Implementation of Flow Policy Optimization (FPO)

[NeurIPS 2025] An official implementation of Flow-GRPO: Training Flow Matching Models via Online RL

Directly Aligning the Full Diffusion Trajectory with Fine-Grained Human Preference

[ICCV2025] UniPortrait: A Unified Framework for Identity-Preserving Single- and Multi-Human Image Personalization

[ICLR 2025] official implementation of "Diffusion-NPO: Negative Preference Optimization for Better Preference Aligned Generation of Diffusion Models"

[Preprint] On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification.

Focus on prompting and generating

Qwen-Image is a powerful image generation foundation model capable of complex text rendering and precise image editing.

CosmicMan: A Text-to-Image Foundation Model for Humans (CVPR 2024)

An official pytorch implementation of "MoLE: Enhancing Human-centric Text-to-image Diffusion via Mixture of Low-rank Experts"

[CVPR 2025 Best Paper Award] VGGT: Visual Geometry Grounded Transformer

Original reference implementation of "3D Gaussian Splatting for Real-Time Radiance Field Rendering"

[ICML 2025] Official Code for "ADHMR: Aligning Diffusion-based Human Mesh Recovery via Direct Preference Optimization"

An open-source implementation for training LLaVA-NeXT.

[CVPR 2025] UniPose: A Unified Multimodal Framework for Human Pose Comprehension, Generation and Editing

Vector (and Scalar) Quantization, in Pytorch

✨✨Latest Advances on Multimodal Large Language Models

Official repository of 'Visual-RFT: Visual Reinforcement Fine-Tuning' & 'Visual-ARFT: Visual Agentic Reinforcement Fine-Tuning'’

The Next Step Forward in Multimodal LLM Alignment

Aligning LMMs with Factually Augmented RLHF

Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)

Solve Visual Understanding with Reinforced VLMs

A straightforward PPO implementation

[ECCV 2024] Official Code for "WHAC: World-grounded Humans and Cameras"

MSCS 申请季总结

[T-PAMI] Official code for "SMPLest-X: Ultimate Scaling for Expressive Human Pose and Shape Estimation"

Official implementation of “JOTR: 3D Joint Contrastive Learning with Transformers for Occluded Human Mesh Recovery“

The repository contains the official implementation of "DPMesh: Exploiting Diffusion Prior for Occluded Human Mesh Recovery", CVPR 2024