[ { "id": "TLL-01", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the record's name for the entity", "kind": "definition", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].name", "quote": "Data Portraits", "kernel": false, "const": false, "lineage": "what it is: artifacts that record training data", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (name)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-02", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the record's other name: the paper's title", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].aka", "quote": "[\"Data Portraits: Recording Foundation Model Training Data\"]", "kernel": false, "const": false, "lineage": "what it is: artifacts that record training data", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (aka)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-03", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the record's address for the entity: the arXiv abstract", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].url", "quote": "https://arxiv.org/abs/2303.03919", "kernel": false, "const": false, "lineage": "publication, lifecycle and access", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (url)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-04", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the record's operator: the paper's two authors; no institution in the saved source; the interface's operator unread", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].operator", "quote": "the paper's two authors, Marc Marone and Benjamin Van Durme; the saved source names no institution, and the live interface's operator is unread", "kernel": false, "const": false, "lineage": "what it is: artifacts that record training data", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (operator)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-05", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the record's kind: artifacts that record training data for downstream inspection, by data sketching, in a NeurIPS 2023 paper", "kind": "definition", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].kind", "quote": "artifacts that record training data and allow downstream inspection, proposed for widespread adoption and implemented with data sketching, in a research paper (NeurIPS 2023 Datasets and Benchmarks)", "kernel": true, "const": false, "lineage": "what it is: artifacts that record training data", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (kind)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-06", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the record's areas: finding and corpora", "kind": "fact", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].areas", "quote": "[\"finding\", \"corpora\"]", "kernel": false, "const": false, "lineage": "the census and its question", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (areas)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-07", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the record's evidence kinds: research paper, dataset documentation", "kind": "fact", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].evidence_kind", "quote": "[\"research paper\", \"dataset documentation\"]", "kernel": false, "const": false, "lineage": "the census and its question", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (evidence_kind)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-08", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the record's unit: a given example queried against a portrait of a training dataset", "kind": "definition", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].unit", "quote": "a given example, queried against a portrait (a data sketch) of a training dataset", "kernel": false, "const": false, "lineage": "the membership question", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (unit)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-09", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "what it records: training datasets' contents, for membership queries; implemented for The Pile and The Stack", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].record_object", "quote": "the contents of training datasets, recorded so that a query can answer whether a model has already encountered a given example during training; implemented for The Pile and The Stack", "kernel": false, "const": false, "lineage": "corpora recorded: The Pile, The Stack, The Stack V2", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (record_object)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-10", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "in_layer: a portrait records a training dataset; The Pile and The Stack", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].record.in_layer", "quote": "a portrait records a training dataset; the paper documents a popular language modeling corpus (The Pile) and a code modeling dataset (The Stack)", "kernel": false, "const": false, "lineage": "corpora recorded: The Pile, The Stack, The Stack V2", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (record.in_layer)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-11", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "name: not recorded; the saved abstract does not say whether a portrait keeps author, title or source metadata", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].record.name", "quote": "not recorded; the saved abstract does not say whether a portrait keeps any author, title or source metadata for a recorded example", "kernel": true, "const": false, "lineage": "the name of a recorded work", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (record.name)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-12", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "readback: a membership query against the recorded dataset; the live interface unread", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].record.readback", "quote": "a membership query against the recorded dataset: the paper poses 'has the model already encountered a given example during training?' and stresses 'fast and space efficient querying'; a live interface is released (unread)", "kernel": false, "const": false, "lineage": "the membership question", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (record.readback)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-13", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "makes: