Miles-diffusion is an post-training framework for large-scale diffusion model training and production workloads, forked from and co-evolving with miles.

Official implementation of SimToolReal: An Object-Centric Policy for Zero-Shot Dexterous Tool Manipulation

DeepSeek Harness: Everything is a Plugin.

[ICML'26] Code and website for Self-Flow: Self-Supervised Flow Matching for Scalable Multi-Modal Synthesis

VerseCrafter: Dynamic Realistic Video World Model with 4D Geometric Control

World Modeling by Forecasting Vision Foundation Model Features

A self-improving RLM agent for coding workflows and long-running autonomous tasks.

[CVPR 2024] Official PyTorch implementation of SuGaR: Surface-Aligned Gaussian Splatting for Efficient 3D Mesh Reconstruction and High-Quality Mesh Rendering

[NeurIPS 2025] Pixel-Perfect Depth

[SIGGRAPH Asia 2026 Journal Track] InfiniSplat: Implicit Gaussian Decoding for Large-Baseline Monocular View Synthesis

[CVPR 2026] InfiniDepth: Arbitrary-Resolution and Fine-Grained Depth Estimation with Neural Implicit Fields

Run the full 2.78-trillion-parameter Kimi K3 model beyond available RAM by streaming activated weights directly from NVMe. A dependency-free, embeddable C inference engine.

Code for SM4RT: Learning Structured Motion Geometry for 4D Reconstruction

VoMP: Predicting Volumetric Mechanical Property Fields

Official Implementation of SIGGRAPH'26 paper "Immediate 3D Gaussian Splat Reconstruction of Unordered Input with Global Consistency"

[CVPR 2026] Fast-FoundationStereo: Real-Time Zero-Shot Stereo Matching

Rust bindings for the Python interpreter

Learn Rust with examples (Live code editor included)

Multimodal RL training framework for diffusion & omni models

Real-time global intelligence dashboard. AI-powered news aggregation, geopolitical monitoring, and infrastructure tracking in a unified situational awareness interface

A generalist video MLLM built for fine-grained motion, long-form reasoning, temporal grounding, and online proactive response.

A Metal-based CUDA compatibility framework for running Numba CUDA workloads on Apple Silicon.

SpaceXAI's coding agent harness and TUI. Fullscreen, mouse interactive, extensible.

[CVPR 2026 Hightlight] OmniVGGT: Omni-Modality Driven Visual Geometry Grounded Transformer

Differentiable rendering without approximation.

Vision as Unified Multimodal Generation

Run frontier MoE models on hardware you already own — pure C, zero deps, experts streamed from disk. Tiny engine, immense model. 🐦

[ICML 2026 Oral] Holi-Spatial: Evolving Video Streams into Holistic 3D Spatial Intelligence