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：
-
- High-gloss reflective terminator? Google NeRF-Casting: Ray tracing can do it!
- NeRF is no longer "afraid" of near specular reflections. Early NeRF variants used multilayer perceptrons (MLPs) to map from 3D coordinates to volumetric density and viewpoint-dependent color, but representing detailed 3D geometry and color required training of large MLPs and evaluation is extremely slow. Recent work has focused on making NeRF more efficient by replacing large MLPs with voxel grid-like data structures or a combination of grids and small MLPs. While scalable to represent detailed large-scale scenes, its advantages are limited to three-dimensional geometry and predominantly diffuse color. Expanding NeRF's ability to model the viewpoint-dependent appearance of reality remains a challenge. The current advanced model for view synthesis of shiny objects has limitations in two aspects: it can only synthesize distant ambient light
- AI 1207 2024-06-07 09:27:53
-
- LLM | Yuan 2.0-M32: Expert Mixture Model with Attention Routing
- Picture 1. The conclusion is written above. Yuan+2.0-M32 is an infrastructure, similar to Yuan-2.0+2B, using an expert hybrid architecture containing 32 experts. 2 of these experts are active. An expert hybrid architecture containing 32 experts is proposed and adopted to select experts more efficiently. Compared with the model using the classic routing network, the accuracy rate is improved by 3.8%. Yuan+2.0-M32 is trained from scratch, using 2000B tokens, and its training consumption is only 9.25% of that of a dense ensemble model of the same parameter size. In order to better select experts, the attention router is introduced, which has the ability to sense quickly and thus enable better selection of experts. Yuan2.0-
- AI 603 2024-06-07 09:06:30
-
- Constructing Scaling Law from 80 models: a new work by a Chinese doctoral student, highly recommended by the author of the thinking chain
- In the field of AI, scaling laws (Scalinglaws) are a powerful tool for understanding LM scaling trends. They provide a guideline for researchers. This law provides an important guide for understanding how the performance of language models changes with scale. But unfortunately, scaling analysis is not common in many benchmarking and post-training studies because most researchers do not have the computational resources to build scaling laws from scratch, and open models are trained on too few scales to make reliable scaling predictions. . Researchers from Stanford University, University of Toronto and other institutions have proposed an alternative observation method: Observational Scaling Laws, which combines the functions of language models (LM) with cross-multiple models.
- AI 406 2024-06-06 20:40:36
-
- Reshaping the cloud 'build” experience in the era of generative AI
- In the era of generative AI, the changes in the cloud computing industry are accelerating across the board. The time to build a new cloud has arrived. As a developer, how can we adapt to technological innovations with ease? How to quickly seize new technology opportunities to get started and grow quickly? Join the 2024 Amazon Cloud Technology China Summit, a must-go event for developers, which can help you answer your questions and get rewards! Amazon Cloud Technology is reshaping the one-stack construction experience of development, operation and maintenance, and optimization in the generative AI era on the cloud. At this summit, we will bring you a new developer generative AI exploration journey, including immersive star products Experience, hands-on special training and peak challenges, skills certification and free learning, technology forward-looking sharing, global community leader dialogue, developer creative market, etc., help developers enjoy unlimited
- AI 826 2024-06-06 18:48:01
-
- Human preference is the ruler! SPPO alignment technology allows large language models to compete with each other and compete with themselves
- 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 Richard Sutton made this evaluation in "TheBitterLesson": "The most important lesson that can be drawn from 70 years of artificial intelligence research is that those general methods that use computing will eventually is the most effective and has the advantage
- AI 447 2024-06-06 18:32:31
-
- The visual representation model vHeat inspired by physical heat transfer is here. It attempts to break through the attention mechanism and has both low complexity and global receptive field.
- 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 article are the original members of VMamba. The first author Wang Zhaozhi is a 2022 jointly trained doctoral student between the University of Chinese Academy of Sciences and Pengcheng Laboratory, and the co-author Liu Yuesi Direct PhD candidate from the University of Chinese Academy of Sciences in 2021. Their main research direction is visual
- AI 526 2024-06-06 17:28:46
-
- DenserRadar: 4D millimeter wave radar point cloud detector based on dense LiDAR point cloud
- Original title: DenserRadar: A4Dmillimeter-waveradarpointclouddetectorbasedondenseLiDARpointclouds Paper link: https://arxiv.org/pdf/2405.05131 Author affiliation: Tsinghua University Paper idea: 4D millimeter wave (mmWave) radar is known for its robustness in extreme environments and broad Detection range and the ability to measure speed and altitude have shown significant potential to enhance perception when autonomous driving systems face corner-cases. However, the inherent sparsity and noise limitations of 4D millimeter wave radar point clouds
- AI 769 2024-06-06 14:10:54
-
- The evaluation results of the bean bag large model are revealed, which is 19% higher than the previous generation 'Skylark'
- Recently, the large bean bag model was officially released at the Volcano Engine Power Conference. While the price reduction trend of large models is promoted at ultra-low prices, Doubao’s model capabilities have also attracted industry attention. In a product information of Volcano Engine, the Doubao Model team released some of the internal test results of the first phase: on the public evaluation sets of 11 mainstream industries such as MMLU, BBH, GSM8K, HumanEval, etc., the total score of Doubao-pro-4k It scored 76.8 points, which is a 19% increase compared to the 64.5 points of the previous generation model Skylark 2. This is also better than other domestic models tested during the same period. This evaluation was completed in May this year and mainly included nine domestic large language models including Universal Model Pro, Skylark2, and
- AI 357 2024-06-06 13:45:41
-
- Yann LeCun: ViT is slow and inefficient. Real-time image processing still depends on convolution.