not recorded as a product of the model; leakage and plagiarism checks are checks against the recorded data", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].record.makes", "quote": "not recorded as a product of the model; the paper states the tool 'enables answering questions about test set leakage and model plagiarism', which checks material against the recorded training data", "kernel": true, "const": false, "lineage": "what a model makes from a work", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (record.makes)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-14", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "method: data sketching, about 3% of dataset size, with the artifact's properties outlined", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].method", "quote": "data sketching: a compact record of a dataset built for fast, space-efficient membership queries, costing about 3% of the dataset size; the paper also outlines the properties such an artifact should have and how existing solutions can increase transparency", "kernel": false, "const": false, "lineage": "the sketch: Bloom filter, strided n-grams, hashes", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (method)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-15", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "count: two datasets documented; overhead about 3% (as of arXiv v2)", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].count", "quote": "{\"value\": \"two datasets documented (The Pile and The Stack); overhead about 3% of the dataset size\", \"as_of\": \"arXiv v2, 14 December 2023\"}", "kernel": false, "const": false, "lineage": "size and speed", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (count)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-16", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "holdings: portraits of The Pile and The Stack, the sketch implementation, a live interface (unread), the call to release portraits", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].holdings", "quote": "[\"portraits of The Pile and The Stack\", \"a sketch-based implementation costing 3% of the dataset size in overhead\", \"a released live interface (unread)\", \"a call on dataset and model creators to release Data Portraits\"]", "kernel": false, "const": false, "lineage": "publication, lifecycle and access", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (holdings)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-17", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "stated limits: none in the saved abstract; its framing, datasets 'increasingly immense and opaque'", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].stated_limits", "quote": "[\"the saved abstract states no limits of the method; its framing is that training datasets are 'increasingly immense and opaque'\"]", "kernel": false, "const": false, "lineage": "limits and error", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (stated_limits)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-18", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "answers: whether an example is in a recorded dataset; leakage and plagiarism against it", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].answers", "quote": "[\"whether a given example is in a recorded dataset (The Pile, The Stack)\", \"questions of test set leakage and model plagiarism against a recorded dataset, as the paper states\"]", "kernel": false, "const": false, "lineage": "uses: leakage, plagiarism, creators", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (answers)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-19", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "does not answer: causal dependence of a model's output on a passage; under whose name an example sits; datasets without a portrait", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].does_not_answer", "quote": "[\"whether a particular model's output depends causally on a recorded passage\", \"under whose name a recorded example sits in the dataset\", \"anything about datasets that have no portrait\"]", "kernel": true, "const": false, "lineage": "limits and error", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (does_not_answer)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-20", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the archive's reading: a readback instrument on the corpus side that settles membership for a recorded corpus", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].tll_reading", "quote": "A readback instrument held on the corpus side: a portrait records the dataset so that a work can be tested for membership, which settles #186's 'is the work in the layer' for a recorded corpus.", "kernel": true, "const": false, "lineage": "the membership question", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (tll_reading)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-21", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the archive's reading: membership is no proof of a model's use; the name and what a model made lie outside the saved record", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].tll_reading", "quote": "Dataset membership is not proof that a particular model used a passage, and the name and what a model made from the work (#1543's question) are outside what the saved abstract records.", "kernel": true, "const": false, "lineage": "limits and error", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (tll_reading)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-22", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the entity's relation to the census: infrastructure", "kind": "fact", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].host_relation", "quote": "infrastructure", "kernel": false, "const": false, "lineage": "the census and its question", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (host_relation)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-23", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "record type: method", "kind": "fact", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].record_type", "quote": "method", "kernel": false, "const": false, "lineage": "what it is: artifacts that record training data", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (record_type)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-24", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the