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:
-
- The marketing effect has been greatly improved, this is how AIGC video creation should be used
- After more than a year of development, AIGC has gradually moved from text dialogue and picture generation to video generation. Looking back four months ago, the birth of Sora caused a reshuffle in the video generation track and vigorously promoted the scope and depth of AIGC's application in the field of video creation. In an era when everyone is talking about large models, on the one hand we are surprised by the visual shock brought by video generation, on the other hand we are faced with the difficulty of implementation. It is true that large models are still in a running-in period from technology research and development to application practice, and they still need to be tuned based on actual business scenarios, but the distance between ideal and reality is gradually being narrowed. Marketing, as an important implementation scenario for artificial intelligence technology, has become a direction that many companies and practitioners want to make breakthroughs. Once you master the appropriate methods, the creative process of marketing videos will be
- AI 551 2024-06-25 00:01:11
-
- 'Encyclopedia' of AI small molecule drug discovery, reviewed by researchers from Cornell, Cambridge, EPFL and others published in Nature sub-journal
- Author | Editor Du Yuanqi of Cornell University | ScienceAI As AI for Science receives more and more attention, people are more concerned about how AI can solve a series of scientific problems and can be successfully used for reference in other similar fields. AI and small molecule drug discovery are one of the most representative and early explored fields. Molecular discovery is a very difficult combinatorial optimization problem (due to the discrete nature of the molecular structure) and the search space is very large and rugged. At the same time, it is very difficult to verify the properties of the searched molecules. It usually requires expensive experiments, at least simulation calculations, Quantum chemical methods to provide feedback. With the rapid development of machine learning and benefiting from early exploration (including the construction of simple and usable optimization goals and effect measures)
- AI 593 2024-06-24 21:20:21
-
- Choosing the smartest AI in the Olympiad: Claude-3.5-Sonnet vs. GPT-4o?
- 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 research team of Shanghai Jiao Tong University’s Generative Artificial Intelligence Laboratory (GAIRLab)’s main research directions are: large model training, alignment and evaluation. Team homepage: https://plms.ai/AI technology is changing with each passing day. Recently, Anthr
- AI 1189 2024-06-24 17:01:06
-
- From RLHF to DPO to TDPO, large model alignment algorithms are already 'token-level'
- 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 In the development process of artificial intelligence, the control and guidance of large language models (LLM) has always been one of the core challenges, aiming to ensure that these models are both powerful and safe serve human society. Early efforts focused on reinforcement learning methods through human feedback (RL
- AI 829 2024-06-24 15:04:43
-
- Why have you given up on LangChain?
- Perhaps from the day it was born, LangChain was destined to be a product with polarizing reputations. Those who are optimistic about LangChain appreciate its rich tools and components and its ease of integration, while those who are not optimistic about LangChain believe that it is doomed to fail - in this era of rapid technological change, it is simply not feasible to build everything with LangChain. Even more exaggerated: "In my consulting work, I spend 70% of my energy convincing people not to use langchain or llamaindex. This solves 90% of their problems." Recently, a LangChain complaint became a hot topic again Discussion focus: The author FabianBoth is an AI testing tool O
- AI 1191 2024-06-24 14:20:32
-
- ICML 2024 | Feature pollution: Neural networks learn irrelevant features and fail to generalize
- Paper title: FeatureContamination: NeuralNetworksLearnUncorrelatedFeaturesandFailtoGeneralize Paper link: https://arxiv.org/pdf/2406.03345 Code link: https://github.com/trzhang0116/feature-contamination With the achievements of large models represented by GPT in recent years A huge success, the machine learning paradigm of deep neural network + SGD + scaling once again proved its dominance in the field of AI. why based on
- AI 923 2024-06-24 14:17:11
-
- CVPR 24|ETH Zurich and other teams: Redefining the small sample 3D segmentation task, the new benchmark opens up broad potential for improvement!
- 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 article, An Zhaochong, is currently studying for a PhD at the University of Copenhagen, under the supervision of Serge Belongie. He graduated from the Federal Institute of Technology in Zurich (ETHZurich) with a master's degree. During his master's degree, he worked under the guidance of his mentor Luc Van
- AI 914 2024-06-24 14:05:51
-
- ByteDouBao's new image Tokenizer: only 32 tokens are needed to generate an image, and the speed is increased by up to 410 times.
- 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 In the rapid development of generative models, ImageTokenization plays a very important role, such as VAE that Diffusion relies on or VQGAN that Transformer relies on. These Tokes
- AI 1233 2024-06-24 14:03:31
-
- Huawei Cloud uses Pangu large model to reshape services, and ModelArts Studio is fully upgraded
- Huawei Developer Conference 2024 will be held in Dongguan. At the conference, Huawei released the latest progress in HarmonyOSNext, Pangu Model 5.0 and other aspects, and introduced a series of application cases of advanced AI technologies. Zhang Yuxin, CTO of Huawei Cloud, said: "Based on the capabilities of the Pangu model, Huawei Cloud has reshaped a series of cloud services and created an AINative cloud to help make thousands of industries intelligent." According to reports, Huawei Cloud combines the capabilities of the Pangu model and Huawei's Combining the data and experience accumulated in various fields such as product research and development, data governance, security protection, and business operation and maintenance, we "jumped our own parachute" and applied it to Huawei Cloud CodeArts, DataArts, MetaStudio, GaussDB, and cloud security.
