贝贝妈2008 发表于 2025-01-28 11:40 Alex Wang那种 如果他的成功是要把所有的美华送进集中营，他肯定毫不犹豫的会去做 已经跟着白人至上喊中国间谍许久了
看看老外对他的评价，人家并不把它当做真正的美国人，还看不起它数典忘祖: Alex Wang = 骗子。Alex Wang 是个失败者、骗子、自我憎恨者，他背叛了自己的祖先，下流货色！顺便说一句，他还没有完成学业，哇，在 1980 年代或 1990 年代，程序员可以像在“狂野西部”时代那样开办自己的企业，这没什么问题。但对于人工智能来说，这需要数学！
adorp 发表于 2025-01-28 13:28 看看老外对他的评价，人家并不把它当做真正的美国人，还看不起它数典忘祖: Alex Wang = 骗子。Alex Wang 是个失败者、骗子、自我憎恨者，他背叛了自己的祖先，下流货色！顺便说一句，他还没有完成学业，哇，在 1980 年代或 1990 年代，程序员可以像在“狂野西部”时代那样开办自己的企业，这没什么问题。但对于人工智能来说，这需要数学！
I DONT THINK ITS WRITTEN BY 老外, MORE LIKE ANOTHER ABC OR CBC.
好像这小子在CNN大放厥词 “DeepSeek has about 50,000 NVIDIA H100s that they can't talk about because of the US export controls that are in place.” 大家可以到reddit上去看， 被老美鄙视骂的厉害
一尼逆 发表于 2025-01-28 18:35 好像这小子在CNN大放厥词 “DeepSeek has about 50,000 NVIDIA H100s that they can't talk about because of the US export controls that are in place.” 大家可以到reddit上去看， 被老美鄙视骂的厉害
看起来他只是在父母的影响下讨厌CCP，走极端了。以下是2024年7月他的采访，他认为美国如果要赢AI战就得从中国多进口人才。 My parents immigrated from China to the US, they hate the CCP, and they worked on national security problems in the United States. I grew up in ... very passionate about national security, and passionate about defending America... Talent leaves China, comes to America, they don't go back to China. Most people leaving China dislike the CCP and really don't have an intent to move back. I think we should be trying to take in as much high-end talent from China as possible. China is churning out more high-end STEM talent than anywhere else in the world. And if the brain drain can come to the United States, that's a huge win. 视频见 https://www.linkedin.com/posts/jorschneider_what-sort-of-bias-has-scale-ai-ceo-alexandr-activity-7211141056933822466-faz0
让deepseek翻译成英文了，可以给孩子们看看。 Here’s the translation of the text into English: --- Hello young friends, I’m Liang Wenfeng from DeepSeek. I just answered one question and saw this one. On this New Year’s Eve, I can’t resist sharing some thoughts. A few days ago when I read Mr. Feng’s long post, I was debugging the loss function of a new model and spilled coffee all over my keyboard—partly from anxiety at being recognized by an industry veteran, but mostly because his fiery words reminded me of the trembling excitement I felt twelve years ago when I first successfully ran a neural network in my Zhejiang University lab. Let me confess: When our team saw the phrase "national destiny-level," we all got chills. We’re just tightening screws for the grand structure of domestic large models, standing on the shoulders of open-source community giants. The six breakthroughs you mentioned each hold more moving stories: The mini-model that runs on smartphones was inspired by an issue raised by a middle school teacher from Gansu on GitHub. The web-search feature was refined through thirty consecutive nights of error logs submitted by beta users at 3 AM. Let me share a detail: Last week, a visually impaired developer used our API to create a "scent