4月23日，商汤科技在中国北京举行新品发布会，正式发布人工智能大模型“日日新5.0”。该模型采用先进的MOE（Mixture of Experts）混合专家架构，并基于超过10TB的数据tokens进行深度学习训练。此外，日日新5.0大模型还拥有高达200K的推理上下文窗口。
据悉，日日新5.0大模型的开发旨在全面对标OpenAI的GPT-4 Turbo。这一举措标志着我国在人工智能领域的发展迈出了重要的一步。商汤科技一直以来都是人工智能领域的先锋，其发布的日日新5.0大模型不仅在技术上对标了GPT-4 Turbo，更在某种程度上代表了中国在人工智能领域的决心和实力。
MOE（Mixture of Experts）混合专家架构是一种将多个专家网络的输出进行组合的方法，可以提高模型的泛化能力和鲁棒性。日日新5.0大模型正是采用了这种结构，使得其在处理各种复杂任务时具有更高的准确性和效率。
基于超过10TB的数据tokens进行深度学习训练，使得日日新5.0大模型在处理各种任务时具有更丰富的知识和理解能力。同时，高达200K的推理上下文窗口也使得模型能够处理更长的文本，从而满足更多复杂场景的需求。
此次商汤科技发布的日日新5.0大模型，是对其在人工智能领域技术实力的再次展示，也是对OpenAI的GPT-4 Turbo的一次有力挑战。我们期待商汤科技在未来能够带给我们更多的惊喜，推动我国人工智能领域的发展。
英语如下：
**News Headline: SenseTime Releases DailyX 5.0 Large Model: Benchmarking GPT-4 Turbo**
Keywords: SenseTime, DailyX 5.0, Benchmarking GPT-4
**News Content:**
### SenseTime Unveils DailyX 5.0 Large Model, Benchmarking GPT-4 Turbo
On April 23, SenseTime, a leading AI company, held a new product launch event in Beijing, China, and officially released the AI large model “DailyX 5.0.” The model employs an advanced MOE (Mixture of Experts) architecture and is trained with over 10TB of data tokens through deep learning. Additionally, the DailyX 5.0 large model boasts an impressive inference context window of up to 200K.
It is understood that the development of the DailyX 5.0 large model is aimed at fully benchmarking OpenAI’s GPT-4 Turbo. This move signifies an important step in China’s development in the field of artificial intelligence. SenseTime has always been at the forefront of the AI sector, and its release of the DailyX 5.0 large model not only technically benchmarks GPT-4 Turbo but also represents China’s determination and strength in the field of artificial intelligence to some extent.
The MOE (Mixture of Experts) architecture is a method that combines the outputs of multiple expert networks, which can enhance the model’s generalization ability and robustness. DailyX 5.0 large model employs this structure, making it more accurate and efficient in handling various complex tasks.
Trained with over 10TB of data tokens through deep learning, the DailyX 5.0 large model possesses a wealth of knowledge and understanding capabilities, enabling it to handle various tasks effectively. Moreover, the high-capacity inference context window of up to 200K allows the model to process longer texts, meeting the needs of more complex scenarios.
The release of the DailyX 5.0 large model by SenseTime is another demonstration of its technical strength in the field of artificial intelligence and a significant challenge to OpenAI’s GPT-4 Turbo. We look forward to SenseTime bringing us more surprises in the future and driving the development of China’s AI sector.
【来源】https://www.guandian.cn/article/20240423/402856.html
Views: 2