Stars
[ICLR 2026] pi-Flow: Policy-Based Few-Step Generation via Imitation Distillation
Code and website for Self-Flow: Self-Supervised Flow Matching for Scalable Multi-Modal Synthesis
Official PyTorch Implementation of "Diffusion Transformers with Representation Autoencoders"
Official implementation of UnifiedReward & [NeurIPS 2025] UnifiedReward-Think & UnifiedReward-Flex
GPU Cluster Monitoring (GCM): Large-Scale AI Research Cluster Monitoring
Diffusion-SDPO: Safeguarded Direct Preference Optimization for Diffusion Models
A PyTorch native platform for training generative AI models
We bring the spirit of nanogpt-speedrun into the omni-modal world
Official implementation of GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization
A Python-embedded DSL that makes it easy to write fast, scalable ML kernels with minimal boilerplate.
(CVPR 2025) From Slow Bidirectional to Fast Autoregressive Video Diffusion Models
Sharp Monocular View Synthesis in Less Than a Second
Official implementation of HPSv3: Towards Wide-Spectrum Human Preference Score (ICCV2025)
Ongoing research training transformer models at scale
DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.
Joint Distillation for Fast Likelihood Evaluation and Sampling in Flow-based Models
cuTile is a programming model for writing parallel kernels for NVIDIA GPUs
An extremely fast Python type checker and language server, written in Rust.
Directly Aligning the Full Diffusion Trajectory with Fine-Grained Human Preference
Provide with pre-build flash-attention package wheels on Linux and Windows platforms using GitHub Actions
[ICLR 2026] Self-Representation Alignment for Diffusion Transformers (SRA)
Official inference repo for FLUX.2 models