近日，万兴科技发布了我国首个音视频多媒体大模型——“天幕”。该大模型聚焦数字创意垂类创作场景，以音视频生成式AI技术为基础，支持全球不同语言。其训练数据来源于15亿创作者及100亿本土化高质量音视频，相关能力已在海外规模化商用。
值得一提的是，万兴科技还将与华为云及马投算力达成三方算力合作，天幕大模型的训练将在国内算力和服务器基础上进行。这一合作将进一步提升大模型的性能和应用范围。
此次万兴“天幕”的发布，标志着我国在音视频多媒体大模型领域取得了重要突破。作为中国首个音视频多媒体大模型，其研发中心的落户地为马栏山。未来，万兴科技将继续深化与华为云及马投算力的合作，共同推动我国音视频多媒体大模型的发展。
英文翻译：
News Title: Wanxing Technology Releases China’s First Audio-visual Multimedia Large Model
Keywords: Audio-visual large model, Huawei Cloud, Malanshan
News Content:
Recently, Wanxing Technology has released China’s first audio-visual multimedia large model – “Tianmu”. The large model focuses on the creative scenarios of digital art and is based on audio-visual generative AI technology. It supports different languages around the world. The training data comes from 150 million creators and 10 billion local high-quality audio-visual videos, and the relevant capabilities have been commercially scaled overseas.
It is worth mentioning that Wanxing Technology will also form a tripartite cooperation with Huawei Cloud and Matou Computing Power, and the training of Tianmu large model will be carried out on domestic computing power and servers. This cooperation will further enhance the performance and application scope of the large model.
The release of Wanxing’s “Tianmu” marks an important breakthrough in the field of audio-visual multimedia large models in China. As China’s first audio-visual multimedia large model, its research and development center will be located in Malanshan. In the future, Wanxing Technology will continue to deepen its cooperation with Huawei Cloud and Matou Computing Power to jointly promote the development of audio-visual multimedia large models in China.
【来源】https://www.tmtpost.com/nictation/6916252.html
Views: 1