Chatgpt 简明教程

ChatGPT – Competitors

多年来，已经开发出 ChatGPT 的各种迭代，每一个都带来了改进和额外功能。主要版本包括 -

Over the years, various iterations of ChatGPT have been developed, each bringing improvements and additional features to the table. The key versions include −

  1. GPT-1 − Introduced in 2018, GPT-1 served as the inaugural model in the GPT series, focusing on text generation.

  2. GPT-2 − Unveiled in 2019, GPT-2 elevated the game with 1.5 billion parameters. It garnered attention for its highly persuasive text generation capabilities, albeit sparking concerns about disinformation.

  3. GPT-3 − Launched in 2020, GPT-3 stands as the most recent and advanced version in the GPT series, boasting 175 billion parameters. Applauded for its heightened ability to generate more natural text and perform diverse natural language processing tasks.

  4. GPT-4 − Unveiled in 2023, OpenAI claims that "GPT-4 can solve challenging problems with greater accuracy, thanks to its broader general knowledge and advanced reasoning capabilities."

每一次迭代都在提高自动化文本生成质量和精度的过程中发挥着至关重要的作用，促进了聊天机器人和用户之间更加自然和无缝的沟通。

Each iteration has played a crucial role in elevating the quality and precision of automated text generation, facilitating more natural and seamless communication between chatbots and users.

事实上，预计 ChatGPT 的能力将在未来几年内经历重大发展，特别是 OpenAI 正在积极开发下一代 GPT-5 语言模型。

In fact, ChatGPT’s capabilities are expected to undergo significant evolution in the coming years, especially with OpenAI actively working on the next-generation GPT-5 language model.

Competitors of ChatGPT

虽然 ChatGPT 占据着显赫的地位，但包括谷歌、Meta、Anthropic 和亚马逊在内的众多竞争对手正在使用大语言模型 (LLM)、深度学习和微调来在市场上确立自己的主导地位。

While ChatGPT holds a prominent position, various competitors, including Google, Meta, Anthropic, and Amazon, are using Large Language Models (LLMs), deep learning, and fine-tuning to establish their dominance in the market.

在人工智能大语言模型 (LLM) 领域，ChatGPT 有一些很有前途的竞争对手。在本章中，我们将单独探讨 ChatGPT 的每一种替代方案。

Within the AI Large Language Model (LLM) domain, there are several promising competitors to ChatGPT. In this chapter, we will explore each of these alternatives to ChatGPT individually.

Google Gemini (Formerly Google Bard)

谷歌 Gemini 是 ChatGPT 的一个强大的竞争对手。它于 2023 年 3 月首次亮相，是一款采用机器学习 (ML)、自然语言处理 (NLP) 和生成式人工智能来理解用户提示并提供文本响应的对话式人工智能聊天机器人。

Google Gemini is a formidable competitor to ChatGPT. It made its debut in March 2023 as a conversational AI chatbot employing machine learning (ML), natural language processing (NLP), and generative AI to comprehend user prompts and furnish text responses.

与 ChatGPT 不同，Gemini 拥有访问互联网的独特功能，可将从最近发布的内容中获取的信息整合到其响应中。

In contrast to ChatGPT, Gemini has the unique capability to access the Internet, integrating information scraped from recently published content into its responses.

最初在 Google LaMDA（一种大语言模型 (LLM)）上进行训练，Gemini 在 2023 年 5 月经历了一次改造性再训练，过渡到更先进的 Pathways Language Model 2（PaLM 2）。谷歌声称，与 LaMDA 相比，PaLM 2 处理信息的速度提高了 500 倍，准确率提高了十倍。

Initially trained on Google LaMDA, a large language model (LLM), Gemini underwent a transformative re-training in May 2023, transitioning to the more advanced Pathways Language Model 2 (PaLM 2). Google asserts that PaLM 2 processes information up to 500 times faster than LaMDA and achieves a remarkable tenfold increase in accuracy.

