Large Model Daily, January 9

News4months agorelease AIWindVane
22 0
Large Model Daily, January 9

[Large Model Daily, January 9] The first self-taught embodied 3D graphics and text large model system that can be used in a variety of furniture and home appliances; the AI wave swept CES 2024, and NVIDIA’s stock price hit a record high! In the second quarter, H20 chips will be mass-produced for domestic use; Baichuan Intelligent releases a large model Baichuan-NPC that can create game characters with simple text descriptions; Nvidia will cooperate with miHoYo, Tencent, NetEase and other game companies in AI digital human business

The first self-taught 3D graphic and text-based large model system that can be used in a variety of furniture and home appliances.
Just by showing it a demonstration video and 10 hours of training, Figure-01 can learn to use the coffee machine, from inserting the coffee capsule to pressing the start button, all in one go. But I want the robot to be self-taught and be able to use it proficiently without any demonstration videos when seeing various furniture and home appliances for the first time. This is a difficult problem to solve. It not only requires the robot to have strong visual perception, decision-making and planning capabilities, but also precise manipulation skills. Now, a three-dimensional embodied graphic and text large model system provides new ideas for the above problems. This system combines a precise geometric perception model based on three-dimensional vision with a two-dimensional graphic and text large model that is good at planning. It can solve complex long-term tasks related to furniture and home appliances without the need for sample data. This research was completed by the team of Professor Leonidas Guibas of Stanford University, Professor Wang He of Peking University, and Zhiyuan Artificial Intelligence Research Institute.

The success rate has increased fourfold. Dongda University and Zhejiang Normal University have proposed a general framework for materials synthesis that integrates AI, high-throughput experiments and prior knowledge of chemistry.
Over the past few years, data-driven machine learning (ML) techniques have become a powerful tool for the design and discovery of advanced materials. However, materials synthesis is often much more complex than property and structure predictions due to the need to consider the availability of precursors, experimental conditions, and reactants, and few computational predictions can be realized experimentally. To address these challenges, a research team from Southeast University and Zhejiang Normal University proposed a general framework that integrates high-throughput experiments, chemical prior knowledge, and machine learning techniques such as subgroup discovery and support vector machines to guide The experimental synthesis of materials can reveal the structure-property relationships hidden in high-throughput experiments and quickly screen out materials with high synthetic feasibility from the vast chemical space. By applying the proposed method to solve the challenging and subsequent synthesis problems of two-dimensional silver/bismuth (AgBi) organic-inorganic hybrid perovskites, the success rate of synthetic feasibility is increased by four times compared to traditional methods. This study provides a practical approach to solving multidimensional chemical acceleration problems using small data sets from typical laboratories with limited available experimental resources.

The wave of industrial AI sweeps through CES 2024, and NVIDIA’s stock price hits a record high! Mass production of domestically produced H20 chips in the second quarter
In a major speech held before the opening of CES this Monday, NVIDIA focused on generative AI models and their role in creating realistic characters, and also launched a new GPU designed for gamers and content creators. design. NVIDIA’s stock price rose 6.43% that day to close at US$522.53 per share, with its market value reaching US$1.29 trillion, a record high. NVIDIA also launched the GeForce RTX 40 SUPER series of GPUs, including the RTX 4080 SUPER, RTX 4070 Ti SUPER, and RTX 4070 SUPER, all delivering powerful gaming and generative AI performance. The GeForce RTX 4080 SUPER supports 4K panoramic ray tracing gaming, and its performance surpasses the GeForce RTX 3080 Ti in some aspects.

Baichuan Intelligent Releases Large Model Baichuan-NPC You can build game characters with just a simple text description
Baichuan Intelligent released the large character model Baichuan-NPC, and launched a customized solution of “character creation platform + search enhanced knowledge base”. Through this solution, game manufacturers can quickly build characters and achieve character customization through simple text descriptions without writing any code.

Getty and NVIDIA launch AI image generation tool for iStock gallery platform
Getty Images further cooperates with NVIDIA to launch AI text-to-image generation function for its library platform iStock, priced at $14.99. Users can generate 100 times, providing four different images each time. The resulting images can be used commercially and come with legal copyright protection up to $10,000.

Benchmarking GPTs but not creating a threshold, AI Agent players at the application layer are finally here
At the beginning of the new year, major manufacturers began to roll out AI agents. In the past year, the trends in the artificial intelligence industry have continued to change. At first, the large language models and AIGC applications that sprung up attracted much attention, and then many AI-assisted applications based on large language models appeared (such as Microsoft Copilot). . Nowadays, AI Agent has become a fierce competition point and is the key to the next stage of AIGC. Even though it is promising in every possible way, it is not difficult to find that AI Agent is mostly played by developers, geeks and enterprise users. It requires a certain amount of professional knowledge, which limits the harvest of a large number of users at the application level. This requires manufacturers to consider how to lower the barriers to creation and use when launching AI Agents. If it can be created directly on an existing product, that would be even better. On January 9, at DingTalk’s 2024 Annual Product Launch – My Super Assistant event, DingTalk, a national-level office software with 700 million users, released a new version 7.5, and launched an intelligent platform that has the opportunity to benchmark GPTs. The product – AI assistant, allows each person and every company to customize a personalized and exclusive super assistant.

NVIDIA will cooperate with miHoYo, Tencent, NetEase and other game companies on AI digital human business
At the CES 2024 exhibition in the United States, NVIDIA released a series of new products, among which the NVIDIA ACE microservice made its debut. This is a technology platform that uses generative AI to create virtual digital humans. Current partners with ACE include MiHoYo and NetEase Games , Zhangqu Technology, Tencent Games, Ubisoft and other companies.

After a year of hard work, the first stable version of LangChain is finally released. LangGraph builds intelligent agents into graphs.
Unconsciously, LangChain has been around for a year. As an open source framework, LangChain provides the modules and tools needed to build AI applications based on large models, greatly lowering the threshold for AI application development, allowing anyone to build their own creative applications based on large models such as GPT-4. LangChain itself has been evolving over the past year. Just now, LangChain officially announced that their first stable version, LangChain v0.1.0, has been released.

NVIDIA partners with Deepcell to accelerate generative AI applications in single-cell research
On January 8, Deepcell, a pioneer in AI-driven single cell analysis, announced a research collaboration with NVIDIA to accelerate the development and adoption of advanced computer vision solutions in the life sciences. Deepcell already uses NVIDIA A4000 and NVIDIA AI technology and will integrate NVIDIA AI into its single-cell analysis technology, working with NVIDIA to co-develop new uses of generative AI and multi-modal applications in cell biology. The joint collaboration aims to advance understanding of cell morphology and ultimately accelerate the use of AI-driven cell analysis in a wide range of cell biology and translational research, including cancer, stem cells and cell therapy.

© Copyright notes

Related posts

No comments

No comments...