Search AI Products and News

Recently, NVIDIA unveiled its brand-new Blackwell platform and showcased its initial performance in the MLPerf Training 4.1 benchmark tests. According to the test results, Blackwell achieved a doubling of performance in certain aspects compared to the previous Hopper platform, garnering widespread attention in the industry. In the MLPerf Training 4.1 benchmark tests, the Blackwell platform excels in LLM (Large Language Models).

Hugging Face recently released LightEval, a lightweight AI evaluation tool designed to assist enterprises and research institutions in more accurately assessing large language models. This tool emphasizes the importance of evaluation in the application of AI technologies, helping to ensure the accuracy, impartiality, and alignment of model outputs with business objectives. Addressing the limitations of traditional evaluation methods, LightEval offers customized evaluation solutions and integrates with other tools from Hugging Face, such as Datat.

Cerebras Systems has launched Cerebras Inference, claiming it to be the fastest AI inference service in the world, achieving performance that surpasses traditional GPU-based systems by 20 times with significantly improved cost-effectiveness, particularly suitable for processing large language models (LLMs). Its 8B version processes 1800 tokens per second, while the 70B version processes 450 tokens, with speed and cost-performance far exceeding NVIDIA GPU solutions.

Hugging Face and NVIDIA have launched an inference-as-a-service powered by NVIDIA's NIM technology, enhancing prototype development and efficient deployment of open-source AI models. This service notably supports robust models like Llama2 and Mistral AI, offering five times faster processing than traditional methods. Through NVIDIA DGX Cloud, developers can train models, and Hugging Face announced team profitability and expansion to 220 members, ....

US Senators urge limitations on Amazon, Microsoft, and Google's development of large language models. Warren believes tech giants have too much power in the AI field and proposed banning them from developing cutting-edge AI technologies, calling for the splitting of Google and Amazon and restrictions on their operations.

The powerful nature of AI technology and large language models, Bezos's optimistic view on AI, AI's role in ensuring humanity does not self-destruct, and how Amazon's voice assistant is preparing to become more intelligent.

Microsoft executive Eric Boyd revealed that the company plans to launch more large language models to meet the needs of enterprise customers. Boyd emphasized that Microsoft's AI applications and LLMs are safe and usable, but companies focused on text generation can develop more quickly. Microsoft has used OpenAI models in its products but plans to remain competitive and may establish relationships with other companies. This move responds to Amazon's CEO's criticism that cloud service providers are limited to one model provider.

The Amazon CEO stated that generative AI will change every customer experience. Generative AI will make it easier for ordinary developers and business users to utilize. It will further enhance Amazon's ability to predict purchase intent. Amazon's virtual assistant Alexa will become more advanced. Amazon has the opportunity to become a leader in generative AI.

Research from the Santa Fe Institute shows that GPT-4 has only a 33% accuracy rate on graphic reasoning tasks. The multimodal version, GPT-4v, performs even worse with only a 25% accuracy rate, raising concerns about its graphic processing capabilities. Testing with the ConceptARC dataset reveals that the average accuracy of humans on graphic reasoning tasks reaches 91%, significantly higher than GPT-4. Researchers raised questions about the input methods used for participants and GPT-4, including sample randomness and how image transformations could alter concepts. The study highlights the limitations of large

Researchers have developed the 3D-GPT system, powered by large language models (LLM), which can generate complex 3D models and scenes based on brief text descriptions. The system breaks down 3D modeling tasks into manageable components through a multi-agent system, including task scheduling agents, conceptualization agents, and modeling agents. The key role of the task scheduling agent is to select the appropriate program generation functions based on user instructions, improving the efficiency of task coordination. 3D-GPT supports continuous instruction input, allowing users to edit and modify generated scenes, including object control.

📚 Large language models (LLMs) like ChatGPT and GPT-4 face challenges in passing the CFA exam. 📊 Research shows that GPT-4 significantly outperforms ChatGPT but still encounters difficulties, especially in higher-level exams. 🔍 Specialized training and knowledge acquisition targeted at the CFA exam may be key to improving AI's performance in the finance sector.

Research shows that large pre-trained language models (LLMs), such as GPT-3, can improve themselves through the RAIN method to meet human needs without additional data and fine-tuning. The RAIN method is widely applicable to various language generation tasks, requiring no additional models or data storage, nor does it rely on labeled data or training. RAIN enhances the performance of LLMs through self-assessment, reducing the success rate of adversarial attacks, and enabling AI to generate more coherent and secure responses. This study introduces RAIN as a technique for adjusting LLMs to align with human preferences.

Arthur has launched the open source tool ArthurBench for evaluating and comparing the performance of large language models. ArthurBench helps companies test the performance of different language models on specific use cases and provides metrics such as accuracy, readability, and risk mitigation for comparison. Financial services firms, automotive manufacturers, and media platforms have already begun using ArthurBench, accelerating analysis and providing more accurate answers.

{1: Microsoft Research Asia has proposed a method for industrial control using pre-trained large language models (LLM). 2: The research utilizes the GPT-4 model to control air conditioning systems, achieving results superior to traditional reinforcement learning methods with only a few example samples. 3: The goal of this research is to explore the potential of directly using pre-trained LLMs for industrial control tasks and to design methods for addressing industrial control tasks with lower technical debt.}