Stars
- All languages
- ANTLR
- ASP
- Assembly
- Astro
- AutoHotkey
- Batchfile
- C
- C#
- C++
- CMake
- CSS
- Clojure
- CoffeeScript
- Cuda
- Cython
- Dart
- Dockerfile
- Elixir
- Emacs Lisp
- Erlang
- FLUX
- Forth
- Go
- Groovy
- HTML
- Handlebars
- Haskell
- Java
- JavaScript
- Jinja
- Jsonnet
- Jupyter Notebook
- Kotlin
- Less
- Lua
- MLIR
- Makefile
- Markdown
- Mojo
- Mustache
- Objective-C
- Odin
- PHP
- PLpgSQL
- Pascal
- Perl
- PowerShell
- Python
- Rich Text Format
- Roff
- Ruby
- Rust
- SCSS
- Scala
- Scheme
- Shell
- Smarty
- Svelte
- Swift
- Tcl
- TeX
- TypeScript
- Vim Script
- Vue
- Zig
🤗 PEFT: State-of-the-art Parameter-Efficient Fine-Tuning.
🤒 A modern alternative network traffic sniffer.
Comfortably monitor your Internet traffic 🕵️♂️
Instant, easy, and predictable development environments
Build resilient language agents as graphs.
CodeGraph - Tool that create a graph of code to show dependencies between code entities (methods, classes and etc).
《开源大模型食用指南》针对中国宝宝量身打造的基于Linux环境快速微调(全参数/Lora)、部署国内外开源大模型(LLM)/多模态大模型(MLLM)教程
手把手带你实战 Huggingface Transformers 课程视频同步更新在B站与YouTube
Flax is a neural network library for JAX that is designed for flexibility.
Build and share delightful machine learning apps, all in Python. 🌟 Star to support our work!
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)
面向开发者的 LLM 入门教程,吴恩达大模型系列课程中文版
Video+code lecture on building nanoGPT from scratch
Desktop tool for managing virtual machines via libvirt
Weaviate is an open-source vector database that stores both objects and vectors, allowing for the combination of vector search with structured filtering with the fault tolerance and scalability of …
Cross-platform text editor, written in Free Pascal
Aila(AI超元域): The premier AI integration tool for Windows, macOS, and Android. Ask once, get answers from 10+ AIs like ChatGPT, Gemini, Claude3, Copilot, Poe, perplexity and more. Features customiza…
Qwen2.5-VL is the multimodal large language model series developed by Qwen team, Alibaba Cloud.
RAGFlow is an open-source RAG (Retrieval-Augmented Generation) engine based on deep document understanding.
No fortress, purely open ground. OpenManus is Coming.
[NeurIPS'23 Oral] Visual Instruction Tuning (LLaVA) built towards GPT-4V level capabilities and beyond.