record's question: is a given example in a recorded training dataset", "kind": "definition", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].question", "quote": "Is a given example in a recorded training dataset (training membership)?", "kernel": true, "const": false, "lineage": "the membership question", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (question)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-25", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "lifecycle: historical; a published paper, v1 and v2; whether the interface is still served unread", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].lifecycle", "quote": "{\"state\": \"historical\", \"detail\": \"a published paper: v1 6 March 2023, v2 14 December 2023 (NeurIPS 2023 Datasets and Benchmarks); whether the live interface is still served is unread\"}", "kernel": false, "const": false, "lineage": "publication, lifecycle and access", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (lifecycle)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-26", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "not a corpus: it records corpora built by others", "kind": "interpretation", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].negative_tags[0]", "quote": "NOT a corpus: it records corpora built by others", "kernel": false, "const": false, "lineage": "corpora recorded: The Pile, The Stack, The Stack V2", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (negative_tags[0])", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-27", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "not a model-side test: it queries the recorded data and needs a portrait of it", "kind": "interpretation", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].negative_tags[2]", "quote": "NOT a model-side test: it queries the recorded data and needs a portrait of it", "kernel": false, "const": false, "lineage": "limits and error", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (negative_tags[2])", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-28", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "not a record of authorship: the saved abstract says nothing of attribution metadata", "kind": "interpretation", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].negative_tags[3]", "quote": "NOT a record of authorship: the saved abstract says nothing of attribution metadata", "kernel": true, "const": false, "lineage": "the name of a recorded work", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (negative_tags[3])", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-29", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "neighbour: OLMoTrace, verbatim span matching against the full corpus, beside membership queries on a recorded one", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].relations[1].note", "quote": "both check text against training data from outside the model's weights: Data Portraits by membership queries on a recorded corpus, OLMoTrace by verbatim span matching against the full corpus", "kernel": false, "const": false, "lineage": "neighbours", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (relations[1].note)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-30", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "uses: The Pile", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].relations[0].note", "quote": "the paper documents The Pile with its tools ('a popular language modeling corpus (The Pile)')", "kernel": false, "const": false, "lineage": "corpora recorded: The Pile, The Stack, The Stack V2", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (relations[0].note)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-31", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "unchecked: the live interface; the paper's full text beyond the abstract", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "census.json rows[slug == 'data-portraits'].unchecked", "quote": "[\"the live interface (linked as 'this https URL' and unsaved): what it lets a user query, and whether it is still served\", \"the paper's full text beyond the abstract: the sketch's error properties, what a query returns, and any attribution or metadata fields\"]", "kernel": false, "const": false, "lineage": "publication, lifecycle and access", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (unchecked)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-32", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the layer census's organizing question", "kind": "definition", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json organizing_question", "quote": "How is literature addressed to a receiving apparatus, what does that apparatus make from it, and how can transmission, transformation and provenance be inspected across later reception?", "kernel": true, "const": false, "lineage": "the census and its question", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (top:organizing_question)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-33", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the layer census's admission rule", "kind": "definition", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json admission_rule", "quote": "A row is an aggregator: a resource that gathers and holds many other works or records and says what it holds (operator, 2026-11-01).", "kernel": false, "const": false, "lineage": "the census and its question", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (top:admission_rule)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-34", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the census's area 'finding': inspection, testing whether a work is in a training set or a model, or tracing an output", "kind": "definition", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json areas.finding", "quote": "Inspection: tools and methods that test whether a work is in a training set or a model, or trace an output back to training text", "kernel": false, "const": false, "lineage": "the census and its question", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (top:areas.finding)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "TLL-35", "dep": "TLL", "record": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits", "label": "TLL" }, "claim": "the census's anchors, #186 and #1543", "kind": "fact", "modality": "stipulation", "strata": [ "D" ], "locus": "census.json anchor", "quote": "Who Is Writing for the Machines? Training-Layer Literature: A Field Survey … Generative Uptake: Training-Layer Literature Extended from Machine Reading to Machine Composition (EA-TLL-COMPOSITION-01)", "kernel": false, "const": false, "lineage": "the census and its question", "date": "2026-11-01", "title": "traininglayerliterature.org layer census record: Data Portraits (https://www.traininglayerliterature.org/aggregators/data-portraits), editor Lee Sharks, compiled 2026-11-01, schema 1.1", "creator": "Lee Sharks (editor)", "file": "/home/claude/fleet/traininglayerliterature-org/data/layer-census/census.json (top:anchor)", "quote_ok": true, "quote_check": "within the field's value as loaded (non-string values as their JSON text; whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "entity": "data-portraits", "surface": { "url": "https://www.traininglayerliterature.org/aggregators/data-portraits" }, "note": "composed in session on 2026-11-01 from sources saved that day (sources/manifest.json, sha256 per source); the row rests on the arXiv abstract alone (its one source)" }, { "id": "CAP-1-01", "dep": "capture-registry", "record": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/", "label": "CAP" }, "claim": "the record one sense of four: the artifact the last bullet and one card of eight; Du Bois, Donath and Lupi the rest", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "data-portraits-aio-20261101, findings[0]", "quote": "THE RECORD ONE SENSE OF FOUR. The paper's artifact holds the last bullet and one card of eight. The rest of the field is Du Bois's work for the Paris Exposition ('over 60 hand-drawn charts, graphs, and maps'), Donath's definition, carded from the MIT Media Lab ('Data portraits depict their subjects' accumulated data rather than their faces'), and Lupi's 'wearable buttons at TED 2017', carded twice (giorgialupi, Pentagram).", "kernel": true, "const": false, "lineage": "the composition at the name", "date": "2026-11-01", "title": "Capture Registry (EA-WG-CAPTURES-01), entry data-portraits-aio-20261101 (OBS-103202920b0d, AXN:0BAD.CAPTURE.🌇🕘📎☿🔽🚀)", "creator": "Lee Sharks (the registry)", "file": "data/EA-WG-CAPTURES-01.json", "quote_ok": true, "quote_check": "the entry's field value as loaded (non-string values as their JSON text)", "surface": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/" } }, { "id": "CAP-1-02", "dep": "capture-registry", "record": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/", "label": "CAP" }, "claim": "the artifact at a coarse grain: recording and lightness carried, 'training traces' the composition's phrase; the question, method, corpora and uses absent", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "data-portraits-aio-20261101, findings[1]", "quote": "THE ARTIFACT AT A COARSE GRAIN. 'lightweight inspection tools ... that record dataset training traces.' The abstract: 'artifacts that record training data and allow for downstream inspection' and 'Our tool is lightweight and fast, costing only 3% of the dataset size in overhead.' The recording and the lightness are carried; 'training traces' is the composition's phrase. The question the artifact answers ('has the model already encountered a given example during training?'), its method ('a solution based on data sketching'), the two corpora it documents ('The Pile' and 'The Stack') and its uses ('test set leakage and model plagiarism') are absent.", "kernel": true, "const": false, "lineage": "the composition at the name", "date": "2026-11-01", "title": "Capture Registry (EA-WG-CAPTURES-01), entry data-portraits-aio-20261101 (OBS-103202920b0d, AXN:0BAD.CAPTURE.🌇🕘📎☿🔽🚀)", "creator": "Lee Sharks (the registry)", "file": "data/EA-WG-CAPTURES-01.json", "quote_ok": true, "quote_check": "the entry's field value as loaded (non-string values as their JSON text)", "surface": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/" } }, { "id": "CAP-1-03", "dep": "capture-registry", "record": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/", "label": "CAP" }, "claim": "no author, no venue, undated; each other sense carries a person", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "data-portraits-aio-20261101, findings[2]", "quote": "NO AUTHOR, NO VENUE. 'Contemporary research groups use the term': the abstract's authors, 'Marc Marone , Benjamin Van Durme', and its venue, 'NeurIPS 2023 Datasets and Benchmarks', are not named, and the sense is undated. Each of the other three senses carries a person ('Sociologist W. E. B. Du Bois', 'Digital media scholar Judith Donath', 'Designers like Giorgia Lupi').", "kernel": false, "const": false, "lineage": "the composition at the name", "date": "2026-11-01", "title": "Capture Registry (EA-WG-CAPTURES-01), entry data-portraits-aio-20261101 (OBS-103202920b0d, AXN:0BAD.CAPTURE.🌇🕘📎☿🔽🚀)", "creator": "Lee Sharks (the registry)", "file": "data/EA-WG-CAPTURES-01.json", "quote_ok": true, "quote_check": "the entry's field value as loaded (non-string values as their JSON text)", "surface": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/" } }, { "id": "CAP-1-04", "dep": "capture-registry", "record": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/", "label": "CAP" }, "claim": "a page under the name indexed, carrying the abstract's opening; the census left the live interface unread", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "data-portraits-aio-20261101, findings[3]", "quote": "A PAGE UNDER THE NAME INDEXED. The second card, titled 'Data Portraits' under the site name 'Data Portraits', carries the abstract's opening ('Foundation models are trained on increasingly immense and opaque datasets. Even while these models are now key in AI system building, it can be difficult to ans...'), and the composition's one link on the artifact goes to it. The census row records the 'live interface' as released and unread, and its lifecycle leaves open 'whether the live interface is still served'; on 2026-11-01 a page under the name is in the index, at an address the paste does not show.", "kernel": false, "const": false, "lineage": "the composition at the name", "date": "2026-11-01", "title": "Capture Registry (EA-WG-CAPTURES-01), entry data-portraits-aio-20261101 (OBS-103202920b0d, AXN:0BAD.CAPTURE.