I've been looking into FSRS since I'm building a language learning app[1], but I haven't implemented it yet. Can FSRS work if I don't want to have 4 choices - bad, good, hard...? I have found myself to get into a decision paralysis so just bad/good works better for me. Plus I can swipe the cards tinder style! :D
My second reason is that I'm worried about the complexity - both from non-nerdy users perspective and me having to debug it.
Yes. Some months ago, I forget whether it was v5 or maybe before, the 2 button grading approach meant that it actually performed better for users (ie. it performed better for users that explicitly only used two grading buttons).
I don't know if this is replicable with v6, or not. I would be interested to find out!
I’m building a spaced-repetition flashcards app. Unlike the competitors, the cards have audio, images and sentences. Additionally, it is possible to add words from ebooks, websites and YouTube.
I've finished my Bc. in computer science before AI, but even then, sitting through a 1.5h long lecture and reading a textbook was just not the way to learn.
a) better quality lectures were available online - it's much easier to learn linear algebra from top MIT Professor than a random one at my university
b) the text books were absolutely terrible compared to what was available online
I can understand that 20 years ago people were captivated with the physical lectures because it was the only way. Today however, professors are competing with 3blue1brown, Khan academy, pre recorded lectures from top universities and many more great resources. Standing in front of a blackboard slowly going through an unintuitive math proof is just not going to cut it and people will get bored.
In my experience in-person lectures have been a terrible way to actually learn (compared to recorded ones or other ways of learning):
1. Lectures are often not held in front of just a handful of students, but hundreds, where frequent interaction and questioning between student and lecturer becomes practically impossible, awkward and socially intimidating. Sitting between many other students is incredibly distracting, and I've more than once seen students bring binoculars to class, because they sat so far away from the blackboard!
2. Only a small fraction of lecturers are actually good at teaching, of being engaged and engaging, clear and concise, understandable and empathetic. Not to mention nice handwriting or powerpoint style. Being "forced" to listen to someone whose style of teaching you don't understand or don't vibe with sucks.
3. If you lose the thread in a deep in-person lecture, you might as well just leave. If you watch a recorded one, you can rewind as often as you want until you understand it.
4. Lecture hall rooms are often not the beautiful, comfortable, nice places you love to go to that they ought to be (but dilapidated, broken, uncomfortable, tight, stuffy and dirty).
In these contexts, sitting in a lecture hall becomes more hell than heaven. And professors shouldn't expect their students to find their passion while having to endure this.
>1. Lectures are often not held in front of just a handful of students, but hundreds, where frequent interaction and questioning between student and lecturer becomes practically impossible, awkward and socially intimidating. Sitting between many other students is incredibly distracting, and I've more than once seen students bring binoculars to class, because they sat so far away from the blackboard!
Spring's annotations arguably are black magic, but while Spring offers DI, DI is not Spring, and honestly something like Guice is a lot easier to follow since it only does annotation-based DI and not a bunch of other stuff.
Search within your IDE, do a Google search, whatever suits you.
What mentality? And the cognitive load is to RTFM, so that you understand what are you doing. If that leaves any questions you can attempt to do a deep dive. It's not particularly high cognitive load to know that @GET is a get rest endpoint.
How is that different without annotations? Documentation is also your best bet at first in case of a normal library function call. Jumping into that codebase can also be quite involved, depending on what it does.
There was a 'may' in my original comment. It is metaprogramming, so you can't see every usage automatically even with "alien tech" like IDEs, unlike in case of a normal type.
Especially that we are not even talking about own code, but third-party annotations with its third-party consumers. Also, grepping is a pretty standard term, it doesn't necessarily mean literal CLI grep, but go on with your advanced tooling as if no one else would be familiar with an IDE.
Seems to me that he thinks he's being fired for mentioning that grok 3 exists but he's actually being fired for saying that chatgpt is better than grok 3.
Then when you fire them, tell them that. But they threatened him using a different reason, a reason that he has support showing that his action was within guidelines.
He’s also clear in his comments that xAI has the right to do whatever they want. He is giving more data points on how xAI/Twitter/Musk operate. This also doesn’t surprise anyone.
We don't actually know what he was told. We only know what he is saying he was told, with no evidence, and in a context where he's playing to the crowd.
He might have misinterpreted what he was told. He might have been told multiple problems, and only reported one. He might be actively lying.
There was this differentiation and discussion in the thread. He claims he has the support. But you are correct, everything I’m relating is his statements on the Twitter thread. For all we know, the entire story could be false.
I've managed to get llms fail on simple questions, that require thinking graphically - 2D or 3D.
An example would be: you have a NxM grid. How many shapes of XYZ shape can you fit on it?
However, thinking of the transformer video games, AI can be trained to have a good representation of 2D/3D worlds. I wonder how it can be combined so that this graphical representation is used to compute text output.
Great app! I've been building something similar, but for less advanced language learners, who wouldn't understand definitions in their target language.
My app [1] is basically a combination of SRS flashcards with an ebook/YouTube/Website reader. Unlike Anki though, AI creates example sentences, definitions, images and audio.
I find it interesting that you want to get inspired by Duolingo. My approach is to have the most efficient grind possible - no gamification. I've found Duolingo was wasting so much of my time with exercises that did not really teach me anything and took a long time to complete + the XP points/levels etc. were quite distracting.
With all the (somewhat competing, though aimed and monetized differently) products in this thread, are there any promotions in place for extensive testing and comparison?
(E.g., your vocabuo website prominently points to possible promo codes.)
This is one of the reasons why I'm building my app in Flutter and not Swift/Android, which is not mentioned a lot in cross-platform vs. native discussions. Neither of the native frameworks compile to JS/we assembly, making it useless if the app gets randomly deleted from stores.
App with dynamic/flexible spaced repetition flashcards for language learning.
Recently I've added dialog & definition cards, so I can learn German from short dialogs with images and audio.