The first principle is that you must not fool yourself, and you are the easiest person to fool. - Richard Feynman

第一个原则就是不要欺骗自己，你是最容易被欺骗的人. - 理查德.费曼

准备重要的事情, 用清单列表的方式让自己做事井井有条. 迷茫的时候, 问自己问题可以找到方向. 避免陷入想当然的思维误区.

生活就像在大海上开船, 需要时不时打一下方向盘. 让自己在正确的轨迹上航行.

在迷茫的时刻, 我希望我能够回过头来看这份问题清单:

首先是新时代的"吾日三省吾身" (汉明问题(Hamming Questions)摘取自这里):

其次是对汉明问题的拓展:

如果你现在做的事不符合你的追求的话:

如果你现在做的事符合你的追求的话:

如果你觉得你做不了某件事:

二〇二四年六月廿九日

John Carmack让Ilya Sutskever推荐关于了解当下人工智能的阅读资料. Ilya Sutskever给了大约有40篇研究论文的链接.

If you really learn all of these, you’ll know 90% of what matters today. - Ilya Sutskever

如果你真的学会了这些，你就掌握了当今90%重要的内容 - Ilya Sutskever

大概在一个月以前, 推特上流传一份文件夹包含30篇资料(论文, 书籍, 博客)包含了这些资料.

阅读列表

按照发布的时间线从旧到新排列如下:

Time	Paper
1993	Keeping Neural Networks Simple by Minimizing the Description Length of the Weights
2004	A Tutorial Introduction to the Minimum Description Length Principle
2008	Machine Super Intelligence
2011	The First Law of Complexodynamics
2012	ImageNet Classification with Deep Convolutional Neural Networks
2014	Quantifying the Rise and Fall of Complexity in Closed Systems: The Coffee Automaton
2014	Neural Turing Machines
2015	The Unreasonable Effectiveness of Recurrent Neural Networks
2015	Understanding LSTM Networks
2015	Recurrent Neural Network Regularization
2015	Deep Residual Learning for Image Recognition
2015	Neural Machine Translation by Jointly Learning to Align and Translate
2015	Deep Speech 2: End-to-End Speech Recognition in English and Mandarin
2016	Order Matters: Sequence to Sequence for Sets
2016	Multi-Scale Context Aggregation by Dilated Convolutions
2016	Identity Mappings in Deep Residual Networks
2017	Pointer Networks
2017	Neural Message Passing for Quantum Chemistry
2017	Attention Is All You Need
2017	A simple neural network module for relational reasoning
2017	Variational Lossy Autoencoder
2017	Kolmogorov Complexity and Algorithmic Randomness
2018	Relational Recurrent Neural Networks
2019	GPipe: Easy Scaling with Micro-Batch Pipeline Parallelism
2020	Scaling Laws for Neural Language Models
2022	The Annotated Transformer
2024	CS231n: Convolutional Neural Networks for Visual Recognition

计划

写30篇文章深度解析Ilya推荐的每一份资料, 并深度实践每一篇资料的概念.

虽然说是从零, 但是不会包括编程的教学.

看完这30篇文章并实践之后, 你能够: