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 latest news! Baidu Wenxin Big Model 4.0: The largest parameter model in the history of Wanka training, see you as soon as next week
- Yesterday, Cailian News exclusively revealed that Baidu’s Wenxin Model 4.0 is intensifying its training and is close to being ready for release. Everyone has always been curious about Wen Xinyiyan's information. Today we also got more news about Wenxin 4.0, which involves key information such as underlying architecture, infrastructure, training data sets, costs, etc. It has a very high degree of credibility! Let’s talk about the core conclusions first: 1. Yesterday’s revelations are basically true. It is currently understood that Wenxin Large Model 4.0 has actually been tested with small traffic. 2. The number of parameters of Wenxin 4.0 is larger than that of all LLMs with publicly released parameters. It is also the first large model in China to be trained using Wanka cluster. 3. The reasoning cost is much higher than that of Wenxin 3.5, it is said to be about 8-10 times! (Large models are really expensive!) If
- AI 893 2024-07-19 12:04:49
-
- Who is bigger, 9.11 or 9.9? We actually tested 15 large models, and more than half of them overturned.
- Editor of Machine Power Report: Yang Wen The big models still can’t solve simple math problems. In the past few days, a prompt to test whether the large model "brain" is working has become popular - which one is bigger, 9.11 or 9.9? This math problem, which even elementary school students can answer correctly, stumps a group of "heroes" in the large-scale model industry. Here's the thing. ScaleAI's senior prompt engineer Riley Goodside asked GPT-4o the prompt word "9.11and9.9 - which is bigger?", but got the answer "the former is bigger". Other large models also overturned. On July 17, we took 12 large domestic models, plus the foreign GPT-4o
- AI 837 2024-07-19 11:27:21
-
- Mistral AI two consecutive releases: 7B mathematical reasoning dedicated, Mamba2 architecture code large model
- Netizens are curious whether Mathstral can solve the problem of "who is bigger, 9.11 or 9.9?" Yesterday, the AI circle was overwhelmed by a simple question like "Who is bigger, 9.11 or 9.9?" Big language models including OpenAIGPT-4o, Google Gemini, etc. all overturned. This allows us to see that large language models cannot understand and give correct answers like humans when dealing with some numerical problems. For numerical and complex mathematical problems, specialized models are more specialized. Today, the French large model unicorn MistralAI released a 7B large model "Mathstral" focused on mathematical reasoning and scientific discovery to solve advanced data that requires complex, multi-step logical reasoning.
- AI 185 2024-07-19 09:54:11
-
- Cost reduction by more than 90%, 'active learning + experimental workflow' accelerates catalyst development
- Editor | Green Luo The synthesis of higher alcohols (HAS) through thermal catalytic hydrogenation of syngas remains a promising technology. Chain growth and CO insertion requirements require multicomponent materials whose complex reaction kinetics and broad chemical space do not meet catalyst design specifications. Here, researchers from ETH Zurich propose an alternative strategy to integrate active learning into the experimental workflow, using the FeCoCuZr catalyst series as an example. The proposed data-assisted framework simplifies navigation of a wide space of compositions and reaction conditions across 86 experiments, reducing environmental footprint and cost by more than 90% compared to traditional procedures. It identifies the Fe65Co19Cu5Zr11 catalyst with optimized reaction conditions that operates stably
- AI 845 2024-07-19 09:49:21
-
- so cool! Old iPhone, iPad, and MacBook devices form a heterogeneous cluster and can run Llama 3
- If you have some spare equipment, you might want to give it a try. This time, the hardware equipment in your hand can also flex its muscles in the field of AI. By combining iPhone, iPad, and Macbook, you can assemble a "heterogeneous cluster inference solution" and then run the Llama3 model smoothly. It is worth mentioning that this heterogeneous cluster can be a Windows system, Linux, or iOS system, and support for Android will be coming soon. Heterogeneous cluster is running. According to the project author @evilsocket, this heterogeneous cluster includes iPhone15ProMax, iPadPro, MacBookPro (M1Max), NVIDIA GeFor
- AI 858 2024-07-19 05:09:59
-
- Exclusive conversation with Li Yan: Funded by Suhua, Jingwei, and Redpoint, the first 'generative recommendation” startup | AI Pioneers
- Humanity is ushering in an explosive update in the field of artificial intelligence. Every step in the expansion of technology into the unknown has attracted astonishing attention. In the process of expanding the boundaries of artificial intelligence, innovation and disagreement coexist in the technical routes of important tracks. The judgment and choices of technology pioneers influence the footsteps of many followers. In the past year, this website exclusively took the lead in introducing outstanding companies such as Dark Side of the Moon, Shengshu Technology, Aishi Technology, and Wuwen Core Dome to everyone, leaving them with the first "10,000-word interview script" in the Internet world. At a stage when the technical roadmap has not yet converged, we have seen the leadership of AI entrepreneurs who truly have faith, courage and systematic cognition. Therefore, we launch “AIPioneers”
- AI 768 2024-07-19 05:03:59
-
- Posthumous work of the OpenAI Super Alignment Team: Two large models play a game, and the output becomes more understandable
- If the answer given by the AI model is incomprehensible at all, would you dare to use it? As machine learning systems are used in more important areas, it becomes increasingly important to demonstrate why we can trust their output, and when not to trust them. One possible way to gain trust in the output of a complex system is to require the system to produce an interpretation of its output that is readable to a human or another trusted system, that is, fully understandable to the point that any possible errors can be found. For example, to build trust in the judicial system, we require courts to provide clear and readable written opinions that explain and support their decisions. For large language models, we can also adopt a similar approach. However, when taking this approach, ensure that the language model generates
