


On May 18, 2019, the 7th iTechClub North China Internet Technology Elite Summit Forum was held. The Director of Baidu Engineering Performance Department gave a keynote speech on "Toward a New Paradigm of AI Native R&D for Human-Machine Collaboration". He released the latest achievement of Baidu's intelligent code assistant Comate - Comate Code Knowledge Enhancement 2.0. This is the first intelligent code assistant in China that supports real-time retrieval. It has built-in more than 10,000 Github popular code libraries, which has brought great benefits to developers around the world. An unprecedented programming experience.
As one of the highlights of this conference, Comate Code Knowledge Enhancement 2.0 received great attention from the participants. The intelligent code assistant Comate is an intelligent code completion and recommendation tool based on the Baidu Wenxin model. Through deep learning and natural language processing technology, it can analyze developers' programming intentions in real time and automatically recommend appropriate code snippets and library functions, greatly improving programming efficiency and code quality.
Tozhi introduced in detail the three major advantages of Comate Code Knowledge Enhancement 2.0 in his speech. First of all, it has built-in 10,000+ Github popular code libraries, which can support comprehensive search and Q&A. Excellent frameworks covering various languages and technology stacks, such as Spring, Mybatis, FastAPI, React, etc. for engineering, Transformer, PaddlePaddle, etc. for algorithms, as well as the latest AI frameworks such as AutoGPT, Langchain, etc.
Developers often encounter code base-related questions during interviews. Now through Comate's interpretation, you can quickly obtain explanations of specific businesses in the open source framework, helping developers master source code logic and improve programming skills. This is like creating a "code library", assisted by Comate, to help every developer learn excellent coding practices.
For example, in an interview scenario, ask "What is the default scope of Spring Bean? How to change the scope of Bean?"
Comate is based on Web online Search to directly obtain the latest technical knowledge in real time. This is also the first intelligent code assistant in China that supports real-time retrieval. Comate quickly learns new knowledge based on web search, analyzes complex problems through large models, clarifies demand solutions, and quickly implements and modifies code based on the user's existing code. If you directly throw out a web page address, Comate can also understand the content of the web page and give answers based on the request. At the same time, knowledge is no longer an isolated island. By mixing and arranging real-time network search content, specified web page content, locally uploaded files, local code libraries and other knowledge, multi-ability blessings can help you generate code that is more relevant to the actual business. .
If you need to generate an Agent to call Wenxinyishuo 4.0 API implementation, you only need to state your needs, and Comate can search through the web page to generate the framework code, find the latest Wenxinyishuo API, and generate business logic code. What used to take days of research and development work can be completed in just a few sentences using Comate.
#Finally, given any API link, the calling code and corresponding test cases can be generated, which is efficient and high-quality. Currently, Comate supports in-depth understanding of local code bases and private domain knowledge within the organization, such as business interface documents, product requirements documents, test case documents, service deployment documents, etc. By fully grasping the context of the current "programming site" and having an in-depth understanding of the entire R&D link of "business/project/service", Comate can generate usage and test code that is more relevant to the business and more targeted.
Entering the Code Knowledge Enhancement 2.0 stage, Comate can provide convenient support for different scenarios such as code writing, learning, interviews and testing. For example, in a code writing scenario, Comate can retrieve the latest technology implementation, generate framework code, and assist in modifying the code based on the user's existing code; in a testing scenario, Comate can retrieve the latest technology implementation based on the scenario-based test description input by the user. A series of code snippets that meet the intent of the business requirements, and complete automated test code is generated through the large model.
All functions released by Comate can be downloaded from the official website to use the IDE plug-in, or you can experience it online through the web page on the Comate official website. Comate has been committed to improving developers' programming efficiency and code quality. Public data shows that 27% of Baidu's daily new codes are automatically generated by Comate, and the overall adoption rate reaches 46%. The release of Comate Code Knowledge Enhancement 2.0 not only further improves the functions and performance of Comate, but also demonstrates the tremendous changes that large models have brought to the field of AI programming.
Recently, in the smart assistant rating lists of major plug-in markets such as VSCode and Jetbrains, Baidu Comate ranked first on the list with 4.5 and 4.4 respectively. Now Comate is like an intelligent code assistant, helping programmers to develop higher-quality code with higher efficiency. The era of "programming" with just a few clicks has arrived.
The above is the detailed content of With built-in 10,000+ popular Github code libraries, Baidu officially released Comate Code Knowledge Enhancement 2.0. For more information, please follow other related articles on the PHP Chinese website!

