amazon-science/tabsyn

How amazon-science/tabsyn works

TabSyn is a mixed-type tabular-data synthesis research repository. Its principal workflow prepares tabular datasets, trains a VAE to encode them into a continuous latent space, trains score-based diffusion over those embeddings, and samples synthetic tables. The repository also offers six baseline synthesizers, several synthetic-data evaluation metrics, and target-column imputation. The README documents these capabilities; some baseline sampling and evaluation internals were not excerpted, so their links are limited to documented workflows.

Main entry points: vae/main.py VAE training · tabsyn/main.py Latent diffusion training · impute.py Target imputation · data.py Shared data transforms

Components 27

Data preparation

Dataset download download_dataset.py
Dataset processing process_dataset.py
Shared data transforms src/data.py
Training preprocessing utils_train.py

TabSyn synthesis

VAE training tabsyn/vae/main.py
VAE encoder tabsyn/vae/model.py
Latent preparation tabsyn/latent_utils.py
Latent diffusion training tabsyn/main.py
Diffusion model tabsyn/model.py
TabSyn sampling tabsyn/sample.py

Baseline synthesis

SMOTE baseline baselines/smote/main.py
CoDi baseline baselines/codi/diffusion_discrete.py
GOGGLE baseline baselines/goggle/GoggleModel.py
GReaT baseline baselines/great/models/great.py
STaSy baseline baselines/stasy/main.py
TabDDPM baseline baselines/tabddpm/models/gaussian_multinomial_distribution.py

Evaluation

Synthetic data evaluation eval
Density metrics eval/eval_density.py
Quality metrics eval/eval_quality.py
ML efficiency eval/eval_mle.py
Privacy distance eval/eval_dcr.py
Two-sample detection eval/eval_detection.py

Imputation

Target imputation impute.py
Imputation evaluation eval_impute.py

Other components

Researcher
Raw tables
Synthetic tables

Connections 35

Researcher → Dataset download: requests datasets README.md
Dataset download → Raw tables: downloads README.md
Raw tables → Dataset processing: processes README.md
Dataset processing → Shared data transforms: prepares splits
Researcher → VAE training: starts training README.md
Shared data transforms → Training preprocessing: provides dataset data utils_train.py
VAE training → Training preprocessing: loads preprocessing tabsyn/vae/main.py
VAE training → VAE encoder: trains tabsyn/vae/main.py
VAE encoder → Latent preparation: encodes tables tabsyn/latent_utils.py
Researcher → Latent diffusion training: starts diffusion training README.md
Latent diffusion training → Latent preparation: loads embeddings tabsyn/main.py
Latent diffusion training → Diffusion model: trains tabsyn/main.py
Researcher → TabSyn sampling: requests synthesis README.md
TabSyn sampling → Synthetic tables: writes samples README.md
Researcher → SMOTE baseline: selects baseline README.md
Researcher → CoDi baseline: selects baseline README.md
Researcher → GOGGLE baseline: selects baseline README.md
Researcher → GReaT baseline: selects baseline README.md
Researcher → STaSy baseline: selects baseline README.md
Researcher → TabDDPM baseline: selects baseline README.md
SMOTE baseline → Synthetic tables: writes samples
CoDi baseline → Synthetic tables: writes samples
GOGGLE baseline → Synthetic tables: writes samples
GReaT baseline → Synthetic tables: writes samples
STaSy baseline → Synthetic tables: writes samples
TabDDPM baseline → Synthetic tables: writes samples
Researcher → Synthetic data evaluation: evaluates samples README.md
Synthetic tables → Density metrics: supplies samples README.md
Synthetic tables → Quality metrics: supplies samples README.md
Synthetic tables → ML efficiency: supplies samples README.md
Synthetic tables → Privacy distance: supplies samples README.md
Synthetic tables → Two-sample detection: supplies samples README.md
Researcher → Target imputation: requests imputation README.md
Target imputation → Shared data transforms: uses dataset data
Target imputation → Imputation evaluation: evaluates imputations README.md

More Python diagrams: snowfallingplum/shmt · ablab/isoquant · alan-cha/silhouette-card-maker · ark-electronics/ros2_px4_offboard_example · berriai/moyai

Drawn by GitDiagram on September 30, 2026 from the repository's file tree, README and source files. Read as Markdown ·