- AI 438 2024-06-22 11:55:38
-
- First-hand review of Claude 3.5: Is it really better than GPT-4o for playing tricks, seeing a doctor, playing tricks, and doing math problems?
- Editor of the report on the power of machines: Yang Wen plays tricks, sees doctors, plays tricks, and does math problems. Is the ability of "new king" Claude 3.5 really that mysterious? It's coming, it's coming, it's coming with Claude3.5Sonnet! After being dormant for three months, just last night, OpenAI’s “strong rival” Anthropic launched a new generation model-Claude3.5Sonnet! What’s unique about this large model? First, it has a better grasp of nuance, humor, and complex instructions, and the writing tone is more natural and approachable. It’s also Anthropic’s strongest visual model, excelling at tasks like interpreting charts, graphs, or transcribing text from imperfect images. In addition, it is used in reasoning, reading comprehension
- AI 1160 2024-06-22 07:46:19
-
- Covering text, positioning and segmentation tasks, Zhiyuan and Hong Kong Chinese jointly proposed the first multi-functional 3D medical multi-modal large model
- Author | Editor Bai Fan, The Chinese University of Hong Kong | ScienceAI Recently, the Chinese University of Hong Kong and Zhiyuan jointly proposed the M3D series of work, including M3D-Data, M3D-LaMed and M3D-Bench, to promote 3D medical images from all aspects of data sets, models and evaluations Development of analytics. (1) M3D-Data is currently the largest 3D medical image data set, including M3D-Cap (120K3D image and text pairs), M3D-VQA (510K question and answer pairs), M3D-Seg (150K3DMask), M3D-RefSeg (3K inference segmentation ) in total four sub-datasets. (2) M3D-LaMed is currently the most versatile 3D medical multi-modal large model that can
- AI 774 2024-06-22 07:16:49
-
- Whether it's real or AI video, 'Mosca” can reconstruct and restore 4D dynamic renderable scenes.
- 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 Lei Jiahui is a doctoral student in the Department of Computer Science at the University of Pennsylvania (2020-present). His supervisor is Professor KostasDaniilidis. His current main research direction is the modeling and representation of four-dimensional dynamic scene geometry. Algorithms and applications. He is in computer vision
- AI 789 2024-06-22 07:09:52
-
- Automatically identify the best molecules and reduce synthesis costs. MIT develops a molecular design decision-making algorithm framework
- Editor | Ziluo AI’s use in streamlining drug discovery is exploding. Screen billions of candidate molecules for those that may have properties needed to develop new drugs. There are so many variables to consider, from material prices to the risk of error, that weighing the costs of synthesizing the best candidate molecules is no easy task, even if scientists use AI. Here, MIT researchers developed SPARROW, a quantitative decision-making algorithm framework, to automatically identify the best molecular candidates, thereby minimizing synthesis costs while maximizing the likelihood that the candidates have the desired properties. The algorithm also determined the materials and experimental steps needed to synthesize these molecules. SPARROW takes into account the cost of synthesizing a batch of molecules at once, since multiple candidate molecules are often available
- AI 1195 2024-06-22 06:43:34
-
- WAIC Forum Registration|Gathering together many central state-owned enterprises and AI companies to discuss the integrated development of industries
- As the engine of new productivity, artificial intelligence is rapidly integrating with thousands of industries. On March 5, Premier Li Qiang of the State Council introduced the government's work tasks this year in his government work report, proposing to deepen the research and development applications of big data, artificial intelligence, etc., carry out the "artificial intelligence +" action, and create an internationally competitive digital industry cluster. Since then, the strategy of artificial intelligence empowering industrial development has accelerated. In order to further assist the process of "artificial intelligence +", showcase and share the experiences of successful explorers, and facilitate cooperation between supply and demand parties in the field of artificial intelligence, this site hosted the key experiment of the Ministry of Industry and Information Technology for the evaluation of artificial intelligence scenario application and intelligent systems. "2024WAIC Artificial Intelligence Empowers Industrial Integration Development" co-organized by the Office
- AI 1229 2024-06-22 04:59:29
-
- Xiaoice and ShiMeta co-create ShiMeta digital human solution: with the blessing of large models, every large-screen terminal becomes a new interactive carrier
- On June 21, 2024, Xiaoice and ShiMeta, a leading AIoT company, announced a strategic cooperation to jointly launch the ShiMeta digital human solution. The two parties jointly create various forms of digital human intelligent interactive large-screen terminals, integrating Xiaoice's AI large model and digital human technology with Shimatel's rich hardware business ecosystem, allowing digital humans to enter various industry scenarios and serve Qianxing. Baiye injects new productivity, improves user experience and service efficiency, and brings more intelligent changes. XiaoIce × Simatech: The rich hardware business ecological solution combines Xiaoice's large models and advanced digital human technologies such as neural network rendering and super natural speech with Simatech's high-performance motherboards, high-definition cameras, microphone arrays, equipment management platforms, etc. Conducted in-depth integration to realize large-screen terminals
- AI 666 2024-06-22 02:22:30