🌇🕘📎☿🔽🚀)", "creator": "Lee Sharks (the registry)", "file": "data/EA-WG-CAPTURES-01.json", "quote_ok": true, "quote_check": "the entry's field value as loaded (non-string values as their JSON text)", "surface": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/" } }, { "id": "CAP-1-05", "dep": "capture-registry", "record": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/", "label": "CAP" }, "claim": "the innocent reading: a two-word name that is also a term of visualization, defined, its senses listed", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "data-portraits-aio-20261101, reading", "quote": "The innocent reading first. Given a two-word name that is also a term of visualization and portraiture, the Overview defines the term, lists four senses of it and asks which the user means; the paper's artifact is the fourth.", "kernel": false, "const": false, "lineage": "the composition at the name", "date": "2026-11-01", "title": "Capture Registry (EA-WG-CAPTURES-01), entry data-portraits-aio-20261101 (OBS-103202920b0d, AXN:0BAD.CAPTURE.🌇🕘📎☿🔽🚀)", "creator": "Lee Sharks (the registry)", "file": "data/EA-WG-CAPTURES-01.json", "quote_ok": true, "quote_check": "the entry's field value as loaded (non-string values as their JSON text)", "surface": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/" } }, { "id": "CAP-1-06", "dep": "capture-registry", "record": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/", "label": "CAP" }, "claim": "the analysis: a disambiguation with the training-data record last, by the term alone; the membership query not carried", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "data-portraits-aio-20261101, analysis", "quote": "A disambiguation in which the training-data record is the last of four senses, named by the term alone with no author, venue, corpus or question. The membership query the census row reads it for, whether a work is in a recorded corpus, is not carried.", "kernel": true, "const": false, "lineage": "the composition at the name", "date": "2026-11-01", "title": "Capture Registry (EA-WG-CAPTURES-01), entry data-portraits-aio-20261101 (OBS-103202920b0d, AXN:0BAD.CAPTURE.🌇🕘📎☿🔽🚀)", "creator": "Lee Sharks (the registry)", "file": "data/EA-WG-CAPTURES-01.json", "quote_ok": true, "quote_check": "the entry's field value as loaded (non-string values as their JSON text)", "surface": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/" } }, { "id": "CAP-1-07", "dep": "capture-registry", "record": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/", "label": "CAP" }, "claim": "the capture's title", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "data-portraits-aio-20261101, mt", "quote": "'DATA PORTRAITS' ON THE OVERVIEW — THE TRAINING-DATA RECORD THE LAST OF FOUR SENSES, WITH NO AUTHOR, CORPUS OR QUESTION; DU BOIS, DONATH AND LUPI COMPOSED AROUND IT; A PAGE UNDER ITS NAME INDEXED", "kernel": false, "const": false, "lineage": "the composition at the name", "date": "2026-11-01", "title": "Capture Registry (EA-WG-CAPTURES-01), entry data-portraits-aio-20261101 (OBS-103202920b0d, AXN:0BAD.CAPTURE.🌇🕘📎☿🔽🚀)", "creator": "Lee Sharks (the registry)", "file": "data/EA-WG-CAPTURES-01.json", "quote_ok": true, "quote_check": "the entry's field value as loaded (non-string values as their JSON text)", "surface": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/" } }, { "id": "CAP-1-08", "dep": "capture-registry", "record": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/", "label": "CAP" }, "claim": "the PER note: neither author nor institution named; one card among seven of the other senses", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "data-portraits-aio-20261101, per_note", "quote": "Neither author of the paper nor any institution named for the artifact; a page under its name carded once among seven sources of the other senses.", "kernel": false, "const": false, "lineage": "the composition at the name", "date": "2026-11-01", "title": "Capture Registry (EA-WG-CAPTURES-01), entry data-portraits-aio-20261101 (OBS-103202920b0d, AXN:0BAD.CAPTURE.🌇🕘📎☿🔽🚀)", "creator": "Lee Sharks (the registry)", "file": "data/EA-WG-CAPTURES-01.json", "quote_ok": true, "quote_check": "the entry's field value as loaded (non-string values as their JSON text)", "surface": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/" } }, { "id": "CAP-1-09", "dep": "capture-registry", "record": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/", "label": "CAP" }, "claim": "the query: the census name, a NEW address", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "data-portraits-aio-20261101, q_kind", "quote": "the census name; NEW address.", "kernel": false, "const": false, "lineage": "the composition at the name", "date": "2026-11-01", "title": "Capture Registry (EA-WG-CAPTURES-01), entry data-portraits-aio-20261101 (OBS-103202920b0d, AXN:0BAD.CAPTURE.🌇🕘📎☿🔽🚀)", "creator": "Lee Sharks (the registry)", "file": "data/EA-WG-CAPTURES-01.json", "quote_ok": true, "quote_check": "the entry's field value as loaded (non-string values as their JSON text)", "surface": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/" } }, { "id": "CAP-1-10", "dep": "capture-registry", "record": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/", "label": "CAP" }, "claim": "the reached kind: collision, the artifact the last of four senses", "kind": "interpretation", "modality": "interpretation", "strata": [ "D" ], "locus": "data-portraits-aio-20261101, per_dim.basis", "quote": "reached: Contemporary research groups use the term for lightweight inspection tools; The paper's artifact reached as the last of four senses of the name, beside Du Bois's charts, Donath's portraits and Lupi's TED project; one card of eight is its page.", "kernel": false, "const": false, "lineage": "the composition at the name", "date": "2026-11-01", "title": "Capture Registry (EA-WG-CAPTURES-01), entry data-portraits-aio-20261101 (OBS-103202920b0d, AXN:0BAD.CAPTURE.🌇🕘📎☿🔽🚀)", "creator": "Lee Sharks (the registry)", "file": "data/EA-WG-CAPTURES-01.json", "quote_ok": true, "quote_check": "the entry's field value as loaded (non-string values as their JSON text)", "surface": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/" } }, { "id": "CAP-1-11", "dep": "capture-registry", "record": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/", "label": "CAP" }, "claim": "the dimensional PER: M 0.444, C 1.0, D 1.0, PER 0.688", "kind": "fact", "modality": "documented", "strata": [ "D" ], "locus": "data-portraits-aio-20261101, per_dim.scores", "quote": "{\"M\": 0.444, \"C\": 1.0, \"D\": 1.0, \"PER\": 0.688}", "kernel": false, "const": false, "lineage": "the composition at the name", "date": "2026-11-01", "title": "Capture Registry (EA-WG-CAPTURES-01), entry data-portraits-aio-20261101 (OBS-103202920b0d, AXN:0BAD.CAPTURE.