Add Croissant metadata

croissant.json

@@ -0,0 +1,290 @@
+{
+  "@context": {
+    "@language": "en",
+    "@vocab": "https://schema.org/",
+    "sc": "https://schema.org/",
+    "cr": "http://mlcommons.org/croissant/",
+    "rai": "http://mlcommons.org/croissant/RAI/",
+    "dct": "http://purl.org/dc/terms/",
+    "citeAs": "cr:citeAs",
+    "conformsTo": "dct:conformsTo",
+    "data": {
+      "@id": "cr:data",
+      "@type": "@json"
+    },
+    "dataCollection": "rai:dataCollection",
+    "dataCollectionRawData": "rai:dataCollectionRawData",
+    "dataCollectionTimeframe": "rai:dataCollectionTimeframe",
+    "dataCollectionType": "rai:dataCollectionType",
+    "dataPreprocessingProtocol": "rai:dataPreprocessingProtocol",
+    "dataReleaseMaintenancePlan": "rai:dataReleaseMaintenancePlan",
+    "dataUseCases": "rai:dataUseCases",
+    "dataLimitations": "rai:dataLimitations",
+    "dataBiases": "rai:dataBiases",
+    "personalSensitiveInformation": "rai:personalSensitiveInformation",
+    "dataSocialImpact": "rai:dataSocialImpact",
+    "dataAnnotationProtocol": "rai:dataAnnotationProtocol",
+    "dataAnnotationPlatform": "rai:dataAnnotationPlatform",
+    "dataAnnotationAnalysis": "rai:dataAnnotationAnalysis",
+    "examples": {
+      "@id": "cr:examples",
+      "@type": "@json"
+    },
+    "extract": "cr:extract",
+    "field": "cr:field",
+    "fileObject": "cr:fileObject",
+    "fileProperty": "cr:fileProperty",
+    "fileSet": "cr:fileSet",
+    "format": "cr:format",
+    "includes": "cr:includes",
+    "jsonPath": "cr:jsonPath",
+    "path": "cr:path",
+    "recordSet": "cr:recordSet",
+    "source": "cr:source"
+  },
+  "@type": "sc:Dataset",
+  "name": "SFD Archive 1B-Token Source-Format Sample",
+  "alternateName": "sfd-archive-1b-source-format-sample",
+  "description": "Archive-wide SFD source-format analysis sample containing sanitized per-filing parse statistics, sampled accession metadata, aggregate summaries, and source-format figures.",
+  "url": "https://huggingface.co/datasets/sfd-anonymous/sfd-archive-1b-source-format-sample",
+  "sameAs": "https://huggingface.co/datasets/sfd-anonymous/sfd-archive-1b-source-format-sample",
+  "conformsTo": [
+    "http://mlcommons.org/croissant/1.1",
+    "http://mlcommons.org/croissant/RAI/1.0"
+  ],
+  "license": "https://creativecommons.org/licenses/by-nc/4.0/",
+  "creator": [
+    {
+      "@type": "Organization",
+      "name": "SFD Authors"
+    }
+  ],
+  "publisher": {
+    "@type": "Organization",
+    "name": "sfd-anonymous"
+  },
+  "datePublished": "2026-05-07",
+  "version": "1.0",
+  "keywords": [
+    "SEC",
+    "EDGAR",
+    "SFD",
+    "source-format analysis",
+    "finance",
+    "filings"
+  ],
+  "distribution": [
+    {
+      "@type": "cr:FileObject",
+      "@id": "repo_root",
+      "name": "Hugging Face dataset repository",
+      "contentUrl": "https://huggingface.co/datasets/sfd-anonymous/sfd-archive-1b-source-format-sample/tree/main",
+      "encodingFormat": "inode/directory"
+    },
+    {
+      "@type": "cr:FileObject",
+      "@id": "summary_json",
+      "name": "summary.json",
+      "contentUrl": "https://huggingface.co/datasets/sfd-anonymous/sfd-archive-1b-source-format-sample/resolve/main/summary.json",
+      "encodingFormat": "application/json",
+      "contentSize": 2922
+    },
+    {
+      "@type": "cr:FileObject",
+      "@id": "source_summary_checkpoint_json",
+      "name": "source_summary_checkpoint.json",
+      "contentUrl": "https://huggingface.co/datasets/sfd-anonymous/sfd-archive-1b-source-format-sample/resolve/main/source_summary_checkpoint.json",
+      "encodingFormat": "application/json",
+      "contentSize": 32190
+    },
+    {
+      "@type": "cr:FileObject",
+      "@id": "filing_stats_jsonl_gz",
+      "name": "filing_stats.jsonl.gz",
+      "contentUrl": "https://huggingface.co/datasets/sfd-anonymous/sfd-archive-1b-source-format-sample/resolve/main/filing_stats.jsonl.gz",
+      "encodingFormat": "application/octet-stream",
+      "contentSize": 3758901
+    },
+    {
+      "@type": "cr:FileObject",
+      "@id": "sample_manifest_jsonl_gz",
+      "name": "sample_manifest.jsonl.gz",
+      "contentUrl": "https://huggingface.co/datasets/sfd-anonymous/sfd-archive-1b-source-format-sample/resolve/main/sample_manifest.jsonl.gz",
+      "encodingFormat": "application/octet-stream",
+      "contentSize": 5522301
+    },
+    {
+      "@type": "cr:FileObject",
+      "@id": "source_format_live_csv",
+      "name": "source_format_live.csv",
+      "contentUrl": "https://huggingface.co/datasets/sfd-anonymous/sfd-archive-1b-source-format-sample/resolve/main/source_format_live.csv",
+      "encodingFormat": "text/csv",
+      "contentSize": 5945
+    },
+    {
+      "@type": "cr:FileObject",
+      "@id": "source_format_live_png",
+      "name": "source_format_live.png",
+      "contentUrl": "https://huggingface.co/datasets/sfd-anonymous/sfd-archive-1b-source-format-sample/resolve/main/source_format_live.png",
