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:
-
- ICML 2024 AI for Math Workshop call for papers and challenge launched!
- ICML2024, AIforMathWorkshop Workshop on Formal and Natural Language AI Mathematical Reasoning Time: July 26/27, 2024 Location: Vienna, Austria. Held simultaneously on site and online. Workshop homepage: https://sites.google.com/view/ai4mathworkshopicml2024/ Mathematical reasoning is the most challenging and deep part of human intelligence. In the development process of mathematical reasoning, humans have summarized various formal languages, which can strictly describe mathematical problems and proof processes. In recent years, machine learning algorithms and large-scale language models are gradually approaching or even surpassing human performance in some mathematical reasoning.
- AI 650 2024-07-18 05:36:50
-
- Meta develops System 2 distillation technology, and the Llama 2 dialogue model task accuracy is close to 100%
- The researchers said that if System2 distillation can become an important feature of future continuous learning AI systems, it can further improve the performance of inference tasks where System2 does not perform so well. When it comes to large language model (LLM) strategies, there are generally two types, one is immediate System1 (fast response), and the other is System2 (slow thinking). Where System2 reasoning favors thoughtful thinking, generative intermediate thinking allows models (or humans) to reason and plan in order to successfully complete a task or respond to instructions. In System2 reasoning, effortful mental activity is required, especially in situations where System1 (more automatic thinking) can go wrong. Therefore, System1 is
- AI 921 2024-07-18 05:07:20
-
- To directly address the real AGI needs of Party A, the Artificial Intelligence Empowerment Industry Integration Development Forum was successfully held
- On July 6, the "2024 WAIC Artificial Intelligence Empowerment Industry Integration Development Forum" was grandly held at the World Expo Exhibition and Convention Center. The main topic of this forum is to discuss issues related to artificial intelligence empowering new industrialization and promoting the development of industrial integration, including leadership speeches, signing ceremonies, keynote speeches, release of artificial intelligence scenario requirements for central and state-owned enterprises, and roundtable forums. Many enterprises from central state-owned enterprises and artificial intelligence fields participated, including China Electronic Information Industry Development Research Institute, China Mobile Research Institute, Sinopec Shengli Oilfield, State Grid Customer Service Center, China Electronics Yuchuang, China Southern Power Grid Digital Grid Group, Damo Institute, Baidu Smart Cloud, Innovation Qizhi, etc. Guests attending the conference focused on the application practice of artificial intelligence in different fields, the development and application of large models, and intelligent operation and maintenance.
- AI 492 2024-07-18 03:14:57
-
- How can fashionable AIGC marketers achieve a win-win situation between 'lizi' and 'face'?
- Innovation and security of AIGC technology in the marketing field In the past year, AI technology has set off a wave of change in all walks of life. The marketing circle, which has always been “fashionable”, was the first to embrace AIGC technology. Relevant data shows that in 2023, nearly half of my country's advertisers will apply AIGC technology in online marketing activities, and more than 90% of these applications focus on content creation and creative development. This new technology-driven advertising and marketing model is gradually taking shape, bringing more possibilities for advertisers to reduce costs and increase efficiency. However, while AIGC technology is making great use in the marketing field, it also comes with many challenges. For example, AIGC technology may cause content risks when generating marketing materials, and heavily invested marketing activities may accidentally serve as a wedding dress for illegal products. So,
- AI 786 2024-07-18 01:41:21
-
- ICML 2024 | Gradient checkpointing too slow? Without slowing down and saving video memory, LowMemoryBP greatly improves backpropagation video memory efficiency
- The AIxiv column is a column where this site publishes academic and technical content. In the past few years, the AIxiv column of this site has received more than 2,000 reports, covering top laboratories from major universities and companies around the world, effectively promoting academic exchanges and dissemination. If you have excellent work that you want to share, please feel free to contribute or contact us for reporting. Submission email: liyazhou@jiqizhixin.com; zhaoyunfeng@jiqizhixin.com The first author of this paper is Yang Yuchen, a second-year master's student in the School of Statistics and Data Science of Nankai University, and his advisor is Associate Professor Xu Jun in the School of Statistics and Data Science of Nankai University. The research focus of Professor Xu Jun’s team is computer vision, generative AI and efficient machine learning, and they are working on top