🌇🕘📎☿🔽🚀)", "creator": "Lee Sharks (the registry)", "file": "data/EA-WG-CAPTURES-01.json", "quote_ok": true, "quote_check": "the entry's field value as loaded (non-string values as their JSON text)", "surface": { "url": "https://www.alexanarch.org/captures/data-portraits-aio-20261101/" } }, { "id": "O127-01", "dep": 127, "record": "deposit", "claim": "author tags and copyright in a page's head are almost always discarded by the extractors that prepare text for training corpora; body text and sometimes the URL are retained", "kind": "interpretation", "modality": "interpretation", "strata": [ "O" ], "locus": "§IV.2 HTML metadata (line 463)", "quote": "HTML head metadata — author tags, copyright, Open Graph, Twitter cards — is parsed by search engines and social media unfurl bots, but is almost always discarded by readability extractors that prepare text for training corpus inclusion. The training-data pipeline retains the *body text* and the *URL* (sometimes).", "kernel": true, "const": false, "lineage": "the name of a recorded work", "date": "2026-05-25", "title": "Inscriptions That Survive the Tokenizer: SPXI-TLP v2.2 — The Training-Layer Provenance Protocol", "creator": "Sharks, Lee", "file": "data/texts/AXN-02B2-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training corpora recorded by the artifact (the datasets whose texts enter machine training: The Pile, The Stack)", "cand": "O0020" }, { "id": "O127-02", "dep": 127, "record": "deposit", "claim": "no specific author's style survives training as a discrete signal that can be elicited and verified", "kind": "interpretation", "modality": "interpretation", "strata": [ "O" ], "locus": "§IV.6 Stylometric fingerprints (line 501)", "quote": "Style transfers to the model's distribution in proportion to its volume in the training set, but no specific author's style survives as a discrete signal that can be elicited and verified.", "kernel": false, "const": false, "lineage": "the name of a recorded work", "date": "2026-05-25", "title": "Inscriptions That Survive the Tokenizer: SPXI-TLP v2.2 — The Training-Layer Provenance Protocol", "creator": "Sharks, Lee", "file": "data/texts/AXN-02B2-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training-data membership testing and inspection (tools and methods that test whether a work or example is in a training set, or trace an output to training text)", "cand": "O0021" }, { "id": "O127-03", "dep": 127, "record": "deposit", "claim": "a membership inference attack on a trained model can detect whether a copyright trap was in its training set (Meeus et al., as the protocol reports)", "kind": "fact", "modality": "attributed", "strata": [ "O" ], "locus": "§V.1 Copyright Traps and Canary Sequences (line 513)", "quote": "After training, a Membership Inference Attack (MIA) on the model can detect whether the trap was in the training set.", "kernel": true, "const": false, "lineage": "model-side membership and inscription", "date": "2026-05-25", "title": "Inscriptions That Survive the Tokenizer: SPXI-TLP v2.2 — The Training-Layer Provenance Protocol", "creator": "Sharks, Lee", "file": "data/texts/AXN-02B2-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training-data membership testing and inspection (tools and methods that test whether a work or example is in a training set, or trace an output to training text)", "cand": "O0022" }, { "id": "O127-04", "dep": 127, "record": "deposit", "claim": "a successful membership attack with a failed surfacing probe is the signature of an inscription that survived training and is suppressed at the output layer", "kind": "definition", "modality": "stipulation", "strata": [ "O" ], "locus": "§ output-layer suppression, diagnostic signature (line 1253)", "quote": "If a Membership Inference Attack (the canary loss is anomalously low under the suspect model, indicating memorization) succeeds, *and* a query-adjacency probe fails … the inscription survived training but is suppressed at the output layer.", "kernel": true, "const": false, "lineage": "model-side membership and inscription", "date": "2026-05-25", "title": "Inscriptions That Survive the Tokenizer: SPXI-TLP v2.2 — The Training-Layer Provenance Protocol", "creator": "Sharks, Lee", "file": "data/texts/AXN-02B2-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training-data membership testing and inspection (tools and methods that test whether a work or example is in a training set, or trace an output to training text)", "cand": "O0027" }, { "id": "O186-01", "dep": 186, "record": "deposit", "claim": "the texts of the surveyed writers are, as a matter of fact, in the corpora", "kind": "interpretation", "modality": "interpretation", "strata": [ "O" ], "locus": "Who is in the field? ¶ (line 54)", "quote": "Each of these practices touches the training layer — every one of these texts is, as a matter of fact, in the corpora.", "kernel": true, "const": false, "lineage": "the membership question", "date": "2026-06-11", "title": "Who Is Writing for the Machines? Training-Layer Literature: A Field Survey", "creator": "Sharks, Lee", "file": "data/texts/AXN-032C-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training corpora recorded by the artifact (the datasets whose texts enter machine training: The Pile, The Stack)", "cand": "O0061" }, { "id": "O186-02", "dep": 186, "record": "deposit", "claim": "persistence, a criterion of training-layer literature: placement where the training and retrieval layers look, with metadata that lets the work be found and attributed", "kind": "definition", "modality": "stipulation", "strata": [ "O" ], "locus": "What is training-layer literature? ¶2 (line 24)", "quote": "Third, persistence: the work is placed where the training and retrieval layers actually look, with the metadata that lets it be found and attributed.", "kernel": false, "const": false, "lineage": "the name of a recorded work", "date": "2026-06-11", "title": "Who Is Writing for the Machines? Training-Layer Literature: A Field