@context dict | @type string | @id string | name string | version string | dct:conformsTo string | description string | license string | sdLicense string | url string | isAccessibleForFree bool | inLanguage list | keywords list | additionalProperty list | distribution list | recordSet list |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
{
"@language": "en",
"@vocab": "http://schema.org/",
"sc": "http://schema.org/",
"cr": "http://mlcommons.org/croissant/",
"dct": "http://purl.org/dc/terms/"
} | sc:Dataset | https://github.com/Sonne1996/VEX/tree/main/dataset/additional/audit_dataset | VEX Audit Dataset | 1.0.0 | http://mlcommons.org/croissant/1.1 | Derived VEX dataset for annotation and audit analysis. It preserves the answer-level records together with the human annotation, consensus, and manual-audit columns that are removed from cleaner benchmark-facing releases. | https://creativecommons.org/licenses/by/4.0/ | https://creativecommons.org/licenses/by/4.0/ | https://github.com/Sonne1996/VEX/tree/main/dataset/additional/audit_dataset | true | [
{
"@type": "sc:Language",
"name": "German",
"alternateName": "de"
},
{
"@type": "sc:Language",
"name": "English",
"alternateName": "en"
}
] | [
"automatic short answer grading",
"ASAG",
"annotation audit",
"gold labels",
"educational assessment",
"VEX",
"predominantly German",
"mixed-language responses"
] | [
{
"@type": "sc:PropertyValue",
"name": "Language distribution",
"value": "Predominantly German. A small minority of responses may contain English or mixed German-English text."
},
{
"@type": "sc:PropertyValue",
"name": "Primary release artifact",
"value": "audit_dataset/audit_dataset.par... | [
{
"@type": "cr:FileObject",
"@id": "audit_dataset_parquet",
"name": "audit_dataset.parquet",
"description": "Derived VEX audit dataset artifact.",
"contentUrl": "https://raw.githubusercontent.com/Sonne1996/VEX/main/dataset/additional/audit_dataset/audit_dataset.parquet",
"encodingFormat": "a... | [
{
"@type": "cr:RecordSet",
"@id": "audit_records",
"name": "Audit Records",
"description": "One record per student answer in the VEX audit dataset. Each record combines anonymized identifiers, question and answer text, the normalized reference grade, split and provenance fields, and the preserved an... |
{
"@language": "en",
"@vocab": "http://schema.org/",
"sc": "http://schema.org/",
"cr": "http://mlcommons.org/croissant/",
"dct": "http://purl.org/dc/terms/"
} | sc:Dataset | https://github.com/Sonne1996/VEX/tree/main/dataset/additional/feedback_dataset | VEX Feedback Dataset | 1.0.0 | http://mlcommons.org/croissant/1.1 | Derived VEX dataset for feedback studies. It stores one row per answer-model pair, preserving the base answer context together with generated grades and written feedback. | https://creativecommons.org/licenses/by/4.0/ | https://creativecommons.org/licenses/by/4.0/ | https://github.com/Sonne1996/VEX/tree/main/dataset/additional/feedback_dataset | true | [
{
"@type": "sc:Language",
"name": "German",
"alternateName": "de"
},
{
"@type": "sc:Language",
"name": "English",
"alternateName": "en"
}
] | [
"automatic short answer grading",
"ASAG",
"feedback generation",
"educational assessment",
"LLM feedback",
"VEX",
"predominantly German",
"mixed-language responses"
] | [
{
"@type": "sc:PropertyValue",
"name": "Language distribution",
"value": "Predominantly German. A small minority of responses may contain English or mixed German-English text."
