quyapeng / sth-gpt

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

sth-gpt

🔥 Content

Technologies

NLP-GPT-Chatbot

Name Website Overview
Open AI Document
  • Open AI website provides multiple functional entries such as chat, speech to text, image generation, etc., as well as access to Open AI's API Key.
  • Open AI 官网,网站中提供多个功能入口,例如chat,speech to text,image generation等等,以及可以获取Open AI的API Key。
ChatGPT
  • The chatbot based on GPT proposed by open AI. ChatGPT Plus uses the GPT4 engine, which needs to be subscribed to use.
  • 由Open AI 基于GPT技术开发的聊天机器人,chatgpt plus 版本采用的是GPT4引擎,需要订阅购买才能使用。
ChatGPT Shortcut
  • This project compiles and summarizes various ChatGPT shortcut commands to double productivity. The shortcuts are organized by domain and function, and can filter prompts by tags, search for keywords, and copy them with one click.
  • 该项目整理并汇总了多种让生产力加倍的 ChatGPT 快捷指令,按照领域和功能分区,可对提示词(Prompt)进行标签筛选、关键词搜索和一键复制。
ChatGPT-Prompts
  • This project provides a rich and professional set of prompts for using ChatGPT.
  • 该项目为使用ChatGPT提供了丰富的,专业的prompts。
ChatGPT for Google
  • The web extension offered in the Chrome Web Store enables ChatGPT's responses to be displayed simultaneously in search engine results.
  • chrome应用商店提供的网页插件,可以在在搜索引擎结果中同时显示ChatGPT的回答。
GitHub Copilot
  • GitHub Copilot is an AI tool created by Github and OpenAI, which can automatically complete code and comments. Chat: Using GitHub Copilot to chat in the editor is similar to the experience of ChatGPT: introducing a chat interface into the editor, which focuses on developer solutions and integrates with VS Code and Visual Studio natively. This goes far beyond the suggested code. GitHub Copilot Chat is not just a chat window. It can recognize code typed by developers, display error messages, and deeply embed them in the IDE. Pull Request: It helps with PR and project management. CLI: It can write commands and loops, and throw obscure search flags to meet your queries. Doc: It uses a chat interface to provide AI-generated answers to document questions for users, including questions from developers about the language, framework, and technology they are using.
  • GitHub Copilot 是由 Github 和 OpenAI 创造的 AI 工具,可以自动补全代码和注释;Chat:使用 GitHub Copilot 聊天在编辑器中类似 ChatGPT 的体验:将一个聊天界面引入编辑器,该界面专注于开发人员方案,并与 VS Code 和 Visual Studio 本机集成。这远远超出了建议代码的作用。GitHub Copilot Chat不仅仅是一个聊天窗口。它可以识别开发人员键入的代码、显示的错误消息,并将其深度嵌入到 IDE中;Pull Request:对于 PR 和项目管理的帮助;CLI:它可以编写命令和循环,并抛出晦涩的查找标志以满足您的查询;Doc:它使用聊天界面为用户提供 AI 生成的对文档问题的回答,包括开发人员对他们正在使用的语言、框架和技术提出的问题。
Cursor
  • Cursor is designed to provide services such as code completion and error detection for programmers. By analyzing the code and its relevant context input by the user, Cursor can assist programmers in writing correct code more quickly, thereby improving programming efficiency and code quality.
  • Cursor旨在为编程人员提供代码补全和错误检测等服务。通过分析用户输入的代码和相关上下文,Cursor可以帮助程序员更快地编写正确的代码,从而提高编程效率和代码质量。
Monica
  • An AI assistant in the form of a web plugin; with a rich set of features; limited free usages.
  • 一个网页插件形式的AI助手;功能比较丰富;免费次数有限
New Bing
  • New Bing is a search engine developed by Microsoft, and its chat assistant is based on the GPT-4 model. After downloading the Edge Dev browser, New Bing can be accessed directly in the browser's sidebar. Additionally, it can read your web pages and clipboard, and can provide answers based on the content of the webpage.
  • New Bing 是微软的搜索引擎,聊天助手基于 GPT-4 模型。下载edge dev浏览器后,New Bing可以在浏览器的侧边栏直接访问,不仅如此,它还可以读取你的网页和剪贴板,可以针对网页的内容进行回答。
