{
  "purpose": "Intentionally flawed teaching fixtures for the viewparquet LLM data audit article",
  "generated_on": "2026-08-24",
  "generator": "scripts/generate_llm_audit_fixtures.py",
  "duckdb_version": "v1.5.3",
  "evidence_artifacts": [
    {
      "name": "generate_llm_audit_fixtures.py",
      "role": "fixture_generator",
      "bytes": 12145,
      "sha256": "0b662dbaa6922ef0f9a5474e17d72b560ba5bca3f6f7087eba6fd9225cf5c224"
    },
    {
      "name": "llm-training-audit.sql",
      "role": "training_sql_bundle",
      "bytes": 2492,
      "sha256": "fec81497082881dd11f89f6252b0b4774404a9b5f33358e18b59f9a39bd59bb4"
    },
    {
      "name": "rag-embedding-audit.sql",
      "role": "rag_sql_bundle",
      "bytes": 3198,
      "sha256": "32eec55aee9e6022bada8a3f03714de0e49c60dd7d32497cc2e612de18c46241"
    }
  ],
  "limitations": [
    "Synthetic teaching data; not representative of a production corpus",
    "Example tokenizer IDs and embedding values do not belong to a real model",
    "Passing these checks does not establish consent, legal fitness, semantic quality, or model suitability"
  ],
  "files": [
    {
      "name": "llm-training-audit-fixture.parquet",
      "bytes": 2191,
      "sha256": "5e8e4753b1178ece799ae5767b3ed5f92173c50311a7f4ef645b71885ddad1be",
      "schema": [
        {
          "name": "example_id",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "source_document_id",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "split",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "text",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "source_uri",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "license",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "tokenizer_revision",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "token_ids",
          "type": "INTEGER[]",
          "nullable": "YES"
        }
      ],
      "expected_findings": {
        "rows": 8,
        "missing_example_ids": 0,
        "duplicate_id_excess_rows": 1,
        "blank_text_rows": 1,
        "invalid_split_rows": 1,
        "missing_source_rows": 1,
        "missing_license_rows": 1,
        "empty_token_rows": 1,
        "unexpected_tokenizer_rows": 1,
        "cross_split_text_groups": 1
      }
    },
    {
      "name": "rag-embedding-audit-fixture.parquet",
      "bytes": 2262,
      "sha256": "261317be2cc1b889d013ec2a1311ff53e2d0ca83db1a9c0b696a5a7abd0ae944",
      "schema": [
        {
          "name": "chunk_id",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "source_document_id",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "chunk_index",
          "type": "INTEGER",
          "nullable": "YES"
        },
        {
          "name": "text",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "text_revision",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "source_uri",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "license",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "embedding_model_revision",
          "type": "VARCHAR",
          "nullable": "YES"
        },
        {
          "name": "embedding",
          "type": "DOUBLE[]",
          "nullable": "YES"
        }
      ],
      "expected_findings": {
        "rows": 6,
        "missing_chunk_ids": 0,
        "duplicate_chunk_id_excess_rows": 0,
        "missing_embedding_rows": 0,
        "wrong_dimension_rows": 1,
        "non_finite_rows": 1,
        "zero_vector_rows": 1,
        "unexpected_model_rows": 1
      }
    }
  ]
}