上海人工智能实验室携手商汤科技，以及香港中文大学和复旦大学的科研力量，近日共同推出了创新性的大语言模型——书生·浦语2.0，并宣布该模型正式开源。这一举措标志着在人工智能领域的又一重大突破，旨在回归语言建模的基础，提升模型的智能水平。
书生·浦语2.0的设计理念专注于提升语料质量和信息密度，以实现语言建模能力的显著增强。该模型的突出特点是支持高达200K token的上下文处理，能够一次性处理约30万汉字的输入，这在业界堪称一大创新。这样的能力使得书生·浦语2.0能够在海量文本中精准地捕捉关键信息，即使在长文本的“大海捞针”任务中也能游刃有余。
这一开源项目不仅将促进学术界与产业界的深度合作，也将推动AI技术在自然语言处理领域的广泛应用，为科研人员和开发者提供更为强大的工具。书生·浦语2.0的发布，无疑将进一步推动中国在人工智能研究和应用领域的领先地位，为全球科技发展注入新的活力。据来源澎湃新闻，这一开源行动已引起广泛关注，预示着未来在自然语言理解和生成方面的革新性进展。
英语如下：
News Title: “Scholar·Pǔyu 2.0 Open Source: A Return to Language Modeling, Leading AI Text Processing to New Heights”
Keywords: Pǔyu 2.0 Open Source, AI Language Model, Multi-School Collaboration
News Content: The Shanghai Artificial Intelligence Laboratory, in collaboration with SenseTime and the research forces from the Chinese University of Hong Kong and Fudan University, recently jointly launched an innovative large language model called Scholar·Pǔyu 2.0, announcing that the model is now open source. This move signifies another major breakthrough in the field of artificial intelligence, aiming to回归语言建模的本质 and enhance the model’s intelligence.
The design philosophy of Scholar·Pǔyu 2.0 centers on improving the quality of the linguistic corpus and information density, resulting in a significant boost in language modeling capabilities. A standout feature of the model is its support for processing up to 200K tokens of context, allowing it to handle approximately 300,000 Chinese characters in one go – a major innovation in the industry. This ability enables Scholar·Pǔyu 2.0 to precisely capture key information from vast amounts of text, excelling in tasks that involve finding needles in a haystack of lengthy texts.
This open-source project not only fosters deep collaboration between academia and industry but also promotes the widespread application of AI technology in natural language processing. It provides researchers and developers with more powerful tools. The release of Scholar·Pǔyu 2.0 undoubtedly propels China’s leading position in AI research and application, injecting new vitality into global technological development. According to sources from Phoenix News, this open-source initiative has attracted widespread attention, foreshadowing groundbreaking progress in the future of natural language understanding and generation.
【来源】https://www.thepaper.cn/newsDetail_forward_26040295
Views: 1