- AI 679 2024-07-18 01:39:51
-
- WAIC 2024 Embodied Intelligence Collection|Gathering of top celebrities, exhibition of all categories of achievements: Innovation in embodied intelligence technology opens a new era of AI robots
- Embodied Intelligence: The new era of intelligent robots The embodied intelligence industry represents the in-depth integration and development of two cutting-edge technologies, artificial intelligence and robotics, allowing robots to achieve natural interaction with the environment through the collaboration of perception, cognition and action. The rise of embodiment means that artificial intelligence is moving from single information processing to more complex and multi-dimensional scene fields, opening a new era of deep integration of intelligent systems and human society. Key issues, application scenarios and cutting-edge results of embodied intelligence The 2024 World Artificial Intelligence Conference and High-Level Conference on Global Governance of Artificial Intelligence (hereinafter referred to as "WAIC2024") will focus in depth on the development trends of embodied intelligence, with major forums and rich intelligent robots Exhibits of innovative achievements provide a glimpse into the huge development potential of the embodied intelligence industry and jointly chart the era of intelligent robots.
- AI 588 2024-07-18 00:17:51
-
- These Features Make the ChatGPT Desktop App Better Than the Website
- Thanks to the ChatGPT app for macOS, you can now launch ChatGPT from any window on your Mac and use Voice Mode. OpenAI also added a few media attachment options to the app that aren't available on the website, such as screenshots and access t
- AI 1166 2024-07-17 22:50:21
-
- Topping the list of open source AI software engineers, UIUC's agent-less solution easily solves SWE-bench real programming problems
- The AIxiv column is a column where this site publishes academic and technical content. In the past few years, the AIxiv column of this site has received more than 2,000 reports, covering top laboratories from major universities and companies around the world, effectively promoting academic exchanges and dissemination. If you have excellent work that you want to share, please feel free to contribute or contact us for reporting. Submission email: liyazhou@jiqizhixin.com; zhaoyunfeng@jiqizhixin.com The authors of this paper are all from the team of teacher Zhang Lingming at the University of Illinois at Urbana-Champaign (UIUC), including: Steven Code repair; Deng Yinlin, fourth-year doctoral student, researcher
- AI 999 2024-07-17 22:02:05
-
- SOTA performance, Xiamen multi-modal protein-ligand affinity prediction AI method, combines molecular surface information for the first time
- Editor | KX In the field of drug research and development, accurately and effectively predicting the binding affinity of proteins and ligands is crucial for drug screening and optimization. However, current studies do not take into account the important role of molecular surface information in protein-ligand interactions. Based on this, researchers from Xiamen University proposed a novel multi-modal feature extraction (MFE) framework, which for the first time combines information on protein surface, 3D structure and sequence, and uses a cross-attention mechanism to compare different modalities. feature alignment. Experimental results demonstrate that this method achieves state-of-the-art performance in predicting protein-ligand binding affinities. Furthermore, ablation studies demonstrate the effectiveness and necessity of protein surface information and multimodal feature alignment within this framework. Related research begins with "S
- AI 1081 2024-07-17 18:37:10
-
- Exploration and practice of multi-modal remote sensing large models, Wang Jian, head of remote sensing large models of Ant Group, brings in-depth interpretation
