商汤科技发布“日日新SenseNova 4.0”，大模型开启全新AI体验
北京时间2023年3月8日，商汤科技发布“日日新SenseNova 4.0”，全面升级大模型体系，带来更全面的知识覆盖、更可靠的推理能力、更优越的长文本理解力、更稳定的数字推理能力和更强的代码生成能力，并支持跨模态交互。
“日日新SenseNova 4.0”的核心是日日新·商量大语言模型-通用版本（SenseChat V4），支持128K语境窗口长度，综合整体评测成绩水平比肩GPT 4，相较GPT 3.5已经实现全面超越。
SenseChat V4在自然语言处理、计算机视觉、语音识别等多项任务上均取得了显著提升。在自然语言处理方面，SenseChat V4的文本生成能力大幅增强，能够生成更流畅、更符合逻辑、更具创造性的文本。在计算机视觉方面，SenseChat V4的图像理解能力得到提升，能够更准确地识别和分类图像中的物体。在语音识别方面，SenseChat V4的语音识别准确率和鲁棒性均有提高，能够在嘈杂环境中更准确地识别语音。
此外，“日日新SenseNova 4.0”还支持跨模态交互，能够同时处理文本、图像、语音等多种模态的数据，并从中提取有价值的信息。这使得“日日新SenseNova 4.0”能够应用于更广泛的场景，例如多模态搜索、智能客服、内容创作等。
商汤科技表示，“日日新SenseNova 4.0”的发布标志着商汤科技在大模型领域取得了新的突破。未来，商汤科技将继续加大对大模型的研发投入，推动大模型技术在各行各业的落地应用，赋能千行百业的智能化转型。
英语如下：
**Headline: SenseTime’s Large Model Debuts, Rivaling GPT-4**
**Keywords:** Large model, SenseNova, GPT-4
**News Content:** SenseTime Technology Releases “SenseNova 4.0”,Large Model Opens Up New AI Experience
Beijing, March 8, 2023 – SenseTime Technology released “SenseNova 4.0”, comprehensively upgrading its large model system, bringing more comprehensive knowledge coverage, more reliable reasoning ability, superior long text comprehension, more stable digital reasoning ability, and stronger codegeneration ability, and supporting cross-modal interaction.
The core of “SenseNova 4.0” is the SenseChat V4, a large-scale language model with a context window length of 128K. Its comprehensive evaluation results are comparable to GPT-4, and it has surpassed GPT-3.5.
SenseChat V4 has made significant improvements in various tasks such as natural language processing, computer vision, and speech recognition. In terms of natural language processing, SenseChat V4’s text generation capabilities have been greatly enhanced, and it can generate smoother, more logical, and more creative text.In terms of computer vision, SenseChat V4’s image understanding capabilities have been improved, and it can more accurately recognize and classify objects in images. In terms of speech recognition, SenseChat V4’s speech recognition accuracy and robustness have both been improved, and it can more accurately recognize speech in noisy environments.
In addition, “SenseNova 4.0” also supports cross-modal interaction, and can simultaneously process data from multiple modalities such as text, images, and speech, and extract valuable information from it. This enables “SenseNova 4.0” to be applied to a wider range of scenarios, such as multimodal search, intelligent customer service, and content creation.
SenseTime Technology said that the release of “SenseNova 4.0” marks a new breakthrough for SenseTime Technology in the field of large models. In the future, SenseTime Technology will continue to increase its investment in the research and development of large models, promote the application of large model technology in various industries, and empower the intelligent transformation of all walks of life.
【来源】https://tidenews.com.cn/news.html?id=2705294&source=1
Views: 1