Delete croissant.json

croissant.json

@@ -1,151 +0,0 @@
-{
-  "@context": {
-    "@language": "en",
-    "@vocab": "https://schema.org/",
-    "cr": "http://mlcommons.org/croissant/",
-    "rai": "http://mlcommons.org/croissant/RAI/",
-    "dct": "http://purl.org/dc/terms/",
-    "sc": "https://schema.org/"
-  },
-  "@type": "sc:Dataset",
-  "cr:conformsTo": [
-    "http://mlcommons.org/croissant/1.0",
-    "http://mlcommons.org/croissant/RAI/1.0"
-  ],
-  "name": "SimplexTasks-12-data",
-  "description": "A frozen benchmark artifact for evaluating conformal prediction strategies on simplex-valued predictions (SimplexUQ). Contains 12 tasks: 6 synthetic data-generating processes (D1–D6) and 6 real-data domains (CIFAR-10 softmax, 20Newsgroups topics, Samson hyperspectral unmixing, PBMC pseudobulk, UTKFace label distribution learning, AffectiveText emotion compositions). All assets are frozen derived arrays and predictor outputs; restricted raw assets are intentionally excluded.",
-  "url": "https://huggingface.co/datasets/liangyou03/SimplexTasks-12-data",
-  "license": "https://huggingface.co/datasets/liangyou03/SimplexTasks-12-data/blob/main/ASSET_PROVENANCE_AND_TERMS.md",
-  "keywords": [
-    "conformal-prediction",
-    "uncertainty-estimation",
-    "simplex",
-    "benchmark",
-    "task-collection"
-  ],
-  "rai:dataCollection": "Synthetic data (D1–D6) was generated programmatically via code implementing predefined data-generating processes. Real-task derived data originates from the following publicly available sources: CIFAR-10 (Krizhevsky 2009), 20 Newsgroups (Lang 1995), Samson hyperspectral dataset (public remote sensing benchmark), PBMC single-cell RNA-seq (public genomics dataset), UTKFace (Zhang et al. 2017), and AffectiveText SemEval-2007 Task 14 (Strapparava & Mihalcea 2007). All real-task assets in this bundle are frozen derived arrays (softmax outputs, pseudobulk proportions, label distributions), not raw source data.",
-  "rai:dataCollectionType": "Programmatic generation (synthetic); derived from publicly available datasets (real tasks).",
-  "rai:dataCollectionMissingData": "No missing data. All frozen arrays are complete and validated as reported in VALIDATION_REPORT.json.",
-  "rai:dataCollectionRawData": "Raw source data for real tasks (images, text, gene expression matrices) is not included in this bundle. Only frozen derived arrays and predictor outputs are released. Refer to ASSET_PROVENANCE_AND_TERMS.md for provenance and access instructions for restricted assets.",
-  "rai:dataCollectionTimeframe": "Synthetic data generated in 2025. Real-task derived arrays produced in 2025 from publicly available datasets collected across 1995–2017.",
-  "rai:dataImputationProtocol": "No imputation applied. All arrays are complete.",
-  "rai:dataPreprocessingProtocol": "Synthetic data: generated directly from parametric DGPs with fixed random seeds. Real data: standard preprocessing per task (e.g., softmax normalization for CIFAR-10, pseudobulk aggregation for PBMC, label distribution normalization for UTKFace and AffectiveText). Full protocols documented in task_cards/ and BENCHMARK_PROTOCOL.md.",
-  "rai:dataManipulationProtocol": "Frozen arrays are produced by running trained predictors on held-out splits. No further manipulation is applied post-freeze. Split policies are defined in configs/.",
-  "rai:dataAnnotationProtocol": "No human annotation. Labels and targets are inherited from source datasets or derived algorithmically.",
-  "rai:dataAnnotationPlatform": "N/A",
-  "rai:dataAnnotationAnalysis": "N/A",
-  "rai:annotationsPerItem": "N/A",
-  "rai:annotatorDemographics": "N/A",
-  "rai:machineAnnotationTools": "Pretrained models used to produce predictor outputs (e.g., softmax classifiers). Details in task_cards/.",
-  "rai:dataBiases": "Real tasks inherit any biases present in their source datasets. UTKFace contains demographic attributes (age, gender, ethnicity) derived from face images; downstream use should consider known biases in facial attribute datasets. PBMC data reflects specific donor cohorts and may not generalize across populations. Synthetic tasks are unbiased by construction but may not reflect real-world distribution shift patterns.",
-  "rai:dataUseCases": "Intended exclusively for benchmarking and evaluating conformal prediction methods on simplex-valued (compositional) outputs. Not intended for production deployment, clinical decision-making, or any application involving real individuals.",
-  "rai:dataLimitations": "This bundle contains only frozen derived arrays, not raw data. Reproducibility of the full pipeline requires access to original source datasets (see ASSET_PROVENANCE_AND_TERMS.md). Coverage is limited to 12 tasks; generalization of benchmark conclusions to other domains should be made with caution.",