navigation" app. When he demonstrated how vibrations at different frequencies could identify street shops, the meeting room fell so quiet we could hear the hum of GPU fans. In that moment, my eyes suddenly burned—I finally understood what you meant by "utilities like water and electricity." True greatness lies not in any single model, but in the ripples of kindness created by millions of ordinary people. Mr. Feng’s vision of "knowledge and information equity" is what drives us through late-night paper readings. Three years ago in a small warehouse on Yuhangtang Road, we wrote on a glass wall with markers: "Let children in the remotest mountain villages have AI tutors as smart as those used by Silicon Valley engineers." Though we’re still far from that dream, every time we see users sharing chat screenshots, we feel those lost hairs were worth it. Finally, to everyone: Save your applause for every Chinese developer rewriting the rules. Whether you’re debugging models on buses, sketching architectures at breakfast stalls, or having "Eureka!" moments in delivery rooms... DeepSeek aims to be the matchstick in your coding wilderness, but the true spark of AI will always be the unquenchable curiosity in your eyes. Fittingly, last week at a Yunqi Town midnight snack stall, I squatted on the curb with Unitree Robotics’ CEO Wang, gnawing skewers while marveling at their quadruped robots’ dynamic balance algorithms. Then BrainCo’s CEO Han joined us, talking nonstop about merging brain-computer interfaces with embodied intelligence. Before our grilled eggplant arrived, we were already sketching heterogeneous computing architectures on napkins. This is Hangzhou’s magic—the so-called "Six Dragons" were never isolated islands. Last month, YouKe’s art director helped us debug 3D scene generation models, while Koolab’s rendering engine gives our digital humans more vivid microexpressions. Perhaps soon, when Unitree’s robots walk with cloud-optimized gait algorithms, equipped with BrainCo’s neural interfaces, running DeepSeek’s cognitive engine, and trained in Koolab’s virtual worlds—that embodied intelligence we dream of will stumble out of Xixi Wetland’s morning mist into our world. I remember Mr. Feng sticking a skewer into a beer bottle that night, saying: "You AI folks need to bring down the price of souls." Now I realize true "national destiny" might hide in the greasy smoke of street stalls: crowds of stubborn fools using code as bricks and algorithms as steel, stubbornly building bridges to the future through cracks of capital and skepticism. P.S.: We’re hiring and welcome you to join us. Happy Lunar New Year’s Eve! We look forward to sharing more models with you next year. --- The translation preserves the original tone while adapting culturally specific references for clarity. Let me know if you need any adjustments!