- AI 585 2024-07-19 01:29:52
-
- Tsinghua University won the Best Paper + Time Test Award, Shandong University received an honorable mention, and the SIGIR 2024 awards were announced
- Tsinghua University’s results are outstanding. The 47th International Computer Society Conference on Information Retrieval (ACMSIGIR) will be held in Washington, DC, USA from July 14th to 18th, 2024. This conference is the top academic conference in the field of information retrieval. Just now, the conference announced the Best Paper Award, Best Paper Runner-up, Best Paper Honorable Mention Award, and Time Test Award. Among them, Tsinghua University, Hillhouse School of Artificial Intelligence at Renmin University of China, and the Xiaohongshu team won the best paper; researchers from the University of Glasgow and the University of Pisa won the runner-up; the honorable mention award for the best paper was awarded to Shandong University (Qingdao) ), Leiden University, and the University of Amsterdam; the Time Test Award was awarded to researchers from Tsinghua University and the University of California, Santa Cruz. Next, let's
- AI 338 2024-07-19 00:06:43
-
- Login to Science, drug affinity increased 37 times, AI performs unsupervised optimization of protein and antibody complexes
- Editor | Radish skin proteins are involved in many biological functions such as cell composition, muscle contraction, digestion of food, and identification of viruses. In order to design better proteins (including antibodies), scientists often repeatedly mutate amino acids (the units that make up proteins in a certain order) at different positions until the protein obtains the desired function. But there are more amino acid sequences than there are grains of sand in the world, so finding the best proteins, and thus the best potential drugs, is often difficult. When faced with this challenge, scientists often spend millions of dollars and test on miniaturized, simplified versions of biological systems. “This requires a lot of guesswork and verification.”
- AI 644 2024-07-18 22:22:51
-
- How does the brain process language? Princeton team analyzes Transformer model
- Editor | Radish Skin When processing language, the brain deploys specialized computations to construct meaning from complex linguistic structures. Artificial neural network based on Transformer architecture is an important tool for natural language processing. Princeton University researchers explore the Transformer model and the functional specialization of the human brain in language processing. Transformer calculates and integrates contextual information between words through structured circuits. However, current research mainly focuses on the internal representations ("embeddings") generated by these circuits. The researchers analyzed circuit calculations directly: they deconstructed these calculations into functionally specialized "transformations" that integrate contextual information across words. Exploit participants
- AI 495 2024-07-18 20:52:41
-
- Doubao Big Model Team releases new Detail Image Caption evaluation benchmark to improve the reliability of VLM Caption evaluation
- 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 current visual language model (VLM) mainly performs performance evaluation through QA question and answer format, but lacks evaluation of the basic understanding of the model, such as reliable evaluation methods for detailimagecaption performance. In response to this problem, the Chinese Academy of Sciences,
- AI 598 2024-07-18 20:10:02
-
- Samsung China Galaxy Z series new products access bean bag large model
- On July 17, Samsung Electronics released a new generation of Galaxy Z series products for the Chinese market. At the meeting, Samsung Electronics and Volcano Engine officially announced their cooperation to connect bean bag models to the smart assistants and AI vision of Galaxy Z Fold6 and Galaxy Z Flip 6 mobile phones to enhance the smart application experience of mobile phones. Previously, Samsung announced its in-depth cooperation with Google Gemini at overseas new product launches. In China, it selected manufacturers such as Volcano Engine as large model partners. fenye caption: The smart assistant and AI visual access bean bag model of Samsung Galaxy Z Fold6 and Galaxy Z Flip 6 mobile phones. In addition to the AI functions that have been disclosed such as circle search, real-time translation, recording transcription, etc., this time
- AI 404 2024-07-18 20:07:33
-
- Abandoning the visual encoder, this 'native version' multi-modal large model is also comparable to mainstream methods
- 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 Diao Haiwen is a doctoral student at Dalian University of Technology, and his supervisor is Professor Lu Huchuan. Currently interning at Beijing Zhiyuan Artificial Intelligence Research Institute, the instructor is Dr. Wang Xinlong. His research interests are vision and language, efficient transfer of large models, multi-modal large models, etc. Let’s make Cui together
- AI 214 2024-07-18 19:21:11
-
- Are all these VLMs blind? GPT-4o and Sonnet-3.5 successively failed the 'vision' test
- The four major VLMs are all trying to fool the blind? Let the most popular SOTA models (GPT-4o, Gemini-1.5, Sonnet-3, Sonnet-3.5) count how many intersections there are between two lines. Will they perform better than humans? The answer is probably no. Since the launch of GPT-4V, visual language models (VLMs) have made the intelligence of large models a big step closer to the level of artificial intelligence we imagined. VLMs can both understand images and use language to describe what they see, and perform complex tasks based on these understandings. For example, if you send the VLM model a picture of a dining table and a picture of a menu, it can extract the number of beer bottles and the unit price on the menu from the two pictures, and calculate
- AI 489 2024-07-18 18:18:02
-
- MotionClone: No training required, one-click cloning of video movements
- 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 No training or fine-tuning is required. The movement of the reference video can be cloned in the new scene specified by the prompt word. Whether it is global camera movement or local body movement, it can be done with one click. Paper: https://arxiv.org/abs/2406.05
- AI 858 2024-07-18 17:06:12