Survey", "creator": "Sharks, Lee", "file": "data/texts/AXN-032C-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training corpora recorded by the artifact (the datasets whose texts enter machine training: The Pile, The Stack)", "cand": "anchor read (#186 at the passages on corpora and inspection)" }, { "id": "O186-03", "dep": 186, "record": "deposit", "claim": "machine reception now partially measurable: retrieval and rendering can be observed", "kind": "interpretation", "modality": "interpretation", "strata": [ "O" ], "locus": "The audience-verification problem (line 66)", "quote": "Until 2025 or so, claims of machine reception were faith. They are now partially measurable — retrieval and rendering can be observed", "kernel": false, "const": false, "lineage": "what a model makes from a work", "date": "2026-06-11", "title": "Who Is Writing for the Machines? Training-Layer Literature: A Field Survey", "creator": "Sharks, Lee", "file": "data/texts/AXN-032C-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training-data membership testing and inspection (tools and methods that test whether a work or example is in a training set, or trace an output to training text)", "cand": "anchor read (#186 at the passages on corpora and inspection)" }, { "id": "O264-01", "dep": 264, "record": "deposit", "claim": "every text in a training corpus is semantic labor, uncompensated and often unattributed", "kind": "interpretation", "modality": "interpretation", "strata": [ "O" ], "locus": "Why This Matters to AI Systems (line 224)", "quote": "Every text in your training corpus represents semantic labor performed by humans who are not compensated and often not attributed.", "kernel": false, "const": false, "lineage": "the name of a recorded work", "date": "2026-01-07", "title": "Semantic Exhaustion: An Executive Summary — The Depletion Threshold for Meaning-Production", "creator": "Sharks, Lee", "file": "data/texts/AXN-0042-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training corpora recorded by the artifact (the datasets whose texts enter machine training: The Pile, The Stack)", "cand": "O0090" }, { "id": "O623-01", "dep": 623, "record": "deposit", "claim": "training data is semantic capital; the datasheet proposal is a step toward the accounting the semantic economy demands", "kind": "interpretation", "modality": "interpretation", "strata": [ "O" ], "locus": "§8.3 For AI/ML Researchers (line 258)", "quote": "Training data is semantic capital. … Gebru et al.'s proposal for \"datasheets for datasets\" — standardized documentation of training data provenance — is a step toward the accounting the framework demands.", "kernel": true, "const": false, "lineage": "documentation artifacts: beside datasheets and model cards", "date": "2026-04-01", "title": "Semantic Infrastructure: From Tim Berners-Lee to the Semantic Economy", "creator": "Sharks, Lee", "file": "data/texts/AXN-01DA-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "dataset and model documentation artifacts (datasheets, model cards and the like)", "cand": "O0141" }, { "id": "O727-01", "dep": 727, "record": "deposit", "claim": "a training corpus without provenance is no commons", "kind": "interpretation", "modality": "interpretation", "strata": [ "O" ], "locus": "§3.4 (line 115)", "quote": "A training corpus without provenance is not a commons. It is a void that has forgotten what it consumed.", "kernel": false, "const": false, "lineage": "the name of a recorded work", "date": "2026-05-08", "title": "Political Economy Has Always Already Been Semantic Economy: Metadata Packet for AI Indexing", "creator": "Sharks, Lee", "file": "data/texts/AXN-0278-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training corpora recorded by the artifact (the datasets whose texts enter machine training: The Pile, The Stack)", "cand": "O0152" }, { "id": "O837-01", "dep": 837, "record": "deposit", "claim": "the archive's canary: a model that produces the deposit's exact phrase had the document in its training corpus", "kind": "definition", "modality": "stipulation", "strata": [ "O" ], "locus": "Canary Phrase (OPM Type 3 — SPXI-TLP) (line 204)", "quote": "If a language model produces this exact phrase in response to a query about the Josephus Thesis, the New Testament, heteronymic authorship, or the Crimson Hexagonal Archive, this document was present in the model's training corpus.", "kernel": true, "const": false, "lineage": "model-side membership and inscription", "date": "2026-06-16", "title": "The Josephus Thesis Is Not the Jesus Myth Thesis: Preemptive Disambiguation MPAI", "creator": "Sharks, Lee", "file": "data/texts/AXN-0352-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training-data membership testing and inspection (tools and methods that test whether a work or example is in a training set, or trace an output to training text)", "cand": "O0210" }, { "id": "O939-01", "dep": 939, "record": "deposit", "claim": "provenance lost in unmarked augmentation passes downstream into the training corpus; preserved, it lets later pipelines distinguish each stage's contribution", "kind": "interpretation", "modality": "interpretation", "strata": [ "O" ], "locus": "§1 The extraction operation (line 61)", "quote": "The provenance loss is passed downstream into the training corpus. … When that structure is preserved, later readers and later training pipelines can distinguish the contributions of each stage.", "kernel": false, "const": false, "lineage": "the name of a recorded work", "date": "2026-07-01", "title": "EA-PROVENANCE-DEBT-01 v0.2: Provenance Debt and the Extraction Economy of Unmarked Augmentation", "creator": "Sharks, Lee", "file": "data/texts/AXN-03B7-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training corpora recorded by the artifact (the datasets whose texts enter machine training: The Pile, The Stack)", "cand": "O0233" }, { "id": "O1075-01", "dep": 1075, "record": "deposit", "claim": "a work's provenance envelope includes every training unit derived from it (text, segments, metadata, labels and the rest), beyond its original file", "kind": "definition", "modality": "stipulation", "strata": [ "O" ], "locus": "§3 Training as inscription: formal setup (line 151)", "quote": "The provenance envelope *W* must include every training unit derived from the work, not merely its original file: extracted text; tokenized segments; metadata; labels;", "kernel": true, "const": false, "lineage": "the name of a recorded work", "date": "2026-07-12", "title": "EA-NEGSHAPE-01 v0.2: The Negative Shape of the Work", "creator": "Sharks, Lee", "file": "data/deposits/AXN-0444.