chore 发表于 2025-01-28 19:49 看起来他只是在父母的影响下讨厌CCP，走极端了。以下是2024年7月他的采访，他认为美国如果要赢AI战就得从中国多进口人才。 My parents immigrated from China to the US, they hate the CCP, and they worked on national security problems in the United States. I grew up in ... very passionate about national security, and passionate about defending America... Talent leaves China, comes to America, they don't go back to China. Most people leaving China dislike the CCP and really don't have an intent to move back. I think we should be trying to take in as much high-end talent from China as possible. China is churning out more high-end STEM talent than anywhere else in the world. And if the brain drain can come to the United States, that's a huge win. 视频见 https://www.linkedin.com/posts/jorschneider_what-sort-of-bias-has-scale-ai-ceo-alexandr-activity-7211141056933822466-faz0
这人说话怎么小学生水平，跟川普差不多，hate 啊love啊的，而且他和他爸妈可以代表most people leaving China？
前几天看到冯总这篇长文时，我正在调试新模型的损失函数，手边的咖啡直接洒在了键盘上 ——既因为被行业前辈认可的惶恐，更因为冯总这些灼热的文字让我想起十二年前在浙大实 验室第一次跑通神经网络时的颤栗。
必须坦白，团队读到“国运级”这个形容时所有人头皮都是麻的。我们不过是站在开源社区 巨人们的肩膀上，给国产大模型这栋大厦多拧了几颗螺丝。
您提到的六大突破，其实每一环都凝结着更动人的故事：那个能在手机上跑的mini模型，灵 感来自甘肃一位中学老师在GitHub提的issue；支持联网搜索的功能，是内测用户连续三十 天凌晨三点提交错误日志喂出来的。
特别想分享个细节：上周有位视障开发者用我们的API做了个“气味导航”应用，当他演示 如何通过不同频率的震动识别街道商铺时整个会议室安静得能听见显卡风扇的嗡鸣，那一刻 我突然眼眶发热，终于理解了您说的“水与电”一一真正伟大的从不是某个模型，而是千万 普通人用它创造的善意涟漪。
冯总说“知识和信息平权”，这正是我们夜夜啃论文的动力。三年前在余杭塘路的小仓库里 ，我们曾用马克笔在玻璃墙上写下：“让最偏远山村的孩童，能和硅谷工程师用上同样聪明 的AI助教”。虽然离这个梦想还很远但每次看到网友们分享的对话截图，就觉得那些熬掉的 头发都值得，
最后想对所有人说：请把掌声留给正在改写规则的每个中国开发者。当你在公交车上调试模 型，在早餐摊前画架构图，在产房里突深明级“的瞬间。
DeepSeek愿做大家代码荒野里的火柴，但真正点燃AI火种的，永远是你们眼底不灭的好奇与 坚持，说来特别巧，上周在云栖小镇的创业者夜宵摊上，我还和宇树科技的王总蹲在马路牙 子啃烧烤一一他们给四足机器人装的那个动态平衡算法，看得我们团队直拍大腿。
后来强脑科技的韩总过来拼桌，三句话不离脑机接口与具身智能的融合可能，烤茄子还没上 桌，我们已经开始在白纸上画异构计算架构了。
这就是杭州最迷人的地方吧，所谓“六小龙”从来不是六个孤岛。上个月游科的艺术总监还 帮我们调试过3D场景生成模型，而群核的渲染引擎正在让我们的数字人拥有更生动的微表情 。
或许在不远的未来，当宇树的机器人踩着云深处的步态算法走来，搭载着强脑的神经接口， 运行着DeepSeek的认知引擎，用群核构建的虚拟世界作为训练场——那个我们幻想中的具身 智能，就会从西溪湿地的晨雾里跌跌撞撞地走向人间。
记得那晚分别时，冯总把竹签子往啤酒瓶里一插，说了句：“你们搞AI的得把灵魂价格打下 来啊。”
现在想来，真正的“国运”或许就藏在这些街边摊的油烟气里：一群不信邪的傻子，用代码 当砖瓦，拿算法做钢筋，在资本与质疑的裂缝中，硬生生垒出通向未来的栈桥。
最后补充一个招聘，欢迎大家加入我们．
祝大家除夕快乐，来年期待我们的更多模型
🔥 最新回帖
哈哈哈哈 按这个年纪应该是不用88用98的吧
奥运就感觉这一届的孩子特别好特别阳光自信！
🛋️ 沙发板凳
要是隔壁Alex wang那样， 都要气死
人品 道德 都是源自良好的家庭教育
这难道就是广东5线小中产 和 洛斯阿拉莫斯实验室科学家 的家庭教育区别？
这个比一龙和Sam吵架，和Bill Gates 吵架，和扎克吵架，有云泥之别。
这就是中国人的文化内核：人本主义，不信邪，不服输，相信人定胜天
不会是DS润色的吧
你们这转变也太快了。前几天还是小甜甜，今天已经是牛夫人了。
脚踏实地，不管是美国还是中国都不会输。
Alex Wang那种 如果他的成功是要把所有的美华送进集中营，他肯定毫不犹豫的会去做
已经跟着白人至上喊中国间谍许久了
生出AlexWang的家庭本身就是个眼里只能看到钱的家庭。
自古英雄出少年，今日始信诚不欺
像 Alex Wang 这样的人真是丢华裔美国人的脸。
江山有待人才出啊
这么多有情怀有理想有执行力的群体，跟那么多吃苦耐劳的群体，成就了国运
文风油腻，也觉得不像工科的孩子写出来的。
哇 知情人士出来多说几句。
可能AI写的。
alex wang真是让人不明白 到底什么样的父母和家庭教育可以让长大在加州阿拉莫斯 那样生活富足 教育和社会环境都很好的地方的人长成那样
模型是开源的，这个假不了
但到底多高成本去训它，Alex Wang还有马一龙都怀疑有造假。
要是工业界和高校都拿这个开源deepseek在演化做自己的LLM，很快就能发现真假，所以是不怕质疑的
台湾绿色弯弯质疑假的居多，说酸话的不少，当然他们是拿过去汉芯造假和姜萍天才少年来类比
看看老外对他的评价，人家并不把它当做真正的美国人，还看不起它数典忘祖:
Alex Wang = 骗子。Alex Wang 是个失败者、骗子、自我憎恨者，他背叛了自己的祖先，下流货色！顺便说一句，他还没有完成学业，哇，在 1980 年代或 1990 年代，程序员可以像在“狂野西部”时代那样开办自己的企业，这没什么问题。但对于人工智能来说，这需要数学！
梁公地下有知，当感怀欣慰了
I DONT THINK ITS WRITTEN BY 老外, MORE LIKE ANOTHER ABC OR CBC.