},
{
"@type": "sc:PropertyValue",
"name": "Primary release artifact",
"value": "feedback_dataset/merged_feedbac... | [
{
"@type": "cr:FileObject",
"@id": "feedback_dataset_parquet",
"name": "merged_feedback_long.parquet",
"description": "Derived VEX feedback dataset artifact in long format.",
"contentUrl": "https://raw.githubusercontent.com/Sonne1996/VEX/main/dataset/additional/feedback_dataset/merged_feedback_l... | [
{
"@type": "cr:RecordSet",
"@id": "feedback_records",
"name": "Feedback Records",
"description": "One record per student answer and model pair in the VEX feedback dataset. Each record combines anonymized identifiers, question and answer text, the human reference grade, split and provenance fields, a... |
{
"@language": "en",
"@vocab": "http://schema.org/",
"sc": "http://schema.org/",
"cr": "http://mlcommons.org/croissant/",
"dct": "http://purl.org/dc/terms/"
} | sc:Dataset | https://github.com/Sonne1996/VEX/tree/main/dataset/additional/teacher_selection_dataset | VEX Teacher Selection Dataset | 1.0.0 | http://mlcommons.org/croissant/1.1 | Derived VEX dataset for teacher-model selection on the gold subset. It combines answer-level gold context, audit metadata, candidate teacher outputs, and evaluator-model grades used to compare teacher candidates before silver-label generation. | https://creativecommons.org/licenses/by/4.0/ | https://creativecommons.org/licenses/by/4.0/ | https://github.com/Sonne1996/VEX/tree/main/dataset/additional/teacher_selection_dataset | true | [
{
"@type": "sc:Language",
"name": "German",
"alternateName": "de"
},
{
"@type": "sc:Language",
"name": "English",
"alternateName": "en"
}
] | [
"automatic short answer grading",
"ASAG",
"teacher model selection",
"silver labels",
"gold evaluation",
"VEX",
"predominantly German",
"mixed-language responses"
] | [
{
"@type": "sc:PropertyValue",
"name": "Language distribution",
"value": "Predominantly German. A small minority of responses may contain English or mixed German-English text."
},
{
"@type": "sc:PropertyValue",
"name": "Primary release artifact",
"value": "teacher_selection_dataset/gold_... | [
{
"@type": "cr:FileObject",
"@id": "teacher_selection_dataset_parquet",
"name": "gold_with_all_models.parquet",
"description": "Derived VEX teacher-selection dataset artifact.",
"contentUrl": "https://raw.githubusercontent.com/Sonne1996/VEX/main/dataset/additional/teacher_selection_dataset/gold_... | [
{
"@type": "cr:RecordSet",
"@id": "teacher_selection_records",
"name": "Teacher Selection Records",
"description": "One record per student answer in the VEX teacher-selection dataset. Each record combines anonymized identifiers, question and answer text, the normalized gold reference grade, split an... |
{
"@language": "en",
"@vocab": "http://schema.org/",
"sc": "http://schema.org/",
"cr": "http://mlcommons.org/croissant/",
"dct": "http://purl.org/dc/terms/"
} | sc:Dataset | https://github.com/Sonne1996/VEX/tree/main/dataset/additional/vex_metric_dataset | VEX Metric Dataset | 1.0.0 | http://mlcommons.org/croissant/1.1 | Derived VEX dataset used by the VEX evaluation pipeline. It combines the reference grade and answer-level context with prediction columns from all released model families for item-level and virtual-exam-level benchmarking. | https://creativecommons.org/licenses/by/4.0/ | https://creativecommons.org/licenses/by/4.0/ | https://github.com/Sonne1996/VEX/tree/main/dataset/additional/vex_metric_dataset | true | [
{
"@type": "sc:Language",
"name": "German",
"alternateName": "de"
},
{
"@type": "sc:Language",
"name": "English",
"alternateName": "en"
}
] | [
"automatic short answer grading",
"ASAG",
"virtual exams",
"benchmark evaluation",
"model predictions",
"VEX",
"predominantly German",
"mixed-language responses"
] | [
{
"@type": "sc:PropertyValue",
"name": "Language distribution",
"value": "Predominantly German. A small minority of responses may contain English or mixed German-English text."
},
{
"@type": "sc:PropertyValue",
"name": "Primary release artifact",
"value": "vex_metric_dataset/merged_model... | [
{
"@type": "cr:FileObject",
"@id": "vex_metric_dataset_parquet",
"name": "merged_model_predictions.parquet",
"description": "Derived VEX metric dataset artifact for benchmark evaluation.",
"contentUrl": "https://raw.githubusercontent.com/Sonne1996/VEX/main/dataset/additional/vex_metric_dataset/m... | [
{
"@type": "cr:RecordSet",
"@id": "vex_metric_records",
"name": "VEX Metric Records",
"description": "One record per student answer in the VEX metric dataset. Each record combines anonymized identifiers, question and answer text, the normalized human reference grade, split and provenance fields, and... |
No dataset card yet
- Downloads last month
- 45