ChatGPT-Personal web page
  • Github project for one-click free deployment of your personal ChatGPT web application.
  • github项目,一键免费部署你的私人 ChatGPT 网页应用。
Chat-Academic
  • Using ChatGPT for polishing, translating and other editing tasks of a research paper draft.
  • 利用ChatGPT对论文初稿进行润色、翻译等。
Chat-Paper
  • Using ChatGPT to summarize a research paper and assist researchers with initial screening of the paper.
  • 通过ChatGPT实现对论文进行总结,帮助科研人进行论文初筛。
Chat-Reviewer
  • ChatReviewer is an AI assistant for automated paper review based on the ChatGPT-3.5 API interface. (After personal use, the comments seem to repeat the advantages mentioned in the paper and should be taken as reference only.)
  • ChatReviewer是一款基于ChatGPT-3.5的API接口的论文自动审稿AI助手(个人使用后看起来评论像是在重复论文里提到的优点,只能参考)
ChatPDF
  • Different from ChatPaper, ChatPDF allows users to ask questions to the AI assistant, resembling a conversation process.
  • 和ChatPaper不同,它是可以向AI助手提问的,类似一个聊天过程
OpenAI Translator
  • The tool is built on the ChatGPT API and can be applied on various platforms such as Chrome extension, Windows, macOS, Linux, etc. It supports three translation modes: translation, polishing, and summarization.
  • 该工具基于 ChatGPT API 构建,可应用于 Chrome 插件、Windows、macos、Linux 等平台,支持三种翻译模式:翻译、润色、总结。
NotionAI
  • Notion AI utilizes large-scale language models and data collection to help users automatically organize notes, correct spelling errors, list key points of articles, translate, create tables, and more. If you are a student, using Notion with Notion AI to organize and create study notes can improve accuracy and readability; If you are a working professional, Notion AI can quickly generate content such as discussion outlines, meeting agendas, and even reports.
  • Notion AI 利用大规模的语言模型和数据收集,能帮助用户自动整理笔记、改正错别字、列出文章重点、翻译、制作表格等。如果你是学生,使用 Notion 搭配 Notion AI 来整理、制作学习笔记可以提高准确性和易读性;如果你是上班族,通过 Notion AI 更可以快速地生成例如讨论大纲、会议行程,甚至是报告的内容。
Novel AI
  • A website that enables AI-powered story generation, text and image generation, and voice generation.
  • 一个可以用AI做故事生成,文图生成,语音生成的网站。
LLaMA
  • LLaMA is a large language model open-sourced by Meta, including its weights and code. Most of the open-source community's self-trained dialogue bots are based on LLaMA.
  • LLaMA是Meta开源的大语言模型。开源了权重和代码。现有的开源社区中自己训练的对话机器人基本都基于LLaMA。
Stanford Alpaca
  • Stanford Alpaca uses 52K instruction-following examples to fine-tune Meta's large language model LLaMA 7B, which has been made available by Meta in terms of model weights and inference code, to generate Alpaca 7B.
  • Stanford Alpaca使用 52K 个 intruction-following examples 来微调 Meta 的大语言模型 LLaMA 7B (Meta 开放了模型权重以及 Inference 代码),从而生成了 Alpaca 7B。
Alpaca-LoRA
  • Stanford Alpaca is fine-tuned on the entire LLaMA model, which means all parameters in the pre-trained model are fine-tuned. However, this method still requires high hardware costs and is inefficient to train. Alpaca-Lora, on the other hand, uses Lora technology to add additional network layers to the model and only trains the parameters of these new layers while keeping the original LLaMA model parameters frozen. Since the number of these new parameters is small, the cost of fine-tuning is significantly reduced (using an RTX 4090 graphics card, it only takes 5 hours to train a model comparable to Alpaca, bringing the requirements for computing power down to the consumer level), while still achieving similar results to full fine-tuning.
