguistics	Hom Functorial Logics	August 2023	yes	237–239
Section 3: Game Theory	Notes on Lie Superalgebra	September 2023	yes	240–250
Section 3: Game Theory	Theory on Quantum Complexes	October 2023	yes	251–260
Section 3: Game Theory	Quantum Gene Splicing	October 2023	yes	261–272
Section 3: Game Theory	quasi-quanta algebra concept	July 2023	yes	273–277
Section 3: Game Theory	Quantum Communication Games	September 2023	yes	278–293
Section 3: Game Theory	Gamer Iteratives 2	October 2023	yes	294–300
Section 3: Game Theory	Running Quantum, Virtual Games Through Limbertwig	October 2023	yes	301–305
Section 3: Game Theory	Infinity Geometries for Quantum Game Theory: A Notation of Quantum Computer Dimensionality	October 2023	yes	306–316
Section 3: Game Theory	Try to Fix	October 2023	yes	317–329
Section 4: Chaos	Combined Systems; quasi quanta semiotic branch	November 2023	yes	330–346
Section 4: Chaos	Exoplasm	October 2023	yes	347–356
Section 4: Chaos	Chaotic-Coordinate Numerals and Semiotic Notations for A Priori Numeric Energy (Energy Numbers)	November 2023	text layer of the PDF	357–373
Section 4: Chaos	Implicit Glue Macros	October 2023	yes	374–378
Section 4: Chaos	Authomaton	November 2023	yes	379–383
Section 4: Chaos	Conjunctions	December 2023	yes	391–394
Section 4: Chaos	Authaumturgy on a Sigma-Adic Ressurection Numeral	December 2023	yes	395–396
Section 4: Chaos	Inductive Maxims	December 2023	yes	397–399
Section 4: Chaos	Jagged Numbers	November 2023	yes	400–401
Section 4: Chaos	Koszul Plucker	October 2023	yes	402–421, 422–437
Section 4: Chaos	Morse Energy: A Multidimensional Approach	December 2023	yes	438–445
Section 4: Chaos	Morse Energy	December 2023	yes	not printed (archive only)
Section 4: Chaos	Non-Field Structure of the Reals, Projective System Preferred	June 2023	yes	446–450
Section 4: Chaos	Optimized Energy Numbers	December 2023	yes	451–467, 768–784
Section 4: Chaos	Pseudo Example Function (Mathematica notebook print)	2023	notebook print	468–472
Section 4: Chaos	Qudit 2	December 2023	yes	473–485
Section 4: Chaos	Rainbow Wings, Stable Release (Mathematica notebook print)	—	notebook print	486–491
Section 4: Chaos	Redefining Mathematical Structure: From the Real Number Non-Field to the Energy Number Field	January 7, 2024 A.D.	yes	492–499
Section 4: Chaos	Research on Energy Numbers and Associated Mathematical Structures	—	yes	500–515, 516–531
Section 4: Chaos	Sigma-Adic Numbers: Transfinite Cardinals and Mass as Mathematical Operation	December 2023	yes	532–576
Section 4: Chaos	Sigma-Adic Numbers and Transfinite Cardinals as Mathematical Operations	December 2023	yes	not printed (archive only)
Section 4: Chaos	Sigma-Adic Operator Algebras, Cardinal Flow, and Spectral Mass	—	yes	not printed (archive only)
Section 4: Chaos	Sphere packing	December 2023	yes	577–593
Section 4: Chaos	Tantras	December 2023	yes	594–597
Section 4: Chaos	Tessellations and Sweeping Nets: Advancing the Calculus of Geometric Logic	December 2023	yes	598–642
Section 4: Chaos	Programs in Tessellations (Google Colab notebook print)	—	notebook print	643–695
Section 4: Chaos	Mina 2 (Google Colab notebook print)	—	notebook print	696–715
Section 4: Chaos	Limbertwig V. 4 - Ultra (Google Colab notebook print)	—	notebook print	716–767
Section 4: Chaos	authomaton investigations	December 2023	yes	not printed (archive only)

Eleven entries of the contents pages have no pages in the printed book and no file in the archive: "Intro.pdf", "Artifact Creatures", "Novel Forms Numeric Energy Pre Numeric Quanta, Quasi Quanta and Non-Quantifiable", "*Rationale", "Fractal Partitioning", "Anti-Calculus", "Meta Poker", "Layers", "Menus vs. Displays", "Chaos Theory and Game Characters", "Game Theory on Lie algebra".