谷歌于 2024 年 2 月 8 日将 Bard 聊天机器人重新命名为 Gemini 。

Google rebranded the Bard chatbot as Gemini on February 8, 2024.

Midjourney

Midjourney 是一款创新的 AI 工具，专门用于快速将提示转化为图像。Midjourney 每月都会更新模型，不断突破创作型人工智能的界限。

Midjourney is an innovative AI tool, specializing in swiftly transforming prompts into images. With its monthly model updates, Midjourney continues to push the boundaries of creative AI.

由于 Midjourney 采用了自筹资金和封闭源代码的方式运作，它的内部运作方式仍然没有公开。该平台采用机器学习技术，融合了大语言模型和扩散模型。

As Midjourney operated on a self-funded and closed-source basis, its intricate workings remain undisclosed. The platform employs machine learning technologies, incorporating large language and diffusion models.

与 ChatGPT 和 Bing Chat 等类似平台不同，Midjourney 采用了一种独特的订阅式模式，要求图形处理单元 (GPU) 以获得最佳性能。

In contrast to counterparts like ChatGPT and Bing Chat, Midjourney adopts a unique approach with a subscription-based model, requiring graphics processing units (GPUs) for optimal performance.

虽然没有免费试用，但定价为 10 美元的入门计划允许根据命令生成 200 多张图片。

While lacking a free trial, the basic plan, priced at $10, facilitates the generation of over 200 images upon command.

Claude 2

2023 年 7 月，人工智能公司 Anthropic 发布了其最新的聊天机器人 Claude 2，该机器人由大语言模型提供支持。

In July 2023, Anthropic, an AI company, unveiled its latest chatbot named Claude 2 which is powered by a large language model.

Claude 2 代表了 Anthropic 的前代人工智能迭代 Claude 1.3 的重大升级。值得注意的改进包括基于书面指令增强的代码编写能力和扩展的“上下文窗口”。用户现在可以输入整本书，并根据内容向 Claude 2 提出问题。

Claude 2 represents a notable upgrade from Anthropic’s previous AI iteration, Claude 1.3. Noteworthy improvements include enhanced code-writing capabilities based on written instructions and an expanded "context window." Users can now input entire books and pose questions to Claude 2 based on their content.

这些增强使 Claude 2 与领先的模型（如推动 OpenAI 的 ChatGPT 的 GPT-3.5 和 GPT-4）处于同等地位。

These enhancements position Claude 2 on par with leading models like GPT-3.5 and GPT-4, which drive OpenAI’s ChatGPT.

要了解 claude 2，请在 [role="bare"] [role="bare"]https://claude.ai/ 注册。

To meet claude 2, sign up at [role="bare"]https://claude.ai/.

Runway ML

Runway ML 代表了一个开创性的平台，旨在让艺术家、设计师和创作者使用机器学习的潜力。Runway ML 使用户能够使用文本提示制作视频、使用文本或图像更改视频样式以及制作个性化肖像、动物、风格等等。

Runway ML represents a groundbreaking platform designed for artists, designers, and creators to use the potential of machine learning. Runway ML empowers users to craft videos using text prompts, alter video styles using text or images, and craft personalized portraits, animals, styles, and beyond.

Runway ML 简单易用，消除了对深入编程知识的需要。Runway ML 的一个突出特点在于其人工智能魔法工具，它促进了实时视频编辑、协作以及其他多种功能。

Simple and accessible, Runway ML eliminates the necessity for in-depth programming knowledge. A standout feature of Runway ML lies in its AI Magic Tools, which facilitate real-time video editing, collaboration, and a myriad of other functionalities.

Runway ML 的人工智能魔法工具的创造潜力无限，提供各种可能性。

The creative potential with Runway ML’s AI Magic Tools is boundless, offering a diverse range of possibilities.