-  "rai:dataSocialImpact": "The benchmark is designed to improve evaluation rigor for uncertainty quantification methods. Improved conformal prediction methods may benefit high-stakes applications (e.g., clinical, scientific) by providing better-calibrated uncertainty estimates. No direct negative social impact is anticipated from this benchmark artifact.",
-  "rai:personalSensitiveInformation": "UTKFace derived arrays are based on face images with demographic labels (age, gender, ethnicity). This bundle contains only aggregated label distribution arrays, not raw face images or identifiable information. PBMC derived arrays are based on anonymized public single-cell RNA-seq data with no personally identifiable information.",
-  "rai:dataReleaseMaintenance": "The benchmark arrays are frozen at release. Bug fixes and provenance updates will be versioned on Hugging Face. Contact via the Hugging Face repository discussion page.",
-  "distribution": [
-    {
-      "@type": "cr:FileObject",
-      "@id": "repo",
-      "name": "repo",
-      "description": "Hugging Face repository root.",
-      "contentUrl": "https://huggingface.co/datasets/liangyou03/SimplexTasks-12-data",
-      "encodingFormat": "git+https",
-      "sha256": "https://github.com/mlcommons/croissant/issues/80"
-    },
-    {
-      "@type": "cr:FileSet",
-      "@id": "synthetic-data",
-      "name": "synthetic-data",
-      "description": "Frozen synthetic benchmark arrays for DGPs D1–D6.",
-      "containedIn": {"@id": "repo"},
-      "encodingFormat": "application/octet-stream",
-      "includes": "data/synthetic/**/*"
-    },
-    {
-      "@type": "cr:FileSet",
-      "@id": "derived-data",
-      "name": "derived-data",
-      "description": "Frozen derived arrays for 6 real tasks: CIFAR-10 softmax, 20Newsgroups topics, Samson unmixing, PBMC pseudobulk, UTKFace LDL, AffectiveText emotion compositions.",
-      "containedIn": {"@id": "repo"},
-      "encodingFormat": "application/octet-stream",
-      "includes": "data/derived/**/*"
-    },
-    {
-      "@type": "cr:FileSet",
-      "@id": "summaries",
-      "name": "summaries",
-      "description": "Metric summaries and figure/table inputs.",
-      "containedIn": {"@id": "repo"},
-      "encodingFormat": "application/octet-stream",
-      "includes": "data/summaries/**/*"
-    },
-    {
-      "@type": "cr:FileSet",
-      "@id": "task-cards",
-      "name": "task-cards",
-      "description": "One task card per synthetic or real task.",
-      "containedIn": {"@id": "repo"},
-      "encodingFormat": "text/plain",
-      "includes": "task_cards/**/*"
-    },
-    {
-      "@type": "cr:FileSet",
-      "@id": "configs",
-      "name": "configs",
-      "description": "Method specs, benchmark task specs, and split policies.",
-      "containedIn": {"@id": "repo"},
-      "encodingFormat": "text/plain",
-      "includes": "configs/**/*"
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "manifest",
-      "name": "MANIFEST.json",
-      "description": "Top-level manifest listing all benchmark assets.",
-      "containedIn": {"@id": "repo"},
-      "contentUrl": "MANIFEST.json",
-      "encodingFormat": "application/json"
-    },
-    {
-      "@type": "cr:FileObject",
-      "@id": "validation-report",
-      "name": "VALIDATION_REPORT.json",
-      "description": "Validation report for the benchmark release.",
-      "containedIn": {"@id": "repo"},
-      "contentUrl": "VALIDATION_REPORT.json",
-      "encodingFormat": "application/json"
-    }
-  ],
-  "cr:recordSet": [
-    {
-      "@type": "cr:RecordSet",
-      "@id": "tasks",
-      "name": "tasks",
-      "description": "The 12 benchmark tasks spanning 6 synthetic DGPs and 6 real-data domains.",
-      "cr:field": [
-        {
-          "@type": "cr:Field",
-          "@id": "tasks/task_id",
-          "name": "task_id",
-          "description": "Unique task identifier (e.g., D1, D2, cifar10_softmax).",
-          "dataType": "sc:Text"
-        },
-        {
-          "@type": "cr:Field",
-          "@id": "tasks/task_type",
-          "name": "task_type",
-          "description": "Whether the task is synthetic or real-data derived.",
-          "dataType": "sc:Text"
-        },
-        {
-          "@type": "cr:Field",
-          "@id": "tasks/domain",
-          "name": "domain",
-          "description": "Application domain (e.g., computer vision, NLP, genomics, remote sensing).",
-          "dataType": "sc:Text"
-        }
-      ]
-    }
-  ]
-}