  • Stanford Alpaca 是在 LLaMA 整个模型上微调,即对预训练模型中的所有参数都进行微调(full fine-tuning)。但该方法对于硬件成本要求仍然偏高且训练低效。Alpaca-Lora 则是利用 Lora 技术,在冻结原模型 LLaMA 参数的情况下,通过往模型中加入额外的网络层,并只训练这些新增的网络层参数。由于这些新增参数数量较少,这样不仅微调的成本显著下降(使用一块 RTX 4090 显卡,只用 5 个小时就训练了一个与 Alpaca 水平相当的模型,将这类模型对算力的需求降到了消费级),还能获得和全模型微调(full fine-tuning)类似的效果。
ChatGLM
  • ChatGLM-6B is an open-source, Chinese-English bilingual conversation language model with 6.2 billion parameters based on the GLM architecture, provided by Tsinghua University. With model quantization technology, users can deploy it locally on consumer-grade graphics cards. ChatGLM-6B uses techniques similar to ChatGPT and is optimized for Chinese QA and conversations. It has been trained on approximately 1T Chinese-English bilingual data with the help of techniques such as supervised fine-tuning, feedback self-play, and human feedback reinforcement learning.
  • ChatGLM-6B 是清华大学提供的一个开源的、支持中英双语的对话语言模型,基于GLM架构,具有62亿参数。结合模型量化技术,用户可以在消费级的显卡上进行本地部署。ChatGLM-6B 使用了和 ChatGPT 相似的技术,针对中文问答和对话进行了优化。经过约 1T 标识符的中英双语训练,辅以监督微调、反馈自助、人类反馈强化学习等技术的加持。
AgentGPT
  • AgentGPT allows you to configure and deploy your own AI agent. Name your own custom AI and set it to achieve any imaginable goal. It will attempt to accomplish the goal by thinking about the task at hand, performing the task, and learning from the results.
  • AgentGPT 允许您配置和部署自AI 代理。为您自己的自定义 AI 命名,让它开始任何可以想象的目标。它将尝试通过思考要完成的任务、执行任务并从结果中学习来实现目标。
HuggingGPT
  • HuggingGPT allows you to connect to all AI models available on the Hugging Face platform and select the appropriate model for a given task based on the input provided. HuggingGPT is particularly useful for language-related tasks such as text generation, question answering, and language translation.
  • HuggingGPT,可以连接到HuggingFace上所有的AI模型,并根据输入的任务选择相应的模型执行。
Auto-GPT
  • This project can automatically help you accomplish all tasks based on the goals you set using GPT-4. You only need to provide an OpenAI API key with sufficient funds, and it will use various methods such as Google search, website browsing, and script execution to achieve your goals.
  • 该项目可以根据你设置的目标,使用GPT-4自动帮你完成所有的任务。你只要提供OpenAI的API Key,保证里面有钱,那么它就可以根据你设定的目标,采用Google搜索、浏览网站、执行脚本等方式帮你完成目标。
ShareGPT
  • A plugin that allows you to export your conversations with ChatGPT as data. Additionally, the website provides data of conversations between other users and ChatGPT.
  • 一个插件,能够将自己和ChatGPT对话的内容导出为数据。同时网站上还提供其他用户和ChatGPT对话内容的数据。
FastChat(Vicuna)
  • Like Stanford Alpaca, this chatbot is also based on the LLAMA large language model and fine-tuned using data collected and organized from the shareGPT website, resulting in a better chatbot.
  • 和Stanford Alpaca一样,基于LLAMA大语言模型进行微调,采用从shareGPT网站收集并整理的数据进行finetune,得到一个更好的chatbot。
Open Assistant
  • Open Assistant is a project aimed at making chat-based large language models accessible to everyone.
  • Open Assistant 是一个旨在让每个人都能访问基于聊天的大型语言模型的项目。
MiniGPT4
  • The model weights were trained using blip2 and Vicuna, and further fine-tuned to provide chat responses with image inputs for ChatGPT.
  • 采用blip2和Vicuna训练过的模型权重,进一步训练得到的模型,可以提供图片输入的chatGPT。
Vlog
  • Vlog can create a document containing both visual and audio information by processing a long video. By sending this document to ChatGPT, we can have a conversation while watching the video!
  • Vlog可以通过处理长视频,将其转换为包含视觉和音频信息的文档。将这个文档发送给 ChatGPT,我们可以在观看视频的同时进行聊天。
Excel&Formulabot
  • Transform your text instructions into Excel formulas in seconds with the help of AI for free. Become highly proficient in Excel in seconds.
  • 利用AI人工智能技术,将您的文字说明在几秒钟内转化为Excel公式

