Easy Data Preparation with latest LLMs-based Operators and Pipelines.

Flame is an open-source multimodal AI system designed to translate UI design mockups into high-quality React code. It leverages vision-language modeling, automated data synthesis, and structured training workflows to bridge the gap between design and front-end development.

[CVPR 2020--Oral] CycleISP: Real Image Restoration via Improved Data Synthesis

Computer vision utils for Blender (generate instance annoatation, depth and 6D pose by one line code)

GraphGen: Enhancing Supervised Fine-Tuning for LLMs with Knowledge-Driven Synthetic Data Generation

[CVPR 2023] Label-Free Liver Tumor Segmentation

[CVPR 2024] Generalizable Tumor Synthesis - Realistic Synthetic Tumors in Liver, Pancreas, and Kidney

SWE-Factory: Your Automated Factory for Issue Resolution Training Data and Evaluation Benchmarks

[EMNLP 2025] Distill Visual Chart Reasoning Ability from LLMs to MLLMs

Official repository for Montessori-Instruct: Generate Influential Training Data Tailored for Student Learning [ICLR 2025]

Official Code for “EarthSynth: Generating Informative Earth Observation with Diffusion Models”

Source code for LDPTrace: Locally Differentially Private Trajectory Synthesis. VLDB 2023.

[Preprint] Deformation-Recovery Diffusion Model (DRDM): Instance Deformation for Image Manipulation and Synthesis

A data framework for music information retrieval focusing on electronic music.

Official implementaion of EMNLP 2022 paper "Generate, Discriminate, and Contrast: A Semi-Supervised Sentence Representation Learning Framework"

Boosting Document Intelligence

TabPFGen: Synthetic Tabular Data Generation with TabPFN

Blender Python Package for extracting internal data from blender scenes for 3d related data generation purposes.

A Label-Free and Data-Free Synthesis Engine and Training Framework for Vascular Segmentation of sOCT Data with PyTorch.

Code for ICCV'25 paper: Learn2Synth: Learning Optimal Data Synthesis using Hypergradients for Brain Image Segmentation