Stars
alexngng / CUDA-Learn-Note
Forked from xlite-dev/LeetCUDA🎉CUDA 笔记 / 高频面试题汇总 / C++笔记,个人笔记,更新随缘: sgemm、sgemv、warp reduce、block reduce、dot product、elementwise、softmax、layernorm、rmsnorm、hist etc.
A high-throughput and memory-efficient inference and serving engine for LLMs
AI GAL是专门为Galgame场景设计的程序,旨在让得每一名用户都能享受到独一无二的剧情。程序基于renpy框架开发
Firefly: 大模型训练工具,支持训练Qwen2.5、Qwen2、Yi1.5、Phi-3、Llama3、Gemma、MiniCPM、Yi、Deepseek、Orion、Xverse、Mixtral-8x7B、Zephyr、Mistral、Baichuan2、Llma2、Llama、Qwen、Baichuan、ChatGLM2、InternLM、Ziya2、Vicuna、Bloom等大模型
RAG-QA-Generator 是一个用于检索增强生成(RAG)系统的自动化知识库构建与管理工具。该工具通过读取文档数据,利用大规模语言模型生成高质量的问答对(QA对),并将这些数据插入数据库中,实现RAG系统知识库的自动化构建和管理。
Fine-tuning & Reinforcement Learning for LLMs. 🦥 Train Qwen3, Llama 4, DeepSeek-R1, Gemma 3, TTS 2x faster with 70% less VRAM.
Model Context Protocol Servers
PyTorch implementation of FractalGen https://arxiv.org/abs/2502.17437
[CVPR 2025 Oral] Reconstruction vs. Generation: Taming Optimization Dilemma in Latent Diffusion Models
采用 微拟物 FUI & Material 设计的向 bilibili 直播间发送 Vtuber 翻译、歌词的弹幕同传工具
[TPAMI 2025🔥] MagicTime: Time-lapse Video Generation Models as Metamorphic Simulators
An Industrial-Level Controllable and Efficient Zero-Shot Text-To-Speech System
✨ 易上手的多平台 LLM 聊天机器人及开发框架 ✨ 支持 QQ、QQ频道、Telegram、微信、企微、飞书、钉钉 | 知识库、MCP 服务器、OpenAI、DeepSeek、Gemini、硅基流动、月之暗面、Ollama、OneAPI、Dify
搜索、推荐、广告、用增等工业界实践文章收集(来源:知乎、Datafuntalk、技术公众号)
Typescript library for controlling DJI Osmo Action 3/4/5 and Pocket 3 using BLE (Bluetooth Low Energy)
HLS.js is a JavaScript library that plays HLS in browsers with support for MSE.
【Hello CTF】题目配套,免费开源的CTF入门教程,针对0基础新手编写,同时兼顾信息差的填补,对各阶段的CTFer都友好的开源教程,致力于CTF和网络安全的开源生态!
SQLI labs to test error based, Blind boolean based, Time based.
Fast and memory-efficient exact attention
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
A brand new web Visual Novel engine | 全新的网页端视觉小说引擎
The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface.
Google Research