Python trend monthly report, browse the most popular Python open source projects such as GitHub and Gitee on a monthly basis. The selected projects mainly refer to GitHub Trending , and some refer to Gitee and others. The rankings are in no particular order, and they are all relatively popular projects in the current month.
Selection formula = 70% GitHub Trending + 20% Gitee + 10% others
Pay attention to the WeChat public account and receive monthly reports
| # | July 2023 | July 8, 2023 |
|---|---|---|
| 1 | WormGPT | The evil version of ChatGPT is here | WormGPT | Automatically write viruses, phishing emails | Hacker special weapon arsenal | BEC attack | WormGPT. The security issues brought about by large language model technology are urgent. WormGPT has been advertised on underground forums as the perfect tool for executing sophisticated phishing campaigns and Business Email Compromise (BEC) attacks. For cyber attackers, using AI for BEC attacks has many advantages, including the use of impeccable syntax and lower barriers to entry for creating BEC campaigns. Threat actors can also use wormpt to automatically create high-quality phishing software that is so well crafted that recipients can easily be convinced. WormGPT is an AI model based on the GPT-3 architecture, but unlike other models, it does not apply any security measures and filters during training. This gives it greater flexibility when dealing with large amounts of hacker-related data. However, this lack of security filters also makes it possible for the model to be used in various malicious activities such as phishing, commercial email compromise, malware creation, scams, attacks, etc. How WormGPT Works: Like ChatGPT, WormGPT is trained using billions of data from a variety of sources, including legitimate websites, darknet forums, hacking manuals, malware samples, phishing templates, and more. WormGPT uses this data to generate convincing phishing emails, malware codes, hacking attacks, and other cybercriminal content designed to deceive humans and machines. |
| 2 | Final 2x | An open source, cross-platform image magnification tool, supports 9 models, supports CPU, GPU, and has a Chinese interface. It allows you to use multiple models to enlarge the image to get a clearer picture. Similar tools also include: Real-ESRGAN – a resolution magnification tool for photos and animated pictures IntegerScaler – makes old games and small pictures clearer, suitable for 2K/4K displays [Windows/Chrome/Firefox] Anime4KCPP – open source high-performance animation class Image and video resolution magnification tool [Windows/macOS/Linux/Android] Waifu2x-Extension-GUI – high-definition intelligent magnification of pictures, GIFs, and videos, supporting noise reduction [Windows] waifu2x – specifically for two-dimensional pictures, lossless magnification times, the effect is excellent [Web/Win] |
| 3 | CodeGeeX | A free AI programming assistant with functions similar to smart programming assistants such as Github Copilot, Codeium, CodeWhisperer, and Bito. Currently, CodeGeeX supports more than 10 mainstream programming languages such as Python, C++, Java, JavaScript, and Go. It is an all-round AI assistant for developers . It can help programmers only write comments to describe the required code functions, and the underlying large model of CodeGeeX can generate the required codes. CodeGeeX is a multi-programming language code generation pretrained model with 13 billion parameters. CodeGeeX is implemented using the Huawei MindSpore framework. CodeGeeX provides free VS Code and JetBrains IDEs plug-ins to assist users in writing code. Users can experience the new programming experience brought by CodeGeeX's code generation capability in their own IDE. |
| 4 | CodeGeeX2 | CodeGeeX2 is the second generation of the multilingual code generation model CodeGeeX (KDD'23). Different from the first generation of CodeGeeX (which is completely trained on the domestic Huawei Ascend chip platform), CodeGeeX2 is implemented based on the ChatGLM2 architecture and adding code pre-training. Thanks to the better performance of ChatGLM2, CodeGeeX2 has achieved performance improvements in multiple indicators (+107% > CodeGeeX ; only 6 billion parameters, which is nearly 10% of the StarCoder-15B with more than 15 billion parameters) |
| 5 | Built | A Python library for quickly creating graphical user interfaces (GUIs) for deployment of machine learning and deep learning models. Split from the technical side, it consists of three parts: front-end page + back-end interface + AI algorithm model reasoning Gradio has done one thing, which is to encapsulate these three parts into a python interface, and users can implement the encapsulated interface , to present the algorithm model trained by oneself to the public in the form of web service. The advantage of Gradio lies in its ease of use. The code structure is simpler than that of Streamlit. It only needs to simply define the input and output interfaces to quickly build a simple interactive page and deploy the model more easily. It is suitable for developers who have relatively simple scenarios and want to quickly deploy applications. |
| 6 | Real-Time-Voice-Cloning | An AI project that can clone someone's speaking voice. It only takes a few seconds of audio to imitate the human voice of the original audio. Real-time voice cloning can be realized, allowing users to generate a voice similar to the original voice by inputting text. |