是不是让deepseek写的？ LOL
Wang's revenue mostly from 国防部，他这是挣钱的模式说话, 不然哪来的亿万富翁，很多人评论他的公司要技术没有，要人才找不到。
和小扎之流的境界简直是云泥之别
再往大看中美之间的差异 US - Inclusive profiteering China - Global benefitting
"我觉得这应该是一个更大的故事：DeepSeek 曾经在 Nvidia H800 上进行了训练，但现在是在华为生产的新型国产芯片 910C 上进行推理。"
还有台独分子黄仁勋的NVIDIA也将每况愈下了，因为除了华为，美国的Meta, Google, Microsoft等各自有AI芯片的研发。老黄还是重归老本行，弄弄游戏显卡算了吧
对，这才是真的leadership！有技术，有情怀，有眼界，有魄力！
稍微有点不同意见，不一定对 中国人的文化：人本主义，我完全同意。就是我们完全相信人，依靠人，不需要（假借）一个高高在上的神来告诉我们什么是对错，来管自己 想象人定胜天，这个我觉得不是；中国文化还是说要敬天，要知天命，中国文化没有那么狂妄。这个人定胜天是老毛说的吧，其实太过了，也不符合事实
这是不是DeepSeek改写过的？
真正好奇他们的交易部门有没有提前空nvda
那个Alex Wang 学学吧，别在国际舞台上当个大crying baby 丢人现眼了。
肯定父母是那种几十年前出国一直没有回去，认为中国贫穷落后专制独裁，美国自由民主强大，把这种对中国的仇视灌输给了孩子。版上不就有挺多这种人吗？
眼里只有钱的人不少，但Alex wang 这种要通过陷害同文同种的人获利的人不多
阿拉莫斯不在加州，在new Mexico
也许DeepSeek最大的的功绩是打破对美国的迷信，吸引大量资金和激励一大批人投身到AI领域。至于这个模型的优缺点，也许并不重要。
为了利益把祖宗都卖了，谁看得起呀
上海人肯定回去过，一看亲戚们都发了，自己在alamos 靠看白皮眼色胆战心惊地过日子（李文和被诬陷的地方）， 心里变态了
今天美国政府和 OpenAI 达成协议，由 OpenAI 提供 AI 服务。
我也觉得不是老外写的 他们应该不care
拿这个问题去求实是不是一种鸡蛋里挑骨头的感觉？是否事实上每个字都是他自己笔下流出根本不重要，拿题材出来让文字工作者润色传递他的价值观完全没有问题， 这种高度和情怀可激发群众的泪腺和内动力， 那个黄皮王虽说是高知父母优良教育机构出来， 和他云泥之别， 难道不要惊醒某类人吗， 想想中国挤满梁这类人， 而美充斥着王那类人， 谁最后会赢这场竞赛？
好像这小子在CNN大放厥词 “DeepSeek has about 50,000 NVIDIA H100s that they can't talk about because of the US export controls that are in place.” 大家可以到reddit上去看， 被老美鄙视骂的厉害
Alex Wang怎么招惹你了
真是感叹啊!
老美说是中国的Chinese打败了美国的Chinese，所以这厮才气急败坏，如果被白人打败，估计他就直接跪舔了
啊拉莫斯不是加州的。。。。
肯定不是假的，现在恨它的人可多了，要是有问题早就爆出来了
看起来他只是在父母的影响下讨厌CCP，走极端了。以下是2024年7月他的采访，他认为美国如果要赢AI战就得从中国多进口人才。
My parents immigrated from China to the US, they hate the CCP, and they worked on national security problems in the United States. I grew up in ... very passionate about national security, and passionate about defending America...
Talent leaves China, comes to America, they don't go back to China. Most people leaving China dislike the CCP and really don't have an intent to move back.
