RT Paul Graham

RT Paul Graham It would be interesting if AI destroys Google's search business, and YouTube ends up being the most valuable thing they have left. Stranger things have happened. Justine Moore: I don’t think people understand how many Gen Zers spend HOURS a day on YouTube. Passive, long-form content (music, ASMR, video podcasts) is the background audio to their lives.

相关推荐

封面图片

RT Paul Graham

RT Paul Graham Idea: Before writing an essay, have ChatGPT write one on the same topic to show you what would be the conventional thing to say, so you can avoid saying that.

封面图片

#账户解封模板  #解封

#账户解封模板  #解封 联系 Telegram 官方/客服: * App→设置→帮助与反馈 * 客服页面: * 官方 Twitter: * 登陆问题, 找Twitter: * 官方FAQ: * 发邮件给官方:     [email protected]     [email protected]     [email protected]     [email protected] * Telegram 客户端反馈群:     iOS - @tgiostests     macOS - @macswift     Desktop - @TelegramDesktopTalk 申诉反馈模板1: 主题: Please recover my Telegram account 内容: Dear Telegram Support Team,I am writing to appeal the recent ban on my Telegram account. I understand that my account has been blocked for violating the terms of service, but I would like to explain the circumstances surrounding this situation. I can assure you that my actions were not intentional and I did not have any malicious intent. I understand the importance of following the rules and regulations set by Telegram, and I regret any actions that may have led to my account being banned. I would be grateful if you could consider lifting the ban on my account. I have a strong passion for using Telegram and I have used it for many years to communicate with friends and family. Losing access to my account would have a significant impact on my daily life. Thank you for taking the time to consider my appeal. I look forward to hearing back from you and hopefully having my account reinstated. My mobile number is xxxx 申诉反馈模板2: 内容: My Telegram account was banned a long time, i think Anti-Spam system triggered wrong action, its mistake that account was banned. Now i have again got access, but still I'm limited to anything. It's wrong & false triggered action i should say. I have been a loyal to Telegram & even contributed to better Telegram. I am requesting for removal of account limitations. 提醒: 申诉后,官方会核查你的账户,并根据情况恢复你的账户,但不一定就能解封! 有人申诉后解封了,有人申诉了多次也未解封,不能一概而论!

封面图片

Updated to version 4.5

Updated to version 4.5 If anyone wants a finished build with all the NSFW female characters, here it is. It also has most of the fixed NPCs. Most of the characters can switch between skins. There is no censorship underwater or on land. This is a fully finished build. Just run "3DMigoto Loader" and the game. In the folder "Mods" is a file to update the mods and also, notepad where all hotkeys are specified. This is not my mods, it is a collection of other people's mods. I did not specify the authors of the mods, as I did not think that even I will put this collection. But I express my sincere gratitude for their labor. I'm supposed to do a preview, I know. But given the fact that there are too many mods, I have no idea how to do it, so as not to stretch the video to infinity. If anyone needs an individual character, let me know. _ Link: https://mega.nz/file/DvpxGL4a#YYisAR5SKDCK0HYZwLIuDjjL-QpzOw9Te2a-K048BFo #Aloyhalf #Ayaka #Amber #Barbara #Beidou #Candace #Charlotte #Chevreuse #Chiori #Clorinde #Collei #Dehya #Diona #Dori #Eula #Faruzan #Fischl #Fremine #Furina #Ganyu #HuTao #Jean #Keqing #Kirara #Klee #Kokomi #KujouSara #KukiShinobu #Layla #Lisa #Lynette #Mona #Nahida #Navia #Nilou #Ningguang #Noelle #Paimon #Qiqi #RaidenShogun #Rosaria #Sayu #Shenhe #Sucrose #Xiangling #Xianyun #Xinyan #YaeMiko #Yanfei #Yaoyao #Yelan #Yoimiya #YunJin #NPC #埃洛伊 #安柏 #神里绫华 #芭芭拉 #北斗 #坎迪斯 #千织 #迪希雅 #迪奥娜 #琴 #刻晴 #夏洛蒂 #夏沃蕾 #柯莱 #多莉 #优菈 #珐露珊 #菲谢尔 #芙宁娜 #甘雨 #胡桃 #可莉 #珊瑚宫心海 #九条裟罗 #久岐忍 #来伊拉 #丽莎 #莫娜 #纳西妲 #娜维娅 #妮露 #凝光 #诺艾尔 #派蒙 #七七 #雷电将军 #罗莎莉亚 #早柚 #申鹤 #砂糖 #香菱 #辛焱 #八重神子 #烟绯 #瑶瑶 #夜兰 #宵宫 #云堇 #绮良良 #琳妮特 #菲米尼 #闲云 #克洛琳德

