mlsanityData to 14 Oct 2026

Synthesis Without Training: An Inference-Only Pipeline for Tabular, Temporal, and Relational Synthetic Data

arXiv 2609.38414 ↗First posted 29 Sep 2026cs.LGcited by 0

On the map: Synthetic data and self-improvement for LLMs → Privacy-preserving tabular data synthesis also fits: LLM synthetic data generation

Abstract (arXiv)

Synthetic data generation is dominated by the fit-then-sample paradigm: a generative model is trained on a private dataset and then sampled from. Despite its widespread adoption, this paradigm faces three challenges: (1) a new training run is required for every dataset; (2) different data modalities, such as single tables, time series, and relational databases, require task-specific models and feature engineering; and (3) the resulting model is opaque, making its behavior under data constraints difficult to inspect. We propose GENSCRIPT, an inference-only pipeline that eliminates model training. GENSCRIPT computes a deterministic statistical profile of the source data (column types, ranges, missingness, categories, correlations, etc.) and passes it--rather than raw rows--to a language model to infer field semantics and cross-column integrity constraints. A coding agent then compiles the profile and constraints into an executable, auditable sampler. This unified approach supports single-table, temporal, and relational data without task-specific modeling. Across four single-table benchmarks, GENSCRIPT builds generators in 2 minutes and samples 50k rows within 6 seconds, while remaining within a few points of leading methods in marginal fidelity. Notably, it is the only method that perfectly preserves a 1-to-1 mapping between columns in the Adult dataset. On a smart-building dataset, it produces conditional time series that more closely match the real distribution than two baselines and perfectly preserves primary- and foreign-key relationships in the corresponding relational database.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Privacy-preserving tabular data synthesis, one of 13 directions in Synthetic data and self-improvement for LLMs. Builds on 19 of its 25 references that are in the map. First posted on arXiv 29 Sep 2026.

Builds on

19 of 25 references.

Closest ancestors · 2

Same direction, newest first

  1. 2024ClavaDDPM: Multi-relational Data Synthesis with Cluster-guided Diffusion Models28 May
  2. 2023TabuLa: Harnessing Language Models for Tabular Data Synthesis19 Oct

Other references · 17

  1. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  2. 2019Modeling Tabular data using Conditional GANcited by 699
  3. 2022PAL: Program-aided Language Modelscited by 652
  4. 2022TabDDPM: Modelling Tabular Data with Diffusion Modelscited by 377
  5. 2022Language Models are Realistic Tabular Data Generatorscited by 294
  6. 2022TabLLM: Few-shot Classification of Tabular Data with Large Language Modelscited by 277
  7. 2022Can Foundation Models Wrangle Your Data?cited by 175
  8. 2021CTAB-GAN: Effective Table Data Synthesizingcited by 156
  9. 2023REaLTabFormer: Generating Realistic Relational and Tabular Data using Transformerscited by 94
  10. 2022CTAB-GAN+: Enhancing Tabular Data Synthesiscited by 70
  11. 2024TabDiff: a Mixed-type Diffusion Model for Tabular Data Generationcited by 63
  12. 2023ArcheType: A Novel Framework for Open-Source Column Type Annotation using Large Language Modelscited by 15
5 more
  1. 2025TabularARGN: A Flexible and Efficient Auto-Regressive Framework for Generating High-Fidelity Synthetic Datacited by 14
  2. 2025Doubling Your Data in Minutes: Ultra-fast Tabular Data Generation via LLM-Induced Dependency Graphscited by 9
  3. 2024TimeAutoDiff: A Unified Framework for Generation, Imputation, Forecasting, and Time-Varying Metadata Conditioning of Heterogeneous Time Series Tabular Datacited by 5
  4. 2023IRG: Modular Synthetic Relational Database Generation with Complex Relational Schemascited by 3
  5. 2025TabTreeFormer: Tabular Data Generation Using Hybrid Tree-Transformercited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025FASTGEN: Fast and Cost-Effective Synthetic Tabular Data Generation with LLMs
  2. 2026Reasoning-Driven Synthetic Data Generation and Evaluation
  3. 2025BARE: Leveraging Base Language Models for Few-Shot Synthetic Data Generation
  4. 2025Boosting Statistic Learning with Synthetic Data from Pretrained Large Models
  5. 2025Synthetic Data Generation Using Large Language Models: Advances in Text and Code
  6. 2023Boosting Data Analytics With Synthetic Volume Expansion
  7. 2024DataGen: Unified Synthetic Dataset Generation via Large Language Models
  8. 2025Unlocking Post-hoc Dataset Inference with Synthetic Data