GitHub Copilot

GitHub Copilot 于 2021 年推出，是 GitHub 与 OpenAI 合作开发的一款革命性编码助手。

GitHub Copilot, introduced in 2021, is a revolutionary coding assistant developed by GitHub in collaboration with OpenAI.

Copilot 无缝集成到流行的代码编辑器中，为开发人员提供实时代码建议和自动补全。它由 OpenAI 的 Codex 提供支持，从大量公共代码存储库中获取见解，提供与上下文相关的代码片段，以提高编码效率。

Seamlessly integrated into popular code editors, Copilot offers developers real-time code suggestions and autocompletions. Powered by OpenAI’s Codex, it draws insights from a vast array of public code repositories, providing context-aware code snippets to enhance coding efficiency.

Copilot 通过帮助开发人员进行智能代码生成、提高生产力和使编码更容易来改变开发格局。与专注于自然语言对话的 ChatGPT 不同，GitHub Copilot 专为加快代码制作而量身定制。

Copilot transforms the development landscape by assisting developers with intelligent code generation, fostering productivity, and making coding more accessible. Unlike ChatGPT, which focuses on natural language conversations, GitHub Copilot is uniquely tailored to expedite code production.

Perplexity AI

Perplexity AI 于 2022 年 8 月发布，是一款具有搜索引擎功能的 AI 聊天机器人。它建立在 GPT-3 和 GPT-4 之上，采用自然语言处理 (NLP) 和机器学习等复杂技术。这使该平台能够对用户查询提供准确而全面的响应。

Perplexity AI, released in August 2022, functions as an AI chatbot with search engine capabilities. It is built upon GPT-3 and GPT-4 that employ sophisticated technologies like natural language processing (NLP) and machine learning. This enables the platform to deliver precise and thorough responses to user queries.

Perplexity AI 专为实时网络搜索而设计，确保用户可以获取各种主题的最新信息。该平台由强大的语言模型（特别是 OpenAI 的 GPT 技术）提供支持，在理解和生成类人文本方面表现出色。Perplexity AI 被定位为一个答案引擎，致力于增强个人探索和交换信息的方式。

Tailored for real-time web searches, Perplexity AI ensures access to the latest information across diverse topics. Fueled by robust language models, particularly OpenAI’s GPT technology, the platform excels in comprehending and generating human-like text. Positioned as an answer engine, Perplexity AI strives to enhance the way individuals explore and exchange information.

Perplexity AI 便于各种用户群体使用，提供网络版和 iPhone 应用程序版本。用户可以通过访问他们的网站免费使用 Perplexity AI。

Perplexity AI is conveniently accessible to a diverse user base, with both web and iPhone app versions available. Users can freely utilize Perplexity AI by visiting their website.

按照以下步骤与 Perplexity AI 互动 −

Follow these steps to engage with Perplexity AI −

  1. Navigate to www.perplexity.ai.

  2. Pose your question by entering it into the search bar and clicking the blue arrow.

  3. Evaluate Perplexity AI’s response along with the provided sources.

  4. Continue the interaction by asking follow-up questions using the "Ask a follow-up" bar below.

Meta Llama 2

Llama 2 是 Meta 的大型语言模型 (LLM)，该模型可在优化计算能力和资源的同时生成文本和代码。

Llama 2 is Meta’s large language model (LLM) that can generate text and code while optimizing computing power and resources.

Llama 2 在大规模多任务语言理解 (MMLU) 中获得了 68.9 的评分，仅略低于 GPT 3.5 的 70.0。虽然它不及 GPT-4 86.4 的评分，但这种接近性确立了 Llama 2 作为与 GPT 3.5 可信的开源竞争对手。

Llama 2 achieves a Massive Multitask Language Understanding (MMLU) score of 68.9, just slightly trailing GPT 3.5’s 70.0. While it falls short of GPT-4’s 86.4 rating, this proximity establishes Llama 2 as a credible open-source competitor to GPT 3.5.

