Official PyTorch Implementation of "Diffusion Transformers with Representation Autoencoders"

Reasoning in Space via Grounding in the World

[NeurIPS 2025] E-MoFlow: Learning Egomotion and Optical Flow from Event Data via Implicit Regularization

[ICCV 2025] Official impl. of "MV-Adapter: Multi-view Consistent Image Generation Made Easy"

A simple state update rule to enhance length generalization for CUT3R

MapAnything: Universal Feed-Forward Metric 3D Reconstruction

From Flatland to Space (SPAR). Accepted to NeurIPS 2025 Datasets & Benchmarks. A large-scale dataset & benchmark for 3D spatial perception and reasoning in VLMs.

[ICCV 2025] Implementation for Describe Anything: Detailed Localized Image and Video Captioning

tiktoken is a fast BPE tokeniser for use with OpenAI's models.

Code for FastVGGT: Training-Free Acceleration of Visual Geometry Transformer

Official repo and evaluation implementation of VSI-Bench

🚀🚀 Efficient implementations of Native Sparse Attention

The official repo of Qwen-VL (通义千问-VL) chat & pretrained large vision language model proposed by Alibaba Cloud.

4DNeX: Feed-Forward 4D Generative Modeling Made Easy

🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.

本项目旨在分享大模型相关技术原理以及实战经验（大模型工程化、大模型应用落地）

Reference PyTorch implementation and models for DINOv3

[ICLR 2025, Oral] EmbodiedSAM: Online Segment Any 3D Thing in Real Time

✨✨Latest Advances on Multimodal Large Language Models

3D-R1: Enhancing Reasoning in 3D VLMs for Unified Scene Understanding

gpt-oss-120b and gpt-oss-20b are two open-weight language models by OpenAI

Open-source unified multimodal model

PyTorch implementation of paper "Neural Scene Flow Fields for Space-Time View Synthesis of Dynamic Scenes"

siiRL: Shanghai Innovation Institute RL Framework for Advanced LLMs and Multi-Agent Systems

Neural Scene Flow Prior (NeurIPS 2021 spotlight)

One-for-All Multimodal Evaluation Toolkit Across Text, Image, Video, and Audio Tasks

Qwen3-VL is the multimodal large language model series developed by Qwen team, Alibaba Cloud.

Official implementation of Spatial-MLLM: Boosting MLLM Capabilities in Visual-based Spatial Intelligence