I think we should be trying to take in as much high-end talent from China as possible. China is churning out more high-end STEM talent than anywhere else in the world. And if the brain drain can come to the United States, that's a huge win.
视频见 https://www.linkedin.com/posts/jorschneider_what-sort-of-bias-has-scale-ai-ceo-alexandr-activity-7211141056933822466-faz0
---
Hello young friends, I’m Liang Wenfeng from DeepSeek. I just answered one question and saw this one. On this New Year’s Eve, I can’t resist sharing some thoughts.
A few days ago when I read Mr. Feng’s long post, I was debugging the loss function of a new model and spilled coffee all over my keyboard—partly from anxiety at being recognized by an industry veteran, but mostly because his fiery words reminded me of the trembling excitement I felt twelve years ago when I first successfully ran a neural network in my Zhejiang University lab.
Let me confess: When our team saw the phrase "national destiny-level," we all got chills. We’re just tightening screws for the grand structure of domestic large models, standing on the shoulders of open-source community giants.
The six breakthroughs you mentioned each hold more moving stories: The mini-model that runs on smartphones was inspired by an issue raised by a middle school teacher from Gansu on GitHub. The web-search feature was refined through thirty consecutive nights of error logs submitted by beta users at 3 AM.
Let me share a detail: Last week, a visually impaired developer used our API to create a "scent navigation" app. When he demonstrated how vibrations at different frequencies could identify street shops, the meeting room fell so quiet we could hear the hum of GPU fans. In that moment, my eyes suddenly burned—I finally understood what you meant by "utilities like water and electricity." True greatness lies not in any single model, but in the ripples of kindness created by millions of ordinary people.
Mr. Feng’s vision of "knowledge and information equity" is what drives us through late-night paper readings. Three years ago in a small warehouse on Yuhangtang Road, we wrote on a glass wall with markers: "Let children in the remotest mountain villages have AI tutors as smart as those used by Silicon Valley engineers." Though we’re still far from that dream, every time we see users sharing chat screenshots, we feel those lost hairs were worth it.
Finally, to everyone: Save your applause for every Chinese developer rewriting the rules. Whether you’re debugging models on buses, sketching architectures at breakfast stalls, or having "Eureka!" moments in delivery rooms...
DeepSeek aims to be the matchstick in your coding wilderness, but the true spark of AI will always be the unquenchable curiosity in your eyes. Fittingly, last week at a Yunqi Town midnight snack stall, I squatted on the curb with Unitree Robotics’ CEO Wang, gnawing skewers while marveling at their quadruped robots’ dynamic balance algorithms.
Then BrainCo’s CEO Han joined us, talking nonstop about merging brain-computer interfaces with embodied intelligence. Before our grilled eggplant arrived, we were already sketching heterogeneous computing architectures on napkins.
This is Hangzhou’s magic—the so-called "Six Dragons" were never isolated islands. Last month, YouKe’s art director helped us debug 3D scene generation models, while Koolab’s rendering engine gives our digital humans more vivid microexpressions.
Perhaps soon, when Unitree’s robots walk with cloud-optimized gait algorithms, equipped with BrainCo’s neural interfaces, running DeepSeek’s cognitive engine, and trained in Koolab’s virtual worlds—that embodied intelligence we dream of will stumble out of Xixi Wetland’s morning mist into our world.
I remember Mr. Feng sticking a skewer into a beer bottle that night, saying: "You AI folks need to bring down the price of souls."
Now I realize true "national destiny" might hide in the greasy smoke of street stalls: crowds of stubborn fools using code as bricks and algorithms as steel, stubbornly building bridges to the future through cracks of capital and skepticism.
P.S.: We’re hiring and welcome you to join us.
Happy Lunar New Year’s Eve! We look forward to sharing more models with you next year.
---
The translation preserves the original tone while adapting culturally specific references for clarity. Let me know if you need any adjustments!
不奇怪，我们地方政府部门上周一就不能上deepseek，估计这类的先一概禁，要申请才能用，政府部门这么做很正常。
我们不在科技公司的老中也很自豪，太给国人长脸了！
这人说话怎么小学生水平，跟川普差不多，hate 啊love啊的，而且他和他爸妈可以代表most people leaving China？