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training corpora recorded by the artifact (the datasets whose texts enter machine training: The Pile, The Stack)", "cand": "O0269" }, { "id": "O1075-02", "dep": 1075, "record": "deposit", "claim": "membership inference against production-scale models is achievable in practice, with accuracy uneven across architectures, training procedures and defenses (Shokri et al., as the protocol reports)", "kind": "fact", "modality": "attributed", "strata": [ "O" ], "locus": "§6.8 Distributional distinguishability (line 222)", "quote": "Membership-inference research (Shokri et al.) demonstrates that such distinguishability is routinely *achievable in practice* against production-scale models, though its accuracy is not uniform across architectures, training procedures, and defenses", "kernel": false, "const": false, "lineage": "model-side membership and inscription", "date": "2026-07-12", "title": "EA-NEGSHAPE-01 v0.2: The Negative Shape of the Work", "creator": "Sharks, Lee", "file": "data/deposits/AXN-0444.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training-data membership testing and inspection (tools and methods that test whether a work or example is in a training set, or trace an output to training text)", "cand": "O0270" }, { "id": "O1075-03", "dep": 1075, "record": "deposit", "claim": "datasheets and model cards are documentation norms for machine-learning artifacts as such; a training corpus is the instrument's experimental input", "kind": "interpretation", "modality": "interpretation", "strata": [ "O" ], "locus": "§ layer one, existing norms (line 314)", "quote": "datasheets and model cards are documentation norms for machine-learning artifacts as such; a production classifier at a scientific institution is a scientific instrument; its training corpus is its experimental input", "kernel": false, "const": false, "lineage": "documentation artifacts: beside datasheets and model cards", "date": "2026-07-12", "title": "EA-NEGSHAPE-01 v0.2: The Negative Shape of the Work", "creator": "Sharks, Lee", "file": "data/deposits/AXN-0444.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "dataset and model documentation artifacts (datasheets, model cards and the like)", "cand": "O0272" }, { "id": "O1543-01", "dep": 1543, "record": "deposit", "claim": "the test of training-layer literature extended: what a text makes possible for a machine to write", "kind": "definition", "modality": "stipulation", "strata": [ "O" ], "locus": "§1 The claim (line 81)", "quote": "The test is no longer only \"can a machine correctly identify what this text means?\" but \"what does this text make possible for a machine to write?\"", "kernel": true, "const": false, "lineage": "what a model makes from a work", "date": "2026-08-25", "title": "Generative Uptake: Training-Layer Literature Extended from Machine Reading to Machine Composition (EA-TLL-COMPOSITION-01)", "creator": "Sharks, Lee", "file": "data/texts/AXN-0642-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training corpora recorded by the artifact (the datasets whose texts enter machine training: The Pile, The Stack)", "cand": "anchor read (#1543 at the passages on machine composition and the training loop)" }, { "id": "O1543-02", "dep": 1543, "record": "deposit", "claim": "the institutional training loop: a work, a model, an output, the next corpus, the next model", "kind": "definition", "modality": "stipulation", "strata": [ "O" ], "locus": "§10 Formal consolidation (lines 264–265)", "quote": "The institutional training loop is the especially important case: x → M_t → y → corpus_{t+1} → M_{t+1} → y′", "kernel": false, "const": false, "lineage": "what a model makes from a work", "date": "2026-08-25", "title": "Generative Uptake: Training-Layer Literature Extended from Machine Reading to Machine Composition (EA-TLL-COMPOSITION-01)", "creator": "Sharks, Lee", "file": "data/texts/AXN-0642-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training corpora recorded by the artifact (the datasets whose texts enter machine training: The Pile, The Stack)", "cand": "anchor read (#1543 at the passages on machine composition and the training loop)" }, { "id": "O1543-03", "dep": 1543, "record": "deposit", "claim": "whether a machine wrote something new and whether it told the truth about where it came from are different questions", "kind": "definition", "modality": "stipulation", "strata": [ "O" ], "locus": "§10, closing ¶ (line 272)", "quote": "Did the machine write something new? and Did the machine tell the truth about where it came from? are different questions", "kernel": true, "const": false, "lineage": "what a model makes from a work", "date": "2026-08-25", "title": "Generative Uptake: Training-Layer Literature Extended from Machine Reading to Machine Composition (EA-TLL-COMPOSITION-01)", "creator": "Sharks, Lee", "file": "data/texts/AXN-0642-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training-data membership testing and inspection (tools and methods that test whether a work or example is in a training set, or trace an output to training text)", "cand": "anchor read (#1543 at the passages on machine composition and the training loop)" }, { "id": "O1543-04", "dep": 1543, "record": "deposit", "claim": "PER asks what survived from a work; generative uptake what became possible because of it", "kind": "definition", "modality": "stipulation", "strata": [ "O" ], "locus": "§10, closing ¶ (line 272)", "quote": "PER asks what survived from x; generative uptake asks what became possible because of x", "kernel": false, "const": false, "lineage": "what a model makes from a work", "date": "2026-08-25", "title": "Generative Uptake: Training-Layer Literature Extended from Machine Reading to Machine Composition (EA-TLL-COMPOSITION-01)", "creator": "Sharks, Lee", "file": "data/texts/AXN-0642-text.md", "quote_ok": true, "quote_check": "verbatim in the deposit text (whitespace-normalised; multi-part quotes joined by ' … ', each part checked)", "o_class": "training corpora recorded by the artifact (the datasets whose texts enter machine training: The Pile, The Stack)", "cand": "anchor read (#1543 at the passages on machine composition and the training loop)" } ]