Jarrett Ye
@JarrettYe
Followers
6K
Following
15K
Media
439
Statuses
2K
Die-hard fan of spaced repetition. Creator of FSRS. Influencer at ZhiHu with 100k+ followers. Research Engineer at MaiMemo Inc. Founder of Thoughts Memo.
China
Joined March 2022
In case you didn’t know, I received my computer science bachelor’s degree from the Harbin Institute of Technology (Shenzhen) three years ago. However, the vast majority of the knowledge I’ve found useful to this day was self-taught. https://t.co/13E74TLHsp
lesswrong.com
Editor’s Note In case you didn’t know, I received my computer science bachelor’s degree from the Harbin Institute of Technology (Shenzhen) three year…
4
16
244
Bro… this setup is insane, tomorrow will be too late. Do your research, go on NASDAQ, add NXXT, go to institutional investors..THEY ALL LOADING UP on NXXT right before earnings $NXXT sittin’ around $1.80 and they just dropped numbers like a mid-cap. Revenue up +196% YoY,
20
33
220
In today’s society, self-improvement and social recognition are separate. Everything is built on a certain foundation, and when this foundation is not met, no matter how much you improve yourself, you can’t even reach the threshold, and all doors will be closed to you. “I can
lesswrong.com
Editor’s Note Ever since I began translating articles on free learning[1] and critiques of schooling, [2]a difficult question has loomed over me like…
1
0
23
I did not go to Hengshui High School. But nine years ago, when I was a high school student, our principal visited the Hengshui school and decided to use its methods. We were forced to go jogging at 6:45 every morning. This stopped only after a student died on the way to the
lesswrong.com
Editor’s Note I did not go to Hengshui High School. But nine years ago, when I was a high school student, our principal visited the Hengshui school a…
1
1
40
Here's a screenshot of a new feature we're preparing to roll out to some beta testers called Gravity. With it, you'll be able to prioritize one or more topics (as well as any missing prerequisites along their respective learning paths) by "adding gravity" to them. There's a lot
9
3
79
@biedalak I confirm. I see it with my own eyes on a daily basis, but I propose to postpone the debate till the release of the SM20 API in March. I understand that it will then be possible to make a reasonable comparison, and keep all metrics open source :) 💝
0
1
4
After submitting the Vulnerability report to SuperMemo team, we build an ADVERSARIAL algorithm to exploit the vulnerability and it works well.
1
0
8
影响最佳回顾节点的因素有很多 - 记忆内容本身的复杂度 - 时间间隔(比如,30天) - 到期记忆保持期望(比如,回顾出70%) 这三个变量确认后,才能计算出「针对一个确定内容以最低学习成本在一定时间后获得最佳期望」的最佳回顾节点 至于说艾宾浩斯记忆曲线,这是一个被以讹传讹的实验记录罢了
99% 的人,不是没学会,而是“学了就忘”。 后来我用上了“艾宾浩斯记忆法”,明白原来记忆是有“规律”的。 📘简单说,这个方法只有一个核心: 记忆不是靠重复次数,而是靠“重复的时间点”。 举个例子,如果你今天学了50个英语单词: 第1天复习(当晚) 第2天再看一次 第4天再巩固 第7天、14天、30天……
2
8
41
We find the vulnerability of the universal metric proposed by SuperMemo. It's still hackable and we could create an adversarial model to win in the algorithm battle. The details have been submitted to SuperMemopedia.
2
0
10
I add a fancy visualization to the comparison among FSRS families and SuperMemo families!
1
0
6