Meta has recently announced that it will hold the first LlamaCon developer conference on April 29, focusing on the latest advancements in its Llama series of generative AI models. The event aims to provide developers with a platform to share the latest achievements in open-source AI, helping them to better build applications and products. As generative AI technology rapidly evolves, Meta hopes to further promote the development of its ecosystem through this conference. At the upcoming event, Meta plans to showcase several.

The National Supercomputing Internet Platform recently announced the launch of the "AI Ecosystem Partner Acceleration Program," offering a series of benefits for enterprise users, including three months of free access to the DeepSeek API and supporting it with a computing power resource pool of ten million core hours. It is reported that the platform has completed the deployment of the full version of the DeepSeek model image. Currently, the platform has achieved interconnection with over 20 supercomputing and intelligent computing centers across 14 provinces, aggregating over 6,500 types of computing products, including nearly 2 AI model services.

IBM recently announced that its AI development platform watsonx.ai now supports the DeepSeek-R1 distilled versions of the Llama 3.18B and Llama 3.370B models. DeepSeek optimizes multiple Llama and Qwen variants using knowledge distillation technology, leveraging data from the R1 model to further enhance model performance. On the watsonx.ai platform, users can utilize DeepSeek in two ways.

Mark Zuckerberg recently announced that Meta will invest up to $65 billion by 2025 to enhance its artificial intelligence (AI) projects. He posted on Facebook that 2025 will be a crucial year for AI, with expectations that Meta's AI assistants will serve over 1 billion users, and the Llama4 model will become an industry-leading cutting-edge model. Additionally, Meta plans to develop an AI engineer capable of gradually contributing code to the company's R&D efforts.

Recently, a study led by the Austrian Institute of Complexity Science (CSH) revealed that despite large language models (LLMs) excelling in various tasks, they show significant shortcomings when tackling advanced history questions. The research team tested three leading models, including OpenAI's GPT-4, Meta's Llama, and Google's Gemini, with disappointing results. Note: The image was generated by AI, and the image is licensed by Midjourney to evaluate these models in history.

Meta is facing a copyright infringement lawsuit, with the plaintiff's attorney claiming that CEO Mark Zuckerberg approved the use of a dataset containing pirated e-books and articles to train its Llama AI model. This case is one of many copyright lawsuits against several tech giants accused of using copyrighted works for AI model training without authorization. In documents filed in the U.S. District Court for the Northern District of California on Wednesday evening, the plaintiffs reiterated Meta's actions from the end of last year.

IBM has officially launched its next-generation open-source large language model, Granite3.1, aiming to secure a leading position in the enterprise AI sector. This series of models features an extended context length of 128K, embedding models, built-in hallucination detection capabilities, and significant performance improvements. IBM claims that the Granite8B Instruct model outperforms its open-source competitors of similar scale, including Meta's Llama3.1, Qwen2.5, and Google's Gemma2.

In the age of AI, large language models (LLMs) are akin to martial arts secrets; their training processes consume massive computing power and data, much like martial arts experts who have spent years in secluded training. The release of open-source models is like these experts sharing their secrets with the public, but they come with certain licenses (such as Apache 2.0 and LLaMA 2 Community License) to protect their intellectual property (IP). However, the world can be treacherous, and incidents of 'shelling' are bound to occur. Some developers claim they have trained new LLMs when, in fact, they are using other foundational models (like Lla