- On July 5, under the guidance of the World Artificial Intelligence Conference Organizing Committee Office and the People's Government of Xuhui District, Shanghai, the 2024 WAIC Yunfan Award and Artificial Intelligence Youth were hosted by the Shanghai Artificial Intelligence Laboratory, this site, and the Global University Artificial Intelligence Academic Alliance. The forum was successfully held. The forum brought together more than 30 past and new Yunfan graduates from universities, research institutions and enterprises at home and abroad, including Stanford University, Oxford University, UCLA, University of California, ETH Zurich, University of Hong Kong, Tsinghua University, Peking University, Shanghai Jiao Tong University, etc. Award winners attended the conference offline, gathering the wisdom of international young AI scientists, actively exploring the boundaries of AI capabilities, and contributing new energy to China's AI development blueprint. Wang Jianzuo, the person in charge of remote sensing large-scale models of Ant Group, serves as the 2024 WAIC Yunfan
- AI 676 2024-07-17 18:03:14
-
- Quark upgrades 'Super Search Box” and launches one-stop AI service centered on AI search
- In the era of large models, how does generative AI revolutionize search products? Quark, a subsidiary of Alibaba Intelligent Information Business Group, "raised your hand to answer the question." On July 10, Quark upgraded the "Super Search Box" and launched a one-stop AI service centered on AI search, providing users with integrated information service value from retrieval, creation, summary, to editing, storage, and sharing. The new AI search box enables answering, creating, and summarizing the past. The search engine provides sorting of website lists based on keywords. Repeated selection, clicking, reading, and a large number of irrelevant results have become obstacles for users to obtain information efficiently, and it is difficult to get satisfactory answers to complex questions. The leap in AI technology has ignited a new value in search. User opens Quark version 7.0 search
- AI 537 2024-07-17 17:43:33
-
- Essential AI Tools and Tips for Every Student
- Students can use AI in several smart ways to improve their learning and work more efficiently. At this point, there are heaps of AI tools students can use, covering note-taking, tutoring, and much more. But which AI tools are the best for students?
- AI 1132 2024-07-17 16:49:04
-
- Completely change the language model: the new architecture TTT surpasses the Transformer, and the ML model replaces the RNN hidden state
- From 125M to 1.3B large models, the performance has been improved. It's unbelievable that this finally happened. A new large language model (LLM) architecture is expected to replace Transformer, which has been popular in the AI field so far, and its performance is better than Mamba. On Monday this week, the paper on Test-TimeTraining (TTT) became a hot topic in the artificial intelligence community. Paper link: https://arxiv.org/abs/2407.04620 The authors of this study are from Stanford University, University of California, Berkeley, University of California, San Diego, and Meta. They designed a new architecture, TTT, to replace the hidden state of RNN with a machine learning model. The module
- AI 525 2024-07-17 16:08:17
-
- Single-author paper, Google proposes millions of expert Mixture, surpassing dense feedforward and sparse MoE
- Unleashing the potential to further extend the Transformer while maintaining computational efficiency. Feedforward (FFW) layers in standard Transformer architectures result in a linear increase in computational cost and activation memory as the hidden layer width increases. As the size of large language models (LLM) continues to increase, the sparse mixed expert (MoE) architecture has become a feasible method to solve this problem, which separates the model size from the computational cost. Many emerging MoE models can achieve better performance and more powerful performance at the same size. The recently discovered scaling law of fine-grained MoE suggests that higher granularity leads to better performance. However, existing MoE models are limited to a low number of experts due to computational and optimization challenges.
- AI 571 2024-07-17 14:34:17
-
- Axiomatic training allows LLM to learn causal reasoning: the 67 million parameter model is comparable to the trillion parameter level GPT-4
- Show the causal chain to LLM and it learns the axioms. AI is already helping mathematicians and scientists conduct research. For example, the famous mathematician Terence Tao has repeatedly shared his research and exploration experience with the help of AI tools such as GPT. For AI to compete in these fields, strong and reliable causal reasoning capabilities are essential. The research to be introduced in this article found that a Transformer model trained on the demonstration of the causal transitivity axiom on small graphs can generalize to the transitive axiom on large graphs. In other words, if the Transformer learns to perform simple causal reasoning, it may be used for more complex causal reasoning. The axiomatic training framework proposed by the team is a new paradigm for learning causal reasoning based on passive data, with only demonstrations
- AI 1217 2024-07-17 10:14:38