Transformer 是自然语言处理领域中的一种重要模型架构。以下是一些与 Transformer 相关的内容:
[title]皇子:LLM经典论文速读版,看完感觉自己通透了[heading2]1.Transformer[heading2]2.GPT-1[heading2]3.BERT[heading2]4.Transformer-XL[heading2]5.GPT-2[heading2]6.ERNIE[heading2]7.DistilBERT[heading2]8.T5[heading2]9.Retrieval-Augmented Generation(RAG)[heading2]10.GPT-3[heading2]11.AutoPrompt[heading2]12.Rotary Position Embedding[heading2]13.LoRA[heading2]14.Codex[heading2]15.FLAN[heading2]16.GLaM[heading2]17.WebGPT[heading2]18.Chain-of-Thought(CoT)[heading2]19.PaLM[heading2]20.InstructGPT[heading2]21.Verify Step by Step[heading2]22.LLM.int8()[heading2]23.ReAct[heading2]24.Toolformer[heading2]25.LLaMA[heading2]26.GPT-4[heading2]27.QLoRA[heading2]28.Phi-1[heading2]29.RLAIF[heading2]30.Superalignment
从核心本质上看,Sora是一个具有灵活采样维度的扩散变压器[4],如图4所示。它有三个部分:(1)时空压缩器首先将原始视频映射到潜在空间。(2)ViT然后处理标记化的潜在表示,并输出去噪的潜在表示。(3)类似CLIP[26]的条件机制接收LLM增强的用户指令和可能的视觉提示,以指导扩散模型生成风格化或主题化的视频。经过多次去噪图4:逆向工程:Sora框架概览在这一步骤中,生成视频的潜在表示被获得,然后通过相应的解码器映射回像素空间。在本节中,我们的目标是对Sora使用的技术进行逆向工程,并讨论广泛的相关工作。
Letter from Alan Turing to W Ross Ashby-Alan Mathison TuringSoftware 2.0-Andrej KarpathyThe Rise of Software 2.0-Ahmad MustaphaInfrastructure 3.0:Building blocks for the AI revolution-Lenny Pruss,Amplify PartnersWill Transformers Take Over Artificial Intelligence?-Stephen OrnesAI Revolution-Transformers and Large Language Models(LLMs)-Elad GilWhat Is a Transformer Model?-RICK MERRITT[AI时代的巫师与咒语](https://mp.weixin.qq.com/s?__biz=MzI3NDQzNTk2Mw==&mid=2247484347&idx=1&sn=ce81d3c4532660d3bbbf675f71246f48&scene=21#wechat_redirect)-Rokey ZhangGenerative AI:A Creative New World-SONYA HUANG,PAT GRADY AND GPT-3What Real-World AI From Tesla Could Mean-CleanTechNicaA Look at Tesla's Occupancy Networks-Think AutonomousBy Exploring Virtual Worlds,AI Learns in New Ways-Allison WhittenSelf-Taught AI Shows Similarities to How the Brain Works-Anil AnanthaswamyHow Transformers Seem to Mimic Parts of the Brain-Stephen OrnesAttention Is All You Need-PAPER by Ashish Vaswani,Noam Shazeer,Niki Parmar,Jakob Uszkoreit,Llion Jones,Aidan N.Gomez,Lukasz Kaiser,Illia PolosukhinOn the Opportunities and Risks of Foundation Models-PAPER by CRFM&HAI of Stanford UniversityMaking Things Think-BOOK by Giuliano GiacagliaA Thousand Brains(中文版:千脑智能)-BOOK by Jeff Hawkins