+      "encodingFormat": "image/png",
+      "contentSize": 125750
+    }
+  ],
+  "recordSet": [
+    {
+      "@type": "cr:RecordSet",
+      "@id": "filing_stats",
+      "name": "Per-filing parse statistics",
+      "description": "One row per attempted filing parse, with status, form metadata, token counts, source-format token counts, table estimates, and PDF page counts.",
+      "field": [
+        {
+          "@type": "cr:Field",
+          "name": "accession",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "filing_stats_jsonl_gz"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "name": "cik",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "filing_stats_jsonl_gz"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "name": "form_type",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "filing_stats_jsonl_gz"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "name": "date_filed",
+          "dataType": "sc:Date",
+          "source": {
+            "fileObject": {
+              "@id": "filing_stats_jsonl_gz"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "name": "status",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "filing_stats_jsonl_gz"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "name": "final_token_count",
+          "dataType": "sc:Integer",
+          "source": {
+            "fileObject": {
+              "@id": "filing_stats_jsonl_gz"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "name": "source_token_counts",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "filing_stats_jsonl_gz"
+            }
+          }
+        }
+      ]
+    },
+    {
+      "@type": "cr:RecordSet",
+      "@id": "sample_manifest",
+      "name": "Sample manifest",
+      "description": "Candidate filing manifest used to construct the source-format analysis sample.",
+      "field": [
+        {
+          "@type": "cr:Field",
+          "name": "accession",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "sample_manifest_jsonl_gz"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "name": "cik",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "sample_manifest_jsonl_gz"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "name": "form_type",
+          "dataType": "sc:Text",
+          "source": {
+            "fileObject": {
+              "@id": "sample_manifest_jsonl_gz"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "name": "date_filed",
+          "dataType": "sc:Date",
+          "source": {
+            "fileObject": {
+              "@id": "sample_manifest_jsonl_gz"
+            }
+          }
+        }
+      ]
+    }
+  ],
+  "dataCollectionRawData": "Public SEC EDGAR filings and derived benchmark artifacts built from those filings.",
+  "dataUseCases": [
+    "Financial language modeling and long-context pretraining research.",
+    "Document parsing, table extraction, OCR evaluation, and filing-grounded financial reasoning.",
+    "Dataset analysis and benchmark reproducibility for the associated NeurIPS submission."
+  ],
+  "dataLimitations": [
+    "SEC filings are not representative of all businesses or all global financial disclosures.",
+    "The artifacts reflect parser behavior and sampling choices at construction time.",
+    "Benchmarks are intended for research evaluation, not financial, legal, or investment advice."
+  ],
+  "dataBiases": [
+    "The archive over-represents U.S.-regulated registrants, public companies, investment funds, and entities required to file with the SEC.",
+    "Filing volume and token mass are skewed by form type, year, issuer size, and filing-agent workflows."
+  ],
+  "personalSensitiveInformation": [
+    "Underlying SEC filings are public regulatory records and may contain names, signatures, ownership disclosures, addresses, or other business-related personal information.",
+    "Synthetic benchmark transformations replace many entities and values, but users should still treat all artifacts as derived from public regulatory filings."
+  ],
+  "dataReleaseMaintenancePlan": "Artifacts are versioned on Hugging Face. Updates, replacements, or corrections will be documented in repository commits and dataset cards.",
+  "dataSocialImpact": "The datasets are intended to improve reproducible research on factual financial documents, while requiring care because downstream uses can affect financial analysis and compliance workflows.",
+  "dataCollection": "Filings were sampled from the SEC EDGAR archive and parsed through SFD to measure source-format composition and parser statistics.",
+  "dataCollectionType": [
+    "Document analysis",
+    "Software Collection",
+    "Secondary Data analysis"
+  ],
+  "dataCollectionTimeframe": "SEC EDGAR archive-wide sample; filing dates span multiple EDGAR generations.",
+  "dataPreprocessingProtocol": "Raw filings were downloaded, parsed into SFD MultiMarkdown/statistics, summarized by source format, and sanitized by removing parser stdout tails and local paths before release."
+}