For the first time, the performance of Taizhiyun's FFM-Llama3-70B in a large-scale Traditional Chinese language model surpasses GPT 4.0_0.
Following the release of the "Traditional Chinese Enhanced Version FFM-Llama 2 Large Language Models", Taizhi Cloud today took the lead in the industry by launching the latest large language models (LLM) FFM-Llama3-70B and FFM-Llama3-8B, as well as the AFS ModelSpace model API service. These models are immediately available on cloud services, accelerating the deployment of generative AI applications by enterprises.
According to TMMLU+ testing, Taizhiyun's FFM-Llama3-70B outperformed GPT 4.0 for the first time in Traditional Chinese Large Language Model (LLM). The FFM-Mixtral-8x7B, FFM-Mistral-7B, FFM-Llama2-v2, and FFM-Llama3-8B models in the FFM family also surpassed GPT 3.5 in overall score, solidifying Taizhiyun's leading position in Traditional Chinese LLM models. The newly added support for Function Calling allows for quick integration with external applications (such as Google Maps, search engines, transportation booking, and real-time financial stock quotes), enabling developers to easily enhance model capabilities and accelerate the comprehensive development of commercial applications. It is also the only Traditional Chinese large language model with this feature.
Taizhi Cloud today launched the latest Large Language Model (LLM) for FFM-Llama3-70B and FFM-Llama3-8B, along with the AFS ModelSpace API, ahead of the industry.
AFS ModelSpace provides enterprises with a variety of open-source large language models, automation tools, and management services. The newly released open-source foundational models, including FFM-Llama3 (8B, 70B), FFM-Mistral, Code Llama, and TAIDE, simplify management procedures, allowing for direct inference and rapid model switching. The newly added "Public Mode" API billing model uses token-based usage, significantly reducing costs and time for enterprises with low to medium usage.
Taizhi Cloud leads enterprises into the AI 2.0 era, pioneering applications in smart healthcare.
At today's GenAI online technology forum hosted by TaiAbra, TaiAbra's Chief Technology Officer, Chen Zhongcheng, shared: "TaiAbra focuses on providing enterprises with the generative AI they need, including AIHPC high-speed computing power, reliable Traditional Chinese and Large Scale Language Models, and No-code platforms. From the cloud to the ground, we can effectively help enterprises solve various considerations such as cost, time, manpower, and cybersecurity. We will also continue to support excellent open-source models to provide enterprises with the most complete and best solutions."
At the same time, Liu Li, Vice President of Medical Information at Show Chwan Medical System, was invited to share the AFS service of Taiwan Smart Cloud. Through the Function Calling function, high-quality medical record summaries and reports are output in a structured manner, and then GenAI performs inference analysis to provide diagnostic suggestions for reference, effectively reducing the workload of medical staff.
For more information, please see:Formosa Foundation Models"Special Page"