| 7 | MetaGPT | It can simulate the development process of a software company and generate high-quality code. The MetaGPT model can abstract a number of different roles, including product managers, architects, project managers, engineers, etc., and can conduct internal supervision during code generation to improve the quality of the final output code. It can be said that a software development company is directly integrated into the into the model. MetaGPT can conduct "market research, competitive product analysis, architecture design" and other links with one click, and can combine real-world situations to intelligently analyze "what type of code should be generated, what kind of applicable people it is for, and what kind of functional requirements are needed" . |
| 8 | PaddleOCR | PaddlePaddle, an open source OCR project of Baidu's deep learning framework, aims to create a rich, leading, and practical OCR tool library to help users train better models and implement applications. PaddleOCR contains rich text detection, text recognition and end-to-end algorithms. |
| 9 | You-Get | A tiny command-line program that downloads media content (video, audio, images) from websites. The most convenient and powerful tool for downloading videos that I have learned so far. The specific uses are: Download videos/audios from popular websites such as Station B, iQiyi, and Netease Cloud Music. Download (of interest) images by scraping web pages. Stream online videos in a media player. Crawl up to 300 Bilibili barrage |
| 10 | gpt-code-ui | Let AI generate and execute code for you. An open source project of OpenAI's ChatGPT code interpreter, Code interpreter, which can generate and execute code at your request. Features: File upload and download function context-aware (it can refer to your previous messages) Generate code to run code (Python kernel) Model switching (support GPT-3.5 and GPT-4) |
| 11 | to me | (Mathematical Animation Engine) is an open source software library for creating mathematical animations. It was developed by Grant Sanderson (also founder of 3Blue1Brown) to help math educators and enthusiasts present math concepts and principles in a beautiful, visual way. |
| 12 | LLMSurvey | A collection of papers and resources related to large language models |
| 13 | h2ogpt | H2O公司开源的一个类似ChatGPT的应用。是基于EleutherAI发布的GPT-NeoX-20b模型微调的结果。使用的是H2O自己收集的数据集。 h2oGPT的最大特点是具有完全宽松、商业可用的代码、数据和模型。官引介绍：建立在大型语言模型(LLM)如GPT-4上的应用程序代表了AI的革命，因为它们在自然语言处理方面具有人类水平的能力。然而，它们也面临着许多重要风险，例如存在偏见、隐私或有害文本，以及未经授权的包含版权材料。我们介绍了h2oGPT，一个基于生成预训练Transformer(GPT)的开源代码仓库套件，用于创建和使用LLM。该项目的目标是创造世界最好的真正开源替代品，以与难以置信且不可阻挡的开源社区协作，我们开源了从7亿参数到400亿参数的多个优化过的h2oGPT模型，已完全开放Apache 2.0许可证下的商业使用。在我们发布的其中包含100%自然语言私人文档搜索。开源语言模型有助于推动AI发展，使其更加容易可得且可信。它们降低入门门槛，使个人和组织能够根据需求定制这些模型。这种开放性增加了创新、透明度和公平性。 |
| 14 | zulip | 远程团队合作的实时聊天软件，一个不错的开源的Slack 替代品 。Dropbox公司开发一款群组聊天软件，适用于团队沟通与协作。该类软件目前比较流行的是Slac；与之相比，Zulip还处于起步阶段。在Zulip社区，Slack也一直被作为竞争对手看待。国内也有类似软件，感兴趣的同学可以自行搜索。 |
| 15 | mypy | Python 中的静态类型检查器。无需运行代码即可发现程序中潜在的错误。还可以加入到 git hook 中，实现在提交代码前自动检查。Mypy 具有强大且易于使用的类型系统，具有现代功能，如类型推断、泛型、可调用类型、元组类型、联合类型和结构子类型。 |
| 16 | chinese-poetry | 最全中华古诗词数据库, 唐宋两朝近一万四千古诗人, 接近5.5万首唐诗加26万宋诗. 两宋时期1564位词人，21050首词。 |
| 17 | lean-side-bussiness | Lean Side Hustle: How Programmers Do Side Hustles Gracefully. This book expands the basic content of "How Programmers Earn Pocket Money Elegantly" and puts it into a larger sideline perspective; at the same time, it introduces the "Lean Entrepreneurship" process that has been verified by the Internet industry, and optimizes it for the sideline-specific "Lean Entrepreneurship" sideline” process. In the second part of the book, based on actual cases, the content of "independent development and monetization" and "online course monetization practice" was added. This book was written in December 2020, later than the one-person company methodology. After reading this book, you can read the one-person company methodology. |
| 18 | stack-roadmap | Full stack roadmap (sugar cube version). Fangtang's full-stack roadmap focuses on the issue of "how to make an idea into a product, commercialize it and obtain benefits", and selects from a large amount of learning content, trying to let readers complete in a short time, including The "minimum viable skill stack" including business model, design, development, operation, and growth helps readers complete the transformation "from a screw to a one-man enterprise". In the cycle of economic downturn, whether it is more choices of career development, or safer dual-line income from side business is extremely important. This is exactly the value that the roadmap expects to bring. |
| 19 | amis | amis is a low-code front-end framework that uses JSON configuration to generate pages, which can save the workload of page development and improve the efficiency of developing front-end pages. amis can complete complete functional development with only JSON configuration, including functions such as data acquisition, form submission, and verification. It has 100+ built-in UI components, which can meet the needs of various page components, and point out custom components. Amis has been widely used within Baidu, and more than 30,000 pages have been created in more than 4 years. From content review to machine management, from data analysis to model training, amis meets various page needs. |
Category: Monthly Ranking of Python Open Source ProjectsTags : Visual OCR Voice Programming Assistant WormGPT MetaGPT Voice Teamwork
Note: The current article will be updated from time to time. If you have better suggestions for this article, or new materials to recommend, you can click: Welcome to share excellent websites .