近日,UC伯克利大学推出了一款名为「大世界模型」(LargeWorldModel, LWM)的新型人工智能模型,这一模型支持处理高达百万token的上下文信息,并能够生成视频内容。该模型一经推出便登上了GitHub热榜榜首,成为当前开源世界模型中最受瞩目的存在。LWM的上下文窗口长度达到100万token,与谷歌同时推出的Gemini 1.5模型持平。LWM不仅能够处理多模态信息,还在处理大规模文本数据方面表现出色,能够在一小时内准确找到目标文本。此外,LWM还能一口气看完1小时的视频内容,展现了其在视频处理方面的强大能力。这一模型的推出,无疑为人工智能领域带来了新的发展方向和可能性。
Title: UC Berkeley Unveils Million-Token Video Generation Model
Keywords: UC Berkeley, Multimodal Information Processing, Large-World Model
News content:
UC Berkeley recently unveiled a new AI model called “Large World Model” (LWM), capable of processing up to a million tokens of context and generating video content. The model has quickly climbed to the top of the GitHub trending list, becoming the most popular open-source world model. LWM’s context window length reaches 100,000 tokens, on par with Google’s Gemini 1.5 model. In addition to processing multimodal information, LWM performs exceptionally well in handling large-scale text data, accurately finding target text within the million-token context. Furthermore, LWM can consume an hour-long video in a single session, showcasing its powerful video processing capabilities. The introduction of this model brings new directions and possibilities to the field of artificial intelligence.

【来源】https://mp.weixin.qq.com/s/52uUGcgcoT6oGhZvi-Dl-w

Views: 1

发表回复

您的邮箱地址不会被公开。 必填项已用 * 标注