Constarium
← Search

Data · dataset · 2026

Replication Data for: Fine-Tuning Large Language Models for JSON Schema Generation and Modification

Listed in DaRUS

This dataset contains a training dataset consisting of natural language description-JSON Schema pairs, which can be used to fine-tune LLMs and an evaluation dataset, which can be used to assess the JSON Schema generation capability of LLMs.

Description

The evaluation dataset consists of five categories with prompts and their corresponding ground truth schemas. It also contains schemas generated by 5 models on the evaluation dataset.

Links

Where it is published

Catalogue records · 1

Topics

From keywords
Engineering · Humanities
Inferred from text
Data management and data science 70%
Provenance · 1 source records, 11 field assertions
SourceKeyLast seenRaw
DaRUSdoi:10.18419/DARUS-59149 d agoJSON v1
FieldAssertionExtractorEvidence
concepts[field].anzsrc:group:4605enrichment · darus uni stuttgart detaxonomy-embedding@1.1.0title+keywords+description (70%)
concepts[field].dataverse_subject:computer-and-information-sciencesource · darus uni stuttgart deconnector:darus_uni_stuttgart_de@1.0.0/subjects
concepts[field].dataverse_subject:engineeringsource · darus uni stuttgart deconnector:darus_uni_stuttgart_de@1.0.0/subjects
concepts[field].local:field:engineeringmapping · darus uni stuttgart deconnector:darus_uni_stuttgart_de@1.0.0/subjects
concepts[field].local:field:humanitiesmapping · darus uni stuttgart deconnector:darus_uni_stuttgart_de@1.0.0/subjects
created_datesource · darus uni stuttgart deconnector:darus_uni_stuttgart_de@1.0.0
descriptionsource · darus uni stuttgart deconnector:darus_uni_stuttgart_de@1.0.0/description
publication_datesource · darus uni stuttgart deconnector:darus_uni_stuttgart_de@1.0.0
titlesource · darus uni stuttgart deconnector:darus_uni_stuttgart_de@1.0.0/name
updated_datesource · darus uni stuttgart deconnector:darus_uni_stuttgart_de@1.0.0
version_labelsource · darus uni stuttgart deconnector:darus_uni_stuttgart_de@1.0.0