Notebooks (notebooks/)

File	Built from	Cells	Outputs
Chaos_and_Authaumatronics_Limbertwig_V4_Ultra.ipynb	the Colab print, pages 716–767	{'markdown': 1, 'code': 48}	{'stream': 26}
Chaos_and_Authaumatronics_Mina_2.ipynb	the Colab print, pages 696–715	{'code': 14}	{'stream': 8, 'display_data': 10}
Chaos_and_Authaumatronics_Programs_in_Tessellations.ipynb	the Colab print, pages 643–695	{'code': 25}	{'display_data': 26, 'stream': 22}
Chaos_and_Authaumatronics_Pseudo_Example_Function.ipynb	the Mathematica print, pages 468–472 (Wolfram Language kernel)	{'markdown': 1, 'code': 2}	{'display_data': 6, 'stream': 2}
Chaos_and_Authaumatronics_Rainbow_Wings_Stable_Release.ipynb	the Mathematica print, pages 486–491 (Wolfram Language kernel)	{'markdown': 1, 'code': 2}	{'stream': 9, 'display_data': 9, 'execute_result': 2}
Chaos_and_Authaumatronics_morse_energy_a_multidimensional_approach_listings.ipynb	the program listings of the paper named in the file (unexecuted: the paper prints no outputs for them)	{'markdown': 1, 'code': 1}	none
Chaos_and_Authaumatronics_quantum_communication_games_listings.ipynb	the program listings of the paper named in the file (unexecuted: the paper prints no outputs for them)	{'markdown': 1, 'code': 1}	none
Chaos_and_Authaumatronics_sigma_adic_operator_algebras_listings.ipynb	the program listings of the paper named in the file (unexecuted: the paper prints no outputs for them); the Smith-form output listing of the paper is attached	{'markdown': 1, 'code': 2}	{'stream': 1}
Chaos_and_Authaumatronics_tessellations_and_sweeping_nets_listings.ipynb	the program listings of the paper named in the file (unexecuted: the paper prints no outputs for them)	{'markdown': 1, 'code': 8}	none
Chaos_and_Authaumatronics_tessellations_and_sweeping_nets_wolfram_listing.ipynb	the program listings of the paper named in the file (unexecuted: the paper prints no outputs for them)	{'code': 1}	none

The Java listing of "Vector Wave…" is saved as Chaos_and_Authaumatronics_vector_wave_listing.java. The printed notebooks are executed notebooks in the sense that every output the print shows (printed text, messages, output labels, pictures) is attached to its cell in page order. The Colab prints cut some long lines at the right page edge; where a cell matches the author's public copy of the notebook (GitHub, sphereofrealization/PythonCode) line for line, the cut lines are completed from the copy, and the cell metadata says which (lines_completed_from_github_copy, output_lines_completed_from_github_copy); lines that remain cut are listed in lines_cut_at_page_edge_unresolved. The records give the printed text, and separately the completed text (notebook_code_completed, notebook_output_completed).

