Lists (4)
Stars
Code release for book "Efficient Training in PyTorch"
🎨 ML Visuals contains figures and templates which you can reuse and customize to improve your scientific writing.
Experimental GStreamer plugin for encrypting / decrypting H264 streams with AES
A fast GPU memory copy library based on NVIDIA GPUDirect RDMA technology
Curated list of datasets and tools for post-training.
[NeurIPS 2020] Official code for the paper "DeepSVG: A Hierarchical Generative Network for Vector Graphics Animation". Includes a PyTorch library for deep learning with SVG data.
(CVPR 2025) Code of "Chat2SVG: Vector Graphics Generation with Large Language Models and Image Diffusion Models"
A benchmark dataset for evaluating LLM's SVG editing capabilities
A collection of modern/faster/saner alternatives to common unix commands.
PyTorch distributed training from scratch (for educational purposes only)
《Pytorch实用教程》（第二版）无论是零基础入门，还是CV、NLP、LLM项目应用，或是进阶工程化部署落地，在这里都有。相信在本书的帮助下，读者将能够轻松掌握 PyTorch 的使用，成为一名优秀的深度学习工程师。
SGLang is a fast serving framework for large language models and vision language models.
deepstream_tools will serve as a parent repo to hold various tools to be released for DeepStream SDK.
🗂️A file list/WebDAV program that supports multiple storages, powered by Gin and Solidjs. / 一个支持多存储的文件列表/WebDAV程序，使用 Gin 和 Solidjs。
flash attention tutorial written in python, triton, cuda, cutlass
Get up and running with Llama 3.3, DeepSeek-R1, Phi-4, Gemma 3, Mistral Small 3.1 and other large language models.
动手学Ollama，CPU玩转大模型部署，在线阅读地址：https://datawhalechina.github.io/handy-ollama/
Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)
AutoMQ is a stateless/diskless Kafka on S3. 10x Cost-Effective. No Cross-AZ Traffic Cost. Autoscale in seconds. Single-digit ms latency. Multi-AZ Availability.
Unofficial description of the CUDA assembly (SASS) instruction sets.
[NeurIPS 2024] Depth Anything V2. A More Capable Foundation Model for Monocular Depth Estimation
CV-CUDA™ is an open-source, GPU accelerated library for cloud-scale image processing and computer vision.
MNBVC(Massive Never-ending BT Vast Chinese corpus)超大规模中文语料集。对标chatGPT训练的40T数据。MNBVC数据集不但包括主流文化，也包括各个小众文化甚至火星文的数据。MNBVC数据集包括新闻、作文、小说、书籍、杂志、论文、台词、帖子、wiki、古诗、歌词、商品介绍、笑话、糗事、聊天记录等一切形式的纯文本中文数据。
fastllm是后端无依赖的高性能大模型推理库。同时支持张量并行推理稠密模型和混合模式推理MOE模型，任意10G以上显卡即可推理满血DeepSeek。双路9004/9005服务器+单显卡部署DeepSeek满血满精度原版模型，单并发20tps；INT4量化模型单并发30tps，多并发可达60+。