封面图片

(注:神经突触与神经元的动作电位触发时间的先后关系,决定了它们连接强度是增强还是减弱。如果突触的激活时序领先于神经元的动作电位,

(注:神经突触与神经元的动作电位触发时间的先后关系,决定了它们连接强度是增强还是减弱。如果突触的激活时序领先于神经元的动作电位,那么该连接获得强化;如果突触的激活时序滞后于神经元的动作电位,那么该连接获得削弱。It's a particular learning rule that uses Spike timing to figure out how to to determine how to update the synapses. So it's kind of like if the synaptic fires into the neuron before the neuron fires, then it strengthens the synapse. And if the signals fire into the neurons shortly after the neuron fired, then it weakens the synapse.) 神经网络另一个重要的点在于loss函数的提出,它为深度学习提供了可行的训练方法。很有趣的一点是,在现实世界中,我们并没有看到对应loss函数的东西 - 进化论是以loss的方式来迭代的吗?经济系统或社会系统存在loss吗?似乎都不是。 2. 神经网络的本质 Ilya认为,大脑也好,大模型也好,本质上都是把知识压缩到一个高维的隐空间当中。每一个新的观测数据到来的时候,它就会通过连接来更新隐空间中的一些参数。知识就存储在这些连接的权重里。(I guess what is a recurring role that you have a neural network which maintains a high dimensional, hidden state, and then within observation arrives. It updates its high dimensional, hidden state through its connections in some way. You could say the knowledge is stored in the connections.)压缩的过程有点类似于人类的记忆和遗忘过程,你忘掉了绝大部分没用的信息,而只是记住了那些有用的,并且将它们整合记忆。 压缩的过程就是“寻找最小回路”(search for small circuits)的过程。在数学上,有一种理论是“最短描述长度”原则,即如果你能够找到能够产生所需数据的最小程序,那么你就能够用这个程序做出最好的预测。(If you can find the shortest program that outputs the data in your disposal, then you will be able to use it to make the best prediction possible.)这是数学上可以被证明的。但“最短描述长度”原则是一个理论原则,在实践中很难准确实现。所以在实践中,针对给定的数据集,我们只能使用神经网络找到“尽量短小”的回路。因此,可以将神经网络的训练过程理解为,慢慢将训练数据集里的信息熵迁移到神经网络的参数中,而最终沉淀下来的这些回路刚好不算太大。(If you imaine the training process of a neural network as you slowly transmit entropy from the data set to the parameters, then somehow the amount of information in the weights ends up being not very large, which would explain why the general is so well.) 如果你能高效压缩信息,那么你一定已经得到知识了。GPT已经是一个世界模型了,it knows all the intricacies。尽管你做的看似只是predict the next word这么简单的事情,但这只是优化手段而已。 自然语言是最好的latent space,而且是最容易做alignment的latent space。 3. Ilya研究生涯中的两个重要时刻。 第一个时刻,是2012年做AlexNet,Alex Krjevsky用GPU来编写足够快的卷积程序,让CNN训练变得超级快,拉开了CV时代的序幕。这是Ilya的顿悟时刻,觉得神经网络这条路是能走通的。 第二个时刻,Ilya对大模型的信心来自于早年团队的一个发现。当时,团队训练一个LSTM模型来预测Amazon评论中的下一个character,当参数从500到4000的时候,就会出现一个专门的neuron来表达评论的sentiment是正面还是负面。于是,团队猜测,当模型足够大、参数足够多的时候,句法已经被充分表达了(run out of syntax to model),多余的参数开始学会捕捉语义信息。这说明了通过“预测下一个字”的训练方法,可以让模型学到更多隐藏的信息。 4. 关于多模态。 多模态是有用的,尤其是视觉。人类大脑皮层中三分之一都用来处理视觉,缺少了视觉的神经网络作用会相当有限。 人类更多是从图像而不是语言中学习的。人类一生只会听到大概10亿个词,这个数据量是非常有限的,而更多的数据来自于视觉。 很多时候,从视觉学习比从文本学习更容易。例如颜色,尽管通过文字也可以学到颜色之间的关联,比如红色和橙色更近,和蓝色更远,但通过视觉来学习要快得多。 5. AI有逻辑吗?有意识吗? AI当然有逻辑,要不为什么AlphaGo和AlphaZero在最需要逻辑推理能力的围棋游戏中击败了人类? 如何真正说明AI有逻辑推理能力?证明真正困难的定理,写复杂的代码,用创新方法解决开放性问题。如果AI能够证明一个未经证实的定理,那么这个理由就很难辩驳。 如何判断AI是否有意识?做这样一个实验,假如未来人工智能的训练可以从零开始,通过更小的数据集来完成,那么我们可以非常小心地对训练数据进行清洗,确保数据集中不包含任何关于意识的内容,如果系统在训练中需要人类的反馈,在互动中也要非常谨慎,确保不提到任何关于意识的概念。等训练结束的时候,你和AI一起聊天,这时你告诉他关于意识的事情,你向他描述之后,想象一下,如果这个人工智能接着说,”哦,我的上帝,我一直有同样的感觉,但我不知道如何表达它“,这时就可以认为人工智能有意识了。 6. 开源 vs 闭源。 如果模型的能力不强,那么开源是一件伟大的事情。如果模型的能力过强,那么开源就变得危险。尽管目前GPT4模型的能力还算不上”过分强大“,但已经能够看到这个趋势,所以闭源是合理的。(类似于核武器?) 当然,现阶段闭源更重要的原因是商业竞争(而不是安全,Ilya的原话)。 7. 更大的模型一定会带来更好的结果。(Of course the larger neuron nets will be better.) 前些年扩大规模很容易是因为有很多计算资源都没有被充分利用,一旦重新部署过之后就会快速取得进展。但现在规模到达了某种瓶颈,算力的扩张速度变慢了。I expect deploying to continue to make progress in art from other places. The deploying stack is quite deep and I expect that there will be improvements in many layers of the stack and together they will still lead to progress being very robust. 我预期我们将发现deep learning中很多尚未被发现的新属性,而这些新属性的应用将会让模型的效果变得更好。5-10年之后的模型能力一定会远远强过现在的模型。 附三个访谈的链接: 2020年5月 Lex Fridman AI Podcast 2023年3月 黄仁勋 CEO 与 OpenAI 联合创始人及首席科学家 Ilya Sutskever 关于 AI 及 ChatGPT 的对话 2023年4月 OpenAI联合创始人首席科学家AI Ilya Sutskever斯坦福大学内部演讲

🔍 发送关键词来寻找群组、频道或视频。

启动SOSO机器人