这两天大语言模型有什么重要进展

you know it's not just the chip controls that we put on so small yards when we were controlling just a few handful of high end chips that are really good for training large language artificial intelligence models so the vast majority of chips that are made in the world are made by us companies or korean companies japanese companies are free to flow to china without any kind of licensing， but the select few at a certain level of capability we've put controls on and not only did we control those chips we control the manufacturing equipment that can make those chips which is incredibly important as well because given the right equipment trying to produce their own chips i mean they have the capability， they have the capacity they have the engineers so controlling the manufacturing equipment is incredibly important and again it's not all manufacturing equipment it's just a certain level of manufacturing equipment that we're controlling now so that's the small yard high fence thing now with that all that said what's gonna go on next so we're always looking they can make it a while i can make a decent chip it's not quite as good as nine one hundred， but it's a decent ai a chip called in a seven nine ten b they're doing that through brew force it's not the best method in process for making those chips， but they're working on those processes so we're always looking well if they're doing that what should we be doing that also needs to be protected in that small yard high fence。

兄弟们，我这回正在奥利奥斯云集大会的现场，这里就是通易展区。通易大模型系列是阿里自研的统一底座大模型系列，包括通易千万和通易万象两个基础模型，以及通易 human aigc 数字人。咱们先看看这个大语言模型，通易千万到底聪不聪明，他能作为阅读助手一键提炼总结， 也能作为智能翻译看这段实时日语翻译，反应呢是相当快。还有 ppt 创作功能，通过一句话、一段音视频或者一个常文档 生成 ppt， 而且还能智能编排和设计，这功能真的太牛了，运用到工作和学习中，那是多么惬意啊！再来看看阿里通易万象全新上线的 ai 视频生成大模型，它支持纹身视频和图身视频两种模式， 具备强大的画面视觉动态生成能力，擅长概念理解与组合生成，并且通音外向知识复杂与大幅度的运动生成，能够还原真实世界的物 处理规律。一只大熊猫走在天空中的云朵上，让他生成一个视频，哇，这个画面质感绝对是影视级的！又来到了创意互动展区，这个太极生万象，他是通过摄像头捕捉定格动作模型，会识别用户的个性化穿搭，自动匹配多种风格化效果，快速制作同款图像和视频，真的太酷了！ 这边还有个通易高效百万创作人计划，咱们再看看同学们创建的智能体工具，有这位清华帅哥创建的小行星防御，有有机化学顾问这些学科学习类的智能体， 也有实用的电影推荐大师，同学们的想象力真的太丰富了，这种共创 a 的形式我觉得特别好。又来到了通易 home aigc 展区，通易智能三 d 数字人，你看多逼真呢？人物动作视频生成技术、 ai 视频人物替换，在这些技术的加持下，就有了与前世面对面的奇妙体验。你看，此时此刻的我， 化身为穿越距离的男主角，我甚至还能跟他对话，真的太神奇了！晚上还有个星语画未来夜台活动，邀请了多位天文学家，你能想象吗？将 ai 大模型装进望远镜里，就是基于 一千万开元模型打造的星宇三点零新一代天文大模型，只要你输入指令，他就能自动操控望远镜进行观测活动。星宇大模型目前已成功接入国家天文台兴隆观测站望远镜阵列。迷你四天，将来能帮助我们可爱的天文工作者 ai 对话星空，探索星空的奥秘，不虚此行啊！人类利用 ai 究竟能创造出怎样的未来？未来可期！