- In the era of unification of Transformers, is it still necessary to study the CNN direction of computer vision? At the beginning of this year, OpenAI’s large video model Sora made the VisionTransformer (ViT) architecture popular. Since then, there has been an ongoing debate about who is more powerful, ViT or traditional convolutional neural network (CNN). Recently, Turing Award winner and Meta chief scientist Yann LeCun, who has been active on social media, also joined the discussion on the dispute between ViT and CNN. The cause of this incident was that Harald Schäfer, CTO of Comma.ai, was demonstrating his latest research. He (like many recent AI scholars) cue Yann LeCun's expression, although
- AI 1036 2024-06-06 13:25:02
-
- Sanofi partners with OpenAI and Formation Bio to accelerate AI drug discovery
- Editor | This collaboration will be a game changer for the pharmaceutical industry. By combining data, artificial intelligence technology and expertise in drug development, they aim to revolutionize the way new medicines are discovered and brought to market. Sanofi CEO Paul Hudson said: "This collaboration is an important step in our journey to become an AI-powered pharmaceutical company." OpenAI COO Brad Lightcap said: "AI has tremendous potential to accelerate drug development. We
- AI 535 2024-06-06 12:54:35
-
- Improved detection algorithm: for target detection in high-resolution optical remote sensing images
- 01 Outlook Summary Currently, it is difficult to achieve an appropriate balance between detection efficiency and detection results. We have developed an enhanced YOLOv5 algorithm for target detection in high-resolution optical remote sensing images, using multi-layer feature pyramids, multi-detection head strategies and hybrid attention modules to improve the effect of the target detection network in optical remote sensing images. According to the SIMD data set, the mAP of the new algorithm is 2.2% better than YOLOv5 and 8.48% better than YOLOX, achieving a better balance between detection results and speed. 02 Background & Motivation With the rapid development of remote sensing technology, high-resolution optical remote sensing images have been used to describe many objects on the earth’s surface, including aircraft, cars, buildings, etc. Object detection in the interpretation of remote sensing images
- AI 903 2024-06-06 12:33:01
-
- Tsinghua University took over and YOLOv10 came out: the performance was greatly improved and it was on the GitHub hot list
- The benchmark YOLO series of target detection systems has once again received a major upgrade. Since the release of YOLOv9 in February this year, the baton of the YOLO (YouOnlyLookOnce) series has been passed to the hands of researchers at Tsinghua University. Last weekend, the news of the launch of YOLOv10 attracted the attention of the AI community. It is considered a breakthrough framework in the field of computer vision and is known for its real-time end-to-end object detection capabilities, continuing the legacy of the YOLO series by providing a powerful solution that combines efficiency and accuracy. Paper address: https://arxiv.org/pdf/2405.14458 Project address: https://github.com/THU-MIG/yo
- AI 1298 2024-06-06 12:20:45
-
- To improve the utilization of optical data sets, the Tianda team proposed an AI model to enhance spectral prediction effects
- Editor | Dead Leaf Butterfly Recently, the team of Associate Professor Wu Liang and Academician Yao Jianquan of the Institute of Laser and Optoelectronics of Tianjin University and the team of Professor Xiong Deyi of the Natural Language Processing Laboratory reported a solution that uses a deep learning model with multi-frequency supplementary input to enhance the spectral prediction effect. . This scheme can improve the accuracy of spectral prediction by using multi-frequency input data. In addition, this solution can also reduce noise interference in the spectrum prediction process, thereby improving the prediction effect. This solution can improve the utilization of existing optical data sets and enhance the prediction effect of the spectral response corresponding to the metasurface structure without increasing the training cost. Relevant research results are titled "Enhancedspectrumpredictionusingdeep
- AI 620 2024-06-06 12:09:28
-
- A single 4090 inferable, 200 billion sparse large model 'Tiangong MoE' is open source
- In the wave of large models, training and deploying state-of-the-art dense set LLMs poses huge challenges in terms of computational requirements and associated costs, especially at scales of tens or hundreds of billions of parameters. To address these challenges, sparse models, such as Mixture of Experts (MoE) models, have become increasingly important. These models offer an economically viable alternative by distributing computation to various specialized sub-models, or "experts," with the potential to match or even exceed the performance of dense set models with very low resource requirements. On June 3, another important news came from the open source large model field: Kunlun Wanwei announced the open source 200 billion sparse large model Skywork-MoE, which significantly reduces the inference cost while maintaining strong performance. Based on the previous Kunlun Wanwei open source Skywo
