Video Depth without Video Models

HunyuanVideo: A Systematic Framework For Large Video Generation Model

Unified KV Cache Compression Methods for Auto-Regressive Models

Implementation of SmoothCache, a project aimed at speeding-up Diffusion Transformer (DiT) based GenAI models with error-guided caching.

LLM KV cache compression made easy

Official Implementation for our NeurIPS 2024 paper, "Don't Look Twice: Run-Length Tokenization for Faster Video Transformers".

一个关于 3d gaussian splat 的命令行小工具盒，ply 转 splat，或者 splat 转 ply 等命令行操作。convert ply to splat,convert splat to ply,etc.

The current generation of LEO constellations assumes that each LEO satellite covers a ground area regardless of the number of end users or terminals operating in the same area. The satellite soluti…

My learning notes/codes for ML SYS.

Computer networking PA and Wireshark solutions (eight Edition)

A framework for serving and evaluating LLM routers - save LLM costs without compromising quality!

No Pose, No Problem: Surprisingly Simple 3D Gaussian Splats from Sparse Unposed Images

Gaga: Group Any Gaussians via 3D-aware Memory Bank

ReFusion: 3D Reconstruction in Dynamic Environments for RGB-D Cameras Exploiting Residuals

[NeurIPS 2024] Depth Anything V2. A More Capable Foundation Model for Monocular Depth Estimation

一个超轻量级、可以在移动端实时运行的数字人模型

这是一个高效，快捷的arXiv论文爬虫，它可以将指定时间范围，指定主题，包含指定关键词的论文信息爬取到本地，并且将其中的标题和摘要翻译成中文。

Tool for producing high quality forecasts for time series data that has multiple seasonality with linear or non-linear growth.

What would you do with 1000 H100s...

Semantic cache for LLMs. Fully integrated with LangChain and llama_index.

A one stop repository for generative AI research updates, interview resources, notebooks and much more!

「大模型」3小时完全从0训练26M的小参数GPT，个人显卡即可推理训练！

《Real-Time Rendering 4th》 (RTR4) 中文翻译

The code for the paper "Reducing the Memory Footprint of 3D Gaussian Splatting"

We present Object Images (Omages): An homage to the classic Geometry Images.

xDiT: A Scalable Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelism

[CVPR 2024 Highlight] Feature 3DGS: Supercharging 3D Gaussian Splatting to Enable Distilled Feature Fields

A large-scale RWKV v6 inference with FLA . Capable of inference by combining multiple states(Pseudo MoE). Easy to deploy on docker. Supports true multi-batch generation and dynamic State switching.…

Learning Disentangled Avatars with Hybrid 3D Representations. (Face, Body, Hair and Clothing)