Search AI Products and News

AI agent tool Manus announced a strategic partnership with Alibaba's Tongyi Qianwen to launch a Chinese version. This collaboration will leverage Tongyi Qianwen's open-source model and utilize domestic computing platforms to implement Manus' full functionality, aiming to provide Chinese users with a more convenient and efficient AI application experience. Manus is a project incubated by the startup Monica.im, founded in July 2023, which focuses on developing consumer-oriented AI tools and browser extensions. The founding team...

Today, early in the morning, Apple officially launched the iPhone 16 series at the highly anticipated autumn product launch event, while highlighting its latest artificial intelligence technology - Apple Intelligence. During the event, Apple announced a major upgrade for the domestic iPhone, and the Chinese version of Apple Intelligence is expected to be launched next year. This news marks Apple's further expansion in the field of artificial intelligence and its in-depth布局 in the Chinese market.

Apple recently released the iOS 18.1 developer preview Beta update for iPhone users, with the most notable new feature being the introduction of Apple Intelligence. However, this feature is currently in testing phase and only partially available, and it is temporarily unavailable on the iPhone 15 Pro and iPhone 15 Pro Max sold in China.

Meta has launched the Llama3.1 open source model series, which includes a 405B parameter version that performs comparably to GPT-4. Among them, the 8B parameter version Llama3.1-8B-Instruct supports multiple languages with a context length of up to 131072 tokens and has been trained on 25 million synthetic data entries, demonstrating cognitive and reasoning capabilities similar to GPT-3.5 Turbo. Based on this model, OpenBuddy has released OpenBuddy-L, which supports Chinese Q&A and cross-language translation.

By expanding the vocabulary, selecting data, and employing a multi-stage training strategy, the Chinese version of LLaMA2 was trained in 15 hours at a cost of a few thousand yuan. The performance of LLaMA2 on various Chinese tasks has significantly improved, reaching an advanced level compared to models of the same scale. The building process, code, and weights are all open source, making it transferable to other languages and fields. Achieve low-cost large model training by training the Chinese version of LLaMA2 in just 15 hours with a cost of a few thousand yuan. All training code and pre-trained weights are open source and can be directly applied to other languages and fields for rapid low-cost large model training.

The Colossal-AI team has developed a highly performant Chinese version of LLaMA-2 at a low cost. The Chinese LLaMA-2 has excelled in multiple evaluation rankings. Colossal-AI has open-sourced the complete training process, code, and weights. They also provide the evaluation framework ColossalEval. Colossal-AI's solution can be used to build large models for any vertical field.