PowerInfer提高了在消费级硬件上运行AI的效率上海交大团队最新推出了超强CPU/GPULLM高速推理引擎PowerInfer。PowerInfer和llama.cpp都在相同的硬件上运行,并充分利用了RTX4090上的VRAM。这个推理引擎速度有多快?在单个NVIDIARTX4090GPU上运行LLM,PowerInfer的平均token生成速率为13.20tokens/s,峰值为29.08tokens/s,仅比顶级服务器A100GPU低18%,可适用于各种LLM。PowerInfer与

要让大型语言模型(LLM)充分发挥其能力,有效的prompt设计方案是必不可少的,为此甚至出现了promptengineering(提示工程)这一新兴领域。在各种prompt设计方案中,思维链(CoT)凭借其强大的推理能力吸引了许多研究者和用户的眼球,基于其改进的CoT-SC以及更进一步的思维树(ToT)也收获了大量关注。近日,苏黎世联邦理工学院、Cledar和华沙理工大学的一个研究团队提出了更进一步的想法:思维图(GoT)。让思维从链到树到图,为LLM构建推理过程的能力不断得到提升,研究者也通

近期,复旦大学自然语言处理团队(FudanNLP)推出LLM-basedAgents综述论文,全文长达86页,共有600余篇参考文献!作者们从AIAgent的历史出发,全面梳理了基于大型语言模型的智能代理现状,包括:LLM-basedAgent的背景、构成、应用场景、以及备受关注的代理社会。同时,作者们探讨了Agent相关的前瞻开放问题,对于相关领域的未来发展趋势具有重要价值。论文链接:https://arxiv.org/pdf/2309.07864.pdfLLM-basedAgent论文列表:

FATE2.0全面升级,推动隐私计算联邦学习规模化应用FATE开源平台宣布发布FATE2.0版本,作为全球领先的联邦学习工业级开源框架。此次更新实现了联邦异构系统之间的互联互通,持续增强了隐私计算平台的互联互通能力。这一进展进一步推动了联邦学习与隐私计算规模化应用的发展。FATE2.0以全面互通为设计理念,采用开源方式对应用层、调度、通信、异构计算(算法)四个层面进行改造,实现了系统与系统、系统与算法、算法与算法之间异构互通的能力。FATE2.0的设计兼容了北京金融科技产业联盟的《金融业隐私计算

大型语言模型(LLM)被广泛应用于需要多个链式生成调用、高级提示技术、控制流以及与外部环境交互的复杂任务。尽管如此,目前用于编程和执行这些应用程序的高效系统却存在明显的不足之处。研究人员最近提出了一种新的结构化生成语言(StructuredGenerationLanguage),称为SGLang,旨在改进与LLM的交互性。通过整合后端运行时系统和前端语言的设计,SGLang使得LLM的性能更高、更易控制。这项研究也获得了机器学习领域的知名学者、CMU助理教授陈天奇的转发。总的来说,SGLang的

将不同的基模型象征为不同品种的狗,其中相同的「狗形指纹」表明它们源自同一个基模型。大模型的预训练需要耗费大量的计算资源和数据,因此预训练模型的参数成为各大机构重点保护的核心竞争力和资产。然而,与传统软件知识产权保护不同,对预训练模型参数盗用的判断存在以下两个新问题:1)预训练模型的参数,尤其是千亿级别模型的参数,通常不会开源。预训练模型的输出和参数会受到后续处理步骤(如SFT、RLHF、continuepretraining等)的影响,这使得判断一个模型是否基于另一个现有模型微调得来变得困难。无

IBM再度发力。随着AI系统的飞速发展,其能源需求也在不断增加。训练新系统需要大量的数据集和处理器时间,因此能耗极高。在某些情况下,执行一些训练好的系统,智能手机就能轻松胜任。但是,执行的次数太多,能耗也会增加。幸运的是,有很多方法可以降低后者的能耗。IBM和英特尔已经试验过模仿实际神经元行为设计的处理器。IBM还测试了在相变存储器中执行神经网络计算,以避免重复访问RAM。现在,IBM又推出了另一种方法。该公司的新型NorthPole处理器综合了上述方法的一些理念,并将其与一种非常精简的计算运行

去噪扩散模型(DDM)是目前广泛应用于图像生成的一种方法。最近,XinleiChen、ZhuangLiu、谢赛宁和何恺明四人团队对DDM进行了解构研究。通过逐步剥离其组件,他们发现DDM的生成能力逐渐下降,但表征学习能力仍然保持一定水平。这说明DDM中的某些组件对于表征学习的作用可能并不重要。针对当前计算机视觉等领域的生成模型,去噪被认为是一种核心方法。这类方法通常被称为去噪扩散模型(DDM),通过学习一个去噪自动编码器(DAE),能够通过扩散过程有效地消除多个层级的噪声。这些方法实现了出色的图


Hot AI Tools

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Undress AI Tool
Undress images for free

Clothoff.io
AI clothes remover

AI Hentai Generator
Generate AI Hentai for free.

Hot Article

Hot Tools

DVWA
Damn Vulnerable Web App (DVWA) is a PHP/MySQL web application that is very vulnerable. Its main goals are to be an aid for security professionals to test their skills and tools in a legal environment, to help web developers better understand the process of securing web applications, and to help teachers/students teach/learn in a classroom environment Web application security. The goal of DVWA is to practice some of the most common web vulnerabilities through a simple and straightforward interface, with varying degrees of difficulty. Please note that this software

Atom editor mac version download
The most popular open source editor

Dreamweaver Mac version
Visual web development tools

PhpStorm Mac version
The latest (2018.2.1) professional PHP integrated development tool

SecLists
SecLists is the ultimate security tester's companion. It is a collection of various types of lists that are frequently used during security assessments, all in one place. SecLists helps make security testing more efficient and productive by conveniently providing all the lists a security tester might need. List types include usernames, passwords, URLs, fuzzing payloads, sensitive data patterns, web shells, and more. The tester can simply pull this repository onto a new test machine and he will have access to every type of list he needs.
