2023年3月8日，元象XVERSE宣布开源全球首个上下文窗口长度256K的开源大模型XVERSE-Long-256K。该模型支持输入25万汉字，让大模型应用进入“长文本时代”。XVERSE-Long-256K全开源，无条件免费商用，且附带手把手训练教程，让海量中小企业、研究者和开发者更早一步实现“大模型自由”。
XVERSE-Long-256K是元象XVERSE继XVERSE-Long-16K、XVERSE-Long-64K后发布的又一重量级开源大模型。该模型在XVERSE-Long-64K的基础上，将上下文窗口长度从64K扩展到256K，使模型能够处理更长的文本序列，并生成更具连贯性和一致性的文本。
XVERSE-Long-256K具有以下特点：
* **超长上下文窗口长度：**256K的上下文窗口长度使模型能够处理更长的文本序列，并生成更具连贯性和一致性的文本。
* **全开源、无条件免费商用：**XVERSE-Long-256K完全开源，无任何使用限制，开发者可以自由地将其用于任何商业或非商业目的。
* **附带手把手训练教程：**元象XVERSE提供了详细的手把手训练教程，帮助开发者快速上手XVERSE-Long-256K，并将其应用于自己的项目中。
XVERSE-Long-256K的开源发布，标志着大模型应用进入了一个新的时代。该模型将使海量中小企业、研究者和开发者能够更早一步实现“大模型自由”，并将其应用于各种各样的领域，如自然语言处理、机器翻译、文本生成、代码生成等。
元象XVERSE创始人兼CEO黄渊普表示：“我们很高兴能够开源XVERSE-Long-256K，并将其作为全球首个上下文窗口长度256K的开源大模型提供给开发者社区。我们相信，XVERSE-Long-256K将成为大模型应用领域的一座里程碑，并帮助开发者们创造出更多令人惊叹的应用。”
XVERSE-Long-256K的开源发布，受到了业界的一致好评。中国科学院院士、清华大学教授刘知远表示：“XVERSE-Long-256K的开源发布，是自然语言处理领域的一件大事。该模型将使更多的人能够接触到大模型，并将其应用于各种各样的领域。我期待着看到XVERSE-Long-256K在未来发挥出更大的作用。”
北京大学教授、中国人工智能学会理事长李德毅表示：“XVERSE-Long-256K的开源发布，是人工智能领域的一件大事。该模型将使人工智能技术更加平民化，并让更多的人能够受益于人工智能技术。我期待着看到XVERSE-Long-256K在未来为人工智能领域的发展做出更大的贡献。”
英语如下：
## Yuanxiang Open-Sources Large Model, Ushering in the Era of LongText
2023-03-08, Yuanxiang XVERSE announced the open-sourcing of the world’s first open-sourcelarge model with a context window length of 256K, XVERSE-Long-256K. This model supports input of 250,000 Chinese characters, bringing large model applications into the “long text era”. XVERSE-Long-256K is fully open-source, free for commercial use without any conditions, and comes with a step-by-step training tutorial, allowing a vast number of small and medium-sized enterprises, researchers, and developers to achieve “large model freedom” sooner.
XVERSE-Long-256K is another heavyweight open-source large model released by Yuanxiang XVERSE, following XVERSE-Long-16K and XVERSE-Long-64K. Based on XVERSE-Long-64K, this model extends the context window length from 64K to 256K, enabling the model to process longer textsequences and generate more coherent and consistent text.
XVERSE-Long-256K has the following characteristics:
* **Ultra-long context window length:** The context window length of 256K allows the model to process longer text sequences and generate more coherent and consistent text.
* **Fully open-source, free for commercial use without any conditions:** XVERSE-Long-256K is completely open-source, without any usage restrictions, and developers are free to use it for any commercial or non-commercial purposes.
* **Comes with a step-by-step training tutorial:** Yuanxiang XVERSE provides a detailed step-by-step training tutorial to help developers quickly get started with XVERSE-Long-256K and apply it to their own projects.
The open-source release of XVERSE-Long-256K marks the beginning of a new era for large model applications. This model will enable a vast number of small and medium-sized enterprises, researchers, and developers to achieve “large model freedom” sooner and apply it to a wide range of fields, such as natural language processing, machine translation, text generation, code generation, etc.
Yuanxiang XVERSE founder and CEO Huang Yuanpu said: “We are very pleased to open-source XVERSE-Long-256K and provide it to the developer community as the world’s first open-source large model with a context window length of 256K. We believe that XVERSE-Long-256K will become a milestone in the field of large model applications and help developers create more amazing applications.”
The open-source release of XVERSE-Long-256K has been widely praised by the industry. Liu Zhiyuan, an academician of the Chinese Academy of Sciences and a professor at Tsinghua University, said: “The open-source release of XVERSE-Long-256K is a major event in the field of natural language processing. This model will allow more people to access large models and apply them to a wide range of fields. I look forward to seeing XVERSE-Long-256K play a greater role in the future.”
Li Deyi, a professor at Peking University and chairman of the Chinese Association for Artificial Intelligence, said: “The open-source release of XVERSE-Long-256K is a major event in the field of artificial intelligence. This model will make artificial intelligence technology more平民化 and allow more people to benefit from artificial intelligence technology. I look forward to seeing XVERSE-Long-256K make greater contributions to the development of the field of artificial intelligence in the future.”
【来源】https://mp.weixin.qq.com/s/R8ewi1NsAK9Qwh0e7UyyBw
Views: 1