Files

Chaos_and_Authaumatronics_LLM_Training_Dataset_rev1/
  README.md                                   this card
  Chaos_and_Authaumatronics_LLM_Training_Dataset_rev1_manifest.json   every file with its size and SHA-256
  data/qa/        question-and-answer records in five files per split: records (with metadata), openai (chat messages), sharegpt, alpaca, prompt_completion
  data/pretraining/   plain-text records, at most 14,000 characters each
  data/vision/    vision records: records (metadata), llava, openai (picture embedded as base64)
  data/plot_numbers/  the vector-drawing numbers: chat form and pretraining form (gzip JSON Lines), and the decoded coordinate file
  images/         the 98 pictures, named by figure number
  notebooks/      the five printed notebooks rebuilt as .ipynb, and the listing notebooks of the papers
  sources/latex/  the 58 LaTeX source files, one folder per paper
  reports/        source inventory, check report, check results, coverage misses

Record formats

Question-and-answer records (data/qa/). Each record asks for one passage of the book and answers with that passage exactly as written. Kinds (source_kind): latex_source (471), notebook_code (90), notebook_code_completed (16), notebook_output (67), notebook_output_completed (2), notebook_print (7), notebook_text (3), pdf_text (74). Records: 730 (709 train, 21 validation). An answer is at most 6,000 characters; a longer passage is cut into consecutive pieces at a clean place (code after a comma or before a + or − at the outermost bracket level; prose after a sentence), and the pieces are chained: the first is asked for with its description, each later one with that description plus the end of the piece before it (chain_id, piece_index, piece_count). A LaTeX record also carries file, passage, start, end (character offsets in the file) and gap_after (the whitespace between this passage and the next), so the files rebuild byte for byte. The openai file gives each record as messages (system, user, assistant); sharegpt as conversations (system, human, gpt); alpaca as instruction/input/output; prompt_completion as prompt (ending with a blank line) and completion.

Pretraining records (data/pretraining/): text is a header line naming the book, section, paper and part, a blank line, then the text; at most 14,000 characters; none is flagged "long". 216 records.

Vision records (data/vision/): 98 pictures. The question names the picture's place in the book and asks for every piece of text in it; the answer is the transcription (or "There is no text in this picture."). records carries image, context, width, height, printed_pages and uncertain (a note where a label is clipped by the picture edge).

Plot numbers (data/plot_numbers/): the 360 paths of the vector drawings on pages 470, 487 and 488, each with its paint operator, transformation matrix, line width, colours and segments, numbers written as the shortest decimal that reads back to the same 64-bit value in Wolfram notation (1.5*^-7, 2.), cut into pieces at line ends and chained like the other records. Chaos_and_Authaumatronics_plot_coordinates.jsonl.gz holds the same paths as decoded JSON. No number in these files was computed by a tool.

Train/validation split

Deterministic: two whole papers go to validation (the SHA-256 of the paper identifier, modulo 30, is 0: "Sigma-Adic Numbers and Transfinite Cardinals as Mathematical Operations" and "Theory on Quantum Complexes"), together with every twenty-fifth picture and the plot numbers of page 470. The two splits partition every file; no chained group is split across them. To train on every character of the book, train on both splits.

Checks

The check report (reports/Chaos_and_Authaumatronics_rev1_check_report.md) gives the result of every check run on the files as written: coverage of the answers (the whole book, 2,019,897 characters of prose and computed output, has 0 characters left out of every answer), plot numbers read back bit for bit, an independent decoding, file loading, the length rules, conversion correctness and the character-class comparison, and the list of everything not in the training text with the reason for each item.

Changes from the previous revision

There is no earlier revision of this book's dataset: revision 1 is the first. What it contains, item by item:

• the 58 LaTeX source files reproduced verbatim as chained question-and-answer records and as pretraining records, rebuilt byte for byte by the checks;
• the text layer of the 74 printed pages whose text the sources lack (cover and contents pages, the "Quasi-Quanta Language Package" page, the printed edition of "Vector Wave…", "Set Theory and Energy Numbers", "Quasi Quanta Logic 2", "Chaotic-Coordinate Numerals…", and the four pages after the end of the "Koszul Plucker" source), read after rebuilding the broken Unicode maps of the Computer Modern fonts;
• the three Google Colab notebooks and the two Mathematica notebooks rebuilt from their prints as executed .ipynb files, with every code cell, printed output, message, output label, panel label and running head in the records;
• the program listings of the papers as unexecuted notebooks;
• the 98 pictures with the text inside each transcribed, as LLaVA and OpenAI vision records;
• the vector-drawing numbers of the Mathematica output pages as plot-number records;
• a deterministic train/validation split, a manifest with SHA-256 of every file, and the check report.