CV-Draw-3D

Name Website Overview
Stable Diffusion
  • The Runwayml project provides the sd1.5 version and the image editing inpainting version. The Stability project provides the sd2.0 version.
  • Runwayml项目提供sd1.5版本和图像编辑inpainting版本。Stability项目提供sd2.0版本。
Stable Diffusion webui
  • Create a UI interface for the text-to-image functionality of StableDiffusion, while also allowing for the integration of various plugins and user-defined models to achieve richer functionality.
  • 将stablediffusion文生图功能做成UI界面,同时可以接入各种插件和用户自定义模型,完成更丰富的功能
Midjourney
  • Midjourney is an AI program developed by Midjourney Research Lab that generates images based on text. Users can operate it through Discord bot commands to create a variety of image works. The program is currently on version 5.
  • Midjourney是一个由Midjourney研究实验室开发的人工智能程序,可根据文本生成图像,使用者可通过Discord的机器人指令进行操作,可以创作出很多的图像作品。现在已经出到了v5版本。
DALL·E 2
  • The text-to-image models provided by OpenAI.
  • OpenAI所提供的文生图模型。
Stable Diffusion Tutorials
  • This website provides a variety of tutorials, explanations, and introductions to stable diffusion, which are very rich and detailed in explaining and introducing AI text-to-image models.
  • 该网站提供了各种stable diffusion的教程,讲解和介绍AI文生图模型非常丰富和详细。
CivitAI
  • Using LoRA to train models with different styles on the open source stable diffusion. On this website, you can obtain style-transfer models trained by different users, which can be used with the web UI.
  • 利用LoRA在开源的stable diffusion上训练不同风格的模型。在该网站上可以获得不同用户自己训练出来的风格化的模型权重,可以配合webui使用。
Segment-Anything
  • Meta has launched a tool called "Segment Anything" that accurately identifies objects in images using AI. The project includes a model and a dataset, and the model is named the "Segment Anything Model" (SAM).
  • Meta 推出 Segment Anything 工具,可准确识别图像中的对象。该项目包括模型、数据集。该模型名为 Segment Anything Model(SAM)。
Ground-Segment-Anything
  • It integrates most of the popular AI models, including object detection, segmentation, image-text generation, image-text editing, speech recognition, and more. It is a very good AI integration tool.
  • 整合了现有的大部分流行的AI模型,包括目标检测,分割,文图生成,文图编辑,语音识别等等,是一个非常不错的AI整合工具。
Lama Cleaner
  • Free and open-source inpainting tool supported by SOTA AI models.
  • 由SOTA AI模型提供支持的免费开源inpainting工具。
segGPT
  • SegGPT can perform arbitrary segmentation tasks in images or videos, such as object instances, materials, parts, contours, and text, through context reasoning.
  • SegGPT可以通过上下文推理在图像或视频中执行任意分割任务,如对象实例、素材、部分、轮廓和文本。
Visual-ChatGPT
  • Visual-ChatGPT is an image chat application developed by Microsoft based on ChatGPT. It not only allows AI painting, but also enables AI to recognize image content and describe image content. Users can use text commands to retouch images, extract colors, edges, and more.
  • Visual-ChatGPT是微软基于ChatGPT开发的图像聊天应用,不光是AI绘画,还可以让AI识别图片内容和描述图片内容,通过文本命令去修图,提取颜色、边缘等。
luma AI
  • Using nerf technology for 3D modeling, it is possible to create a 3D model of an object simply by capturing a video of it.
  • 通过nerf技术进行3d建模,只需要拍摄一段视频就可完成物体的建模。
Stable-Dreamfusion
  • Based on the stable-diffusion technology, a model has been developed to convert images and text into 3D models.
  • 一个基于stable-diffusion完成image-to-3D和text-to-3D的模型技术。
roomGPT
  • This tool allows you to generate dreamy room effects using AI by uploading a photo of your room or a 3D rendering of it. It supports up to 3 image uploads per day, and the generated images can be downloaded and applied in interior design and decoration. Currently, the tool has generated 780,000 home decor images.
  • 你只需要给你的房间拍一张照,或是房间的 3D 效果图,并将其上传,即可用 AI 生成对应的梦幻房间效果图。该工具每天支持上传 3 张图片,生成的效果图支持下载,并应用于室内设计、室内装修等场景,目前已累积生成 78 万张家居效果图。

Audio-Speech

Name Website Overview
Whisper
  • A universal speech recognition model trained on a large dataset containing various types of audio. It is also a multitask model capable of performing multilingual speech recognition, speech translation, and language identification.
  • 一种通用的语音识别模型。它是在包含各种音频的大型数据集上训练的,也是一个可以执行多语言语音识别、语音翻译和语言识别的多任务模型。
SpeechGPT
  • A website where you can have voice conversations with ChatGPT.
  • 一个可以让你与 ChatGPT 进行语音聊天的网站

About