- AI 900 2024-06-05 22:14:46
-
- HuggingFace teaches you how to make a SOTA visual model
- There was OpenAI's GPT-4o before, and Google's series of kings followed. Advanced multi-modal large models hit the market one after another. Other practitioners were shocked and began to think about how to catch up with these super models again. In this paper by HuggingFace and Sorbonne University in France, they summarized the key experiences in building large visual models and pointed out a way for developers. These experiences in the pictures cover many aspects such as model architecture selection, training methods, and training data. The author gives a detailed summary after multiple comparisons. The core points include: If you want to do a good job in large visual models, the choice of architecture is very important. The language model has a greater impact on overall performance than the visual module. Adopting a staged pre-training strategy is more conducive to building model capabilities. The training data should include
- AI 906 2024-06-05 21:39:58
Tool Recommendations
jQuery enterprise message form contact code
jQuery enterprise message form contact code is a simple and practical enterprise message form and contact us introduction page code.
form button
2024-02-29
HTML5 MP3 music box playback effects
HTML5 MP3 music box playback special effect is an mp3 music player based on HTML5 css3 to create cute music box emoticons and click the switch button.
Player special effects
2024-02-29
HTML5 cool particle animation navigation menu special effects
HTML5 cool particle animation navigation menu special effect is a special effect that changes color when the navigation menu is hovered by the mouse.
Menu navigation
2024-02-29
jQuery visual form drag and drop editing code
jQuery visual form drag and drop editing code is a visual form based on jQuery and bootstrap framework.
form button
2024-02-29
Organic fruit and vegetable supplier web template Bootstrap5
An organic fruit and vegetable supplier web template-Bootstrap5
Bootstrap template
2023-02-03
Bootstrap3 multifunctional data information background management responsive web page template-Novus
Bootstrap3 multifunctional data information background management responsive web page template-Novus
backend template
2023-02-02
Real estate resource service platform web page template Bootstrap5
Real estate resource service platform web page template Bootstrap5
Bootstrap template
2023-02-02
Simple resume information web template Bootstrap4
Simple resume information web template Bootstrap4
Bootstrap template
2023-02-02
Cute summer elements vector material (EPS PNG)
This is a cute summer element vector material, including the sun, sun hat, coconut tree, bikini, airplane, watermelon, ice cream, ice cream, cold drink, swimming ring, flip-flops, pineapple, conch, shell, starfish, crab, Lemons, sunscreen, sunglasses, etc., the materials are provided in EPS and PNG formats, including JPG previews.
PNG material
2024-05-09
Four red 2023 graduation badges vector material (AI EPS PNG)
This is a red 2023 graduation badge vector material, four in total, available in AI, EPS and PNG formats, including JPG preview.
PNG material
2024-02-29
Singing bird and cart filled with flowers design spring banner vector material (AI EPS)
This is a spring banner vector material designed with singing birds and a cart full of flowers. It is available in AI and EPS formats, including JPG preview.
banner picture
2024-02-29
Golden graduation cap vector material (EPS PNG)
This is a golden graduation cap vector material, available in EPS and PNG formats, including JPG preview.
PNG material
2024-02-27
Home Decor Cleaning and Repair Service Company Website Template
Home Decoration Cleaning and Maintenance Service Company Website Template is a website template download suitable for promotional websites that provide home decoration, cleaning, maintenance and other service organizations. Tip: This template calls the Google font library, and the page may open slowly.
Front-end template
2024-05-09
Fresh color personal resume guide page template
Fresh color matching personal job application resume guide page template is a personal job search resume work display guide page web template download suitable for fresh color matching style. Tip: This template calls the Google font library, and the page may open slowly.
Front-end template
2024-02-29
Designer Creative Job Resume Web Template
Designer Creative Job Resume Web Template is a downloadable web template for personal job resume display suitable for various designer positions. Tip: This template calls the Google font library, and the page may open slowly.
Front-end template
2024-02-28
Modern engineering construction company website template
The modern engineering and construction company website template is a downloadable website template suitable for promotion of the engineering and construction service industry. Tip: This template calls the Google font library, and the page may open slowly.
Front-end template
2024-02-28