Citation

Parker Emmerson (Yaohushuason). Chaos & Authaumatronics: Analogous Semiotic Notations for Topology, Game Theory and Chaos Theory. 2023–2024. Dataset: Chaos_and_Authaumatronics_LLM_Training_Dataset_rev1, 2026.

Changes from Chaos_and_Authaumatronics_LLM_Training_Dataset_rev1: every file reachable by the Parquet conversion

Chaos_and_Authaumatronics_LLM_Training_Dataset_rev2 is Chaos_and_Authaumatronics_LLM_Training_Dataset_rev1 with files added and the card's configs extended, so that the Hugging Face Parquet converter (the "parquet-converter" bot, which builds the dataset viewer and the refs/convert/parquet branch from the configurations a card declares) reaches every piece of content in the repository. No file of Chaos_and_Authaumatronics_LLM_Training_Dataset_rev1 was removed or changed. Upload the folder of Chaos_and_Authaumatronics_LLM_Training_Dataset_rev1 and this folder into the same repository; this README replaces the previous one.

What was added:

• images configuration: the 98 pictures under images/ are now an image dataset (the pictures themselves become an image column in the Parquet conversion), with images/metadata.jsonl giving each picture's vision questions and answers.
• notebook_cells configuration: the 10 .ipynb files cell by cell (111 rows), with each cell's source, its printed output and its outputs as JSON (pictures inside outputs as their base64 text).
• source_files configuration: the 59 source files kept in the folder (.java, .tex) as rows with their full text.
• documents configuration: the 1 Markdown and text files of the folder (Chaos_and_Authaumatronics_rev1_check_report.md) as rows with their full text. A document longer than the viewer's cell limit is still complete in the Parquet file.
• plot_coordinates configuration: data/plot_numbers/Chaos_and_Authaumatronics_plot_coordinates.jsonl.gz — present in the previous revision but not declared, so the bot skipped it.
• reports/Chaos_and_Authaumatronics_rev1_coverage_misses.json is an empty list (no rows) and stays undeclared.
• files configuration: the manifest as a table — every file of the revision with its size and SHA-256.

New files:

File	What it holds
images/metadata.jsonl	one row per picture (98): file name, the vision examples asked about it (questions and answers), and the earlier metadata columns where the folder had them
Chaos_and_Authaumatronics_notebook_cells.jsonl	every cell of the 10 Jupyter notebooks (111 cells): source, printed output, the outputs as JSON
Chaos_and_Authaumatronics_source_files.jsonl	the 59 source files (LaTeX, programs, scripts) as rows: path and full text
Chaos_and_Authaumatronics_documents.jsonl	the 1 text documents of the folder (full texts, reports, audits) as rows
Chaos_and_Authaumatronics_LLM_Training_Dataset_rev2_files.jsonl	every file of the revision (196 carried from Chaos_and_Authaumatronics_LLM_Training_Dataset_rev1, 4 new) with its size and SHA-256, as rows

What the bot still does not read, and why it is no loss: the .ipynb files (their cells are in notebook_cells); the source files (their text is in source_files); the Markdown and text files (their text is in documents); the manifest (its rows are in files). The pictures are converted through the images configuration.

Built 2026-10-30; every new configuration was loaded with the Hugging Face datasets library before this card was written (see the check table).

Loading check of the new configurations

Configuration	Split	Result	Rows	Columns
images	train	loads	98	5
notebook_cells	train	loads	111	7
source_files	train	loads	59	5
documents	train	loads	1	4
plot_coordinates	train	loads	360	8
files	train	loads	200	4