值得强调的是，Llama 2 的训练数据一直持续到 2022 年 9 月，其他微调数据近期至 2023 年 7 月。相比之下，GPT 3.5 的训练数据只覆盖到 2021 年 9 月。该差异使 Llama 2 定位为比其 OpenAI 对应产品提供更多当前信息的信息源。

It’s important to highlight that Llama 2’s training data extends up to September 2022, with additional tuning data as recent as July 2023. In contrast, GPT 3.5’s training data only covers up to September 2021. This distinction positions Llama 2 as a source of more current information compared to its OpenAI counterpart.

Amazon CodeWhisperer

亚马逊的 AI 代码生成器 CodeWhisperer 由大量代码库进行训练，提供基于注释和现有代码的实时代码建议，从代码片段到完成的功能。

Amazon CodeWhisperer, an AI Code Generator from AWS, is trained on vast code repositories, offering real-time code suggestions ranging from snippets to complete functions based on comments and existing code.

它简化了编码任务，支持 15 种编程语言，包括 Python、Java 和 JavaScript，并与流行的 IDE（如 VS Code、IntelliJ IDEA、AWS Cloud9、AWS Lambda 控制台、JupyterLab 和亚马逊 SageMaker Studio）集成。

Streamlining coding tasks, it supports 15 programming languages, including Python, Java, and JavaScript, and integrates with popular IDEs like VS Code, IntelliJ IDEA, AWS Cloud9, AWS Lambda console, JupyterLab, and Amazon SageMaker Studio.

免费个人版每月包含无限代码建议、引用跟踪和 50 次安全扫描。CodeWhisperer 内置了可以检测漏洞并提供立即修复建议的安全扫描。

The free Individual Tier includes unlimited code suggestions, reference tracking, and 50 security scans per user monthly. CodeWhisperer has built in security scans that can detect vulnerabilities and provide immediate remediation suggestions.

CodeWhisperer 还集成了一个引用跟踪器，标记出建议中的开源相似之处。用存储库 URL、文件引用和许可证详细信息对这些建议进行注释，使用户可以在实施代码之前对其进行审查。

CodeWhisperer also incorporates a reference tracker flagging open-source similarities in suggestions. Annotating such suggestions with repository URLs, file references, and license details allows users to review the code before its implementation.

D-ID Studio

D-ID 的 Creative Reality Studio （又称为 Studio D-ID ）是一个自助服务平台，高效地使用生成式 AI 工具。它使用户能够制作具有动态对话化身的视频。

D-ID’s Creative Reality Studio, also known as Studio D-ID, stands as a self service platform that uses generative AI tools very efficiently. It empowers users to craft videos featuring dynamic, conversing avatars.

此平台将 D-ID 的深度学习面部动画技术与 GPT 文本生成和 Stable Diffusion 文本到图像的能力无缝集成。D ID 是第一个针对使用 AI 生成创新视频的人员提供的一体化解决方案。

This platform seamlessly integrates D-ID’s deep-learning face animation technology with GPT text generation and Stable Diffusion text-to-image capabilities. D ID is the first all-in-one solution for those aiming to produce innovative videos using AI.

作为一种基于网络的工具，Creative Reality Studio 采用了最先进的面部动画和文本转语音技术，以提供类似于生活的对话式 AI 体验。

Operating as a web-based tool, Creative Reality Studio employs state-of-the-art face animation and text-to-speech technologies to deliver life like conversational AI experiences.

这种多功能技术可用于制作个人的数字版本、历史人物、虚构角色、主持人或品牌大使。事实上，工作室 D-ID 为内容注入生命，为平凡的文件和 PowerPoints 提供了一个动态替代方案。

This versatile technology finds applications in crafting digital renditions of individuals, historical figures, fictional characters, presenters, or brand ambassadors. In fact, studio D-ID breathes life into content, providing a dynamic alternative to mundane documents and PowerPoints.