current location:Home>Technical Articles>Technology peripherals>AI
- Direction:
- All web3.0 Backend Development Web Front-end Database Operation and Maintenance Development Tools PHP Framework Daily Programming WeChat Applet Common Problem Other Tech CMS Tutorial Java System Tutorial Computer Tutorials Hardware Tutorial Mobile Tutorial Software Tutorial Mobile Game Tutorial
- Classify:
-
- ICLR 2024 | Model critical layers for federated learning backdoor attacks
- Federated learning uses multiple parties to train models while data privacy is protected. However, because the server cannot monitor the training process performed locally by participants, participants can tamper with the local training model, thus posing security risks to the overall federated learning model, such as backdoor attacks. This article focuses on how to launch a backdoor attack on federated learning under a defensively protected training framework. This paper finds that the implantation of backdoor attacks is more closely related to some neural network layers, and calls these layers the key layers for backdoor attacks. In federated learning, clients participating in training are distributed on different devices. They each train their own models, and then upload the updated model parameters to the server for aggregation. Since the client participating in the training is not trustworthy, there are certain risks, so
- AI 977 2024-04-07 09:04:08
-
- US$100,000 to train Llama-2 large model! All Chinese people build a new MoE, Jia Yangqing, former CEO of SD, looks on
- To learn more about AIGC, please visit: 51CTOAI.x community https://www.51cto.com/aigc/ "Only" US$100,000 is needed to train a large model at the Llama-2 level. A MoE model with smaller size but no loss in performance is here: it is called JetMoE and comes from research institutions such as MIT and Princeton. The performance is significantly higher than that of Llama-2 of the same size. △Reposted by Jia Yangqing You must know that the latter has an investment cost of billions of dollars. JetMoE is completely open source when released and is friendly to academia: it only uses public data sets and open source code, and can be fine-tuned with consumer-grade GPUs. Needless to say, the cost of building large models is really much cheaper than people think. Ps.S
- AI 481 2024-04-07 09:04:01
-
- Up owners have already started to play tricks. Tencent opens up 'AniPortrait' to let photos sing and speak.
- AniPortrait models are open source and can be played with freely. "A new productivity tool for Xiaopozhan Ghost Zone." Recently, a new project released by Tencent Open Source received such evaluation on Twitter. This project is AniPortrait, which generates high-quality animated portraits based on audio and a reference image. Without further ado, let’s take a look at the demo that may be warned by a lawyer’s letter: Anime images can also speak easily: The project has already received widespread praise after just a few days since it was launched: the number of GitHub Stars has exceeded 2,800. Let’s take a look at the innovations of AniPortrait. Paper title: AniPortrait:Audio-DrivenSynthesisof
- AI 478 2024-04-07 09:01:16
-
- A Diffusion Model Tutorial Worth Your Time, from Purdue University
- Diffusion can not only imitate better, but also "create". The diffusion model (DiffusionModel) is an image generation model. Compared with the well-known algorithms such as GAN and VAE in the field of AI, the diffusion model takes a different approach. Its main idea is a process of first adding noise to the image and then gradually denoising it. How to denoise and restore the original image is the core part of the algorithm. The final algorithm is able to generate an image from a random noisy image. In recent years, the phenomenal growth of generative AI has enabled many exciting applications in text-to-image generation, video generation, and more. The basic principle behind these generative tools is the concept of diffusion, a special sampling mechanism that overcomes the limitations of previous methods.
- AI 365 2024-04-07 09:01:07
-
- The fake one looks like the real thing, Tiangong Music's big model brings a subversive AI experience
- Yesterday, Kunlun Wanwei's large-scale AI music generation model "Tiangong SkyMusic" launched a free invitation test event. The media, industry experts and interested music practitioners are sincerely invited to experience SOTA's music model products. This product enables users to have an immersive experience while experiencing the emotional expression of human voices together. After the invitation test started, the majority of users have high expectations for the "Tiangong SkyMusic" AI music generation large model. The staff received hundreds of thousands of test applications in a very short period of time, including many professional music creators, media and industry experts. At the same time, a large number of test applications are continuously sent to the backend. Among the applications, there are many professional music creators, media and industry experts, and there are also a large number of test applications that need to be sourced.
- AI 2317 2024-04-03 21:25:10
-
- What is generative AI? What feature types are there
- Generative AI is a type of human artificial intelligence technology that can generate various types of content, including text, images, audio and synthetic data. So what is artificial intelligence? What is the difference between artificial intelligence and machine learning? What are the technical features? Artificial intelligence is the discipline, a branch of computer science, that studies the creation of intelligent agents. These are systems in which intelligent agents can reason, learn, and act autonomously. The study of intelligent agents is the study of systems that can reason, learn, and act autonomously. Artificial intelligence is concerned with the theories and methods of building machines that think and act like humans. Within this discipline, machine learning is a field of artificial intelligence. It is a program or system that trains a model based on input data. The trained model can be trained from new or unseen data.
- AI 1758 2024-04-03 17:58:21
-
- Will AI threaten 8 million British jobs? IPPR: Can't just benefit tech giants
- The British Institute for Public Policy Research (IPPR) recently pointed out in a report that 8 million jobs in the UK will be threatened due to the popularity of AI. However, reports show that, albeit at its worst, AI can stimulate economic growth and make existing jobs more powerful through government intervention and concentrated regulation. This concept will be very critical in the next few years. If the government can formulate reasonable policies, AI will bring about GPD growth. If it is not reasonable, a large number of jobs will disappear. The AI impact is divided into two stages. IPPR believes that generating an AI will be divided into two stages when it becomes popular. The first phase is what is happening now, marked by the rise of generative AI platforms, such as GPT-4. In the second stage, companies will integrate generative AI technology
- AI 1960 2024-04-03 17:49:16
-
- Detailed explanation of GQA, the attention mechanism commonly used in large models, and Pytorch code implementation
- GroupedQueryAttention is a multi-query attention method in large language models. Its goal is to achieve the quality of MHA while maintaining the speed of MQA. GroupedQueryAttention groups queries, and queries within each group share the same attention weight, which helps reduce computational complexity and increase inference speed. In this article, we will explain the idea of GQA and how to translate it into code. GQA is in the paper GQA:TrainingGeneralizedMulti-QueryTransformerModelsfromMulti-HeadCheckpoint
- AI 746 2024-04-03 17:40:09
-
- The first domestic AI programmer joins Alibaba Cloud: exclusive job number AI001, KPI is that one person can write 20% of the company's code
- You must have heard a saying: You should start learning computers from childhood. In the past many years, learning to program has been a fashionable thing, but with the development of generative AI technology, the attitude of the technology circle seems to have changed. Nvidia CEO Jensen Huang said: "In the future, just leave programming to AI. From now on, everyone will be a software engineer." Soon after, Devin, the world's first AI software engineer, shocked developers and really aroused people's interest in There is a heated discussion about the future prospects of the career of programmers. Is it possible that the jobs of programmers are really going to be taken away by AI? In fact, this is not the case. AI is still just a work assistant for programmers. In fact, there is an AI programmer in China who has been working at a major Internet company for some time. It is the tens of thousands of workers at Alibaba Cloud
- AI 976 2024-04-03 16:04:04
-
- Open source large-model AI agent operating system: control AI agents like Windos
- This article is reprinted with the authorization of the AIGC open community. Please contact the source for reprinting. To learn more about AIGC, please visit: 51CTOAI.x community https://www.51cto.com/aigc/ Last year, the emergence of AutoGPT made us realize the powerful automation capabilities of AI agents and created a new AI Agency track. However, there are still many problems that need to be solved in sub-task scheduling, resource allocation and collaboration between AI. So researchers at Rutgers University developed AIOS, an AI agent operating system with large models at its core. It can effectively solve the problem of low resource call rate as the number of AI agents increases. It can also promote context switching between agents and realize concurrent agent merging.
- AI 1011 2024-04-03 15:10:01
-
- The open source version of AI programmers is here: GPT-4 blessing, ability comparable to Devin, 1.4k Stars a day
- To learn more about AIGC, please visit: 51CTOAI.x community https://www.51cto.com/aigc/ Recently, many people are worried about AI replacing their jobs. Devin, the "first AI programmer" who became popular in the AI circle last month, has mastered full-stack skills by using large model capabilities. He can automatically complete complex code tasks by only requiring humans to give natural language instructions. The tool capabilities demonstrated by Devin are very amazing, especially for this startup that takes the closed source route. Currently, only a few people can use this closed beta quota. On Tuesday, researchers from Princeton University’s NLP group released SWE-agent, an open source version of the AI program.
- AI 1011 2024-04-03 15:01:20
-
- Katy Perry and many other musicians jointly issued an open letter to resist AI replacing human artists
- According to news on April 3, well-known musicians such as Katy Perry and Billie Eilish joined many colleagues in writing a letter to developers, calling on them not to use artificial intelligence (AI) to replace human artists. The emergence of this website, with the gradual penetration of AI technology, through the music industry, may endanger the livelihood of musicians. The stars chose to express their demands through an open letter, emphasizing that they do not want this technology to replace the raw talent and skills that human artists possess. According to Axios, the joint letter is also signed by well-known musicians from different music genres such as Elvis Costello, Norah Jones, and Nicki Minaj. The letter reads: “We call on all AI developers, technology companies,
- AI 645 2024-04-03 14:40:11
-
- Open source! Beyond ZoeDepth! DepthFM: Fast and accurate monocular depth estimation!
- 0.What does this article do? We propose DepthFM: a versatile and fast state-of-the-art generative monocular depth estimation model. In addition to traditional depth estimation tasks, DepthFM also demonstrates state-of-the-art capabilities in downstream tasks such as depth inpainting. DepthFM is efficient and can synthesize depth maps within a few inference steps. Let’s read about this work together ~ 1. Paper information title: DepthFM: FastMonocularDepthEstimationwithFlowMatching Author: MingGui, JohannesS.Fischer, UlrichPrestel, PingchuanMa, Dmytr
- AI 966 2024-04-03 12:04:01
-
- Is artificial intelligence overrated?
- There is no doubt that artificial intelligence has become one of the hottest topics in recent years, capturing the imagination of technology experts, entrepreneurs and the public. However, amid the hype and excitement surrounding artificial intelligence, there is a growing debate about whether artificial intelligence is overvalued. Some commentators believe that AI is simply an advanced form of curve fitting rather than the revolutionary technology it is portrayed to be. Nonetheless, the development and application of artificial intelligence is still very impressive. From self-driving vehicles to smart homes, artificial intelligence has penetrated into every aspect of our lives. Its potential is limitless. However, we cannot ignore the challenges and problems faced by artificial intelligence. A key question is the growing debate over whether artificial intelligence is overrated. The core of artificial intelligence is to develop algorithms and systems
- AI 495 2024-04-03 12:01:01
-
- How to closely integrate online mapping and trajectory prediction?
- Original title: Producing and Leveraging Online Map Uncertainty in Trajectory Prediction Paper link: https://arxiv.org/pdf/2403.16439.pdf Code link: https://github.com/alfredgu001324/MapUncertaintyPrediction Author affiliation: University of Toronto Vector Institute NVIDIA Research Stanford University Paper idea: High Precision (HD) Maps play a role in the development of the modern autonomous vehicle (AV) technology stack
- AI 1103 2024-04-03 12:00:18