[dead]
Scaffolding to Superhuman: How Curriculum Learning Solved 2048 and Tetris
11–20 of 35 posts
Re: Scaffolding to Superhuman: How Curriculum Learning Solved 2048 and Tetris
#12> To learn, agents must experience high-value states, which are hard (or impossible) for untrained agents to reach. The endgame-only envs were the final piece to crack 65k. The endgame requires tens of thousands of correct moves where a single mistake ends the game, but to practice, agents must first get there. This seems really similar to the motivations around masked language modeling. By providing increasingly-mas…
This is less about masked modelling and more about reverse-curriculum.
e.g. DeepCubeA 2019 (!) paper to solve Rubik cube.
Start with solved state and teach the network successively harder states. This is so "obvious" and "unhelpful in real domains" that perhaps they havent heard of this paper.
Re: Scaffolding to Superhuman: How Curriculum Learning Solved 2048 and Tetris
#13Great, add "curriculum" to the list of words that will spark my interest in human learning, only for it to be about garbage AI. I want HN with a hard rule against AI posts.
Re: Scaffolding to Superhuman: How Curriculum Learning Solved 2048 and Tetris
#14Great, add "curriculum" to the list of words that will spark my interest in human learning, only for it to be about garbage AI. I want HN with a hard rule against AI posts.
Why garbage ai? I thought it was a very interesting post, personally.
Re: Scaffolding to Superhuman: How Curriculum Learning Solved 2048 and Tetris
#15Great, add "curriculum" to the list of words that will spark my interest in human learning, only for it to be about garbage AI. I want HN with a hard rule against AI posts.
> HN with a hard rule against AI posts.
Greasemonkey / Tampermonkey / User Scripts with
Array.from( document.querySelectorAll(".submission>.title") ).filter( e => e.innerText.includes("AI") ).map( e => e.parentElement.style.opacity = .1)
Edit: WTH... how am I getting downvoted for suggesting an actual optional solution? Please clarify.
Re: Scaffolding to Superhuman: How Curriculum Learning Solved 2048 and Tetris
#16Great, add "curriculum" to the list of words that will spark my interest in human learning, only for it to be about garbage AI. I want HN with a hard rule against AI posts.
> HN with a hard rule against AI posts. Greasemonkey / Tampermonkey / User Scripts with Array.from( document.querySelectorAll(".submission>.title") ).filter( e => e.innerText.includes("AI") ).map( e => e.parentElement.style.opacity = .1) Edit: WTH... how am I getting downvoted for suggesting an actual optional solution? Please clarify.
Notably this doesn't match the current thread.
Re: Scaffolding to Superhuman: How Curriculum Learning Solved 2048 and Tetris
#17Great, add "curriculum" to the list of words that will spark my interest in human learning, only for it to be about garbage AI. I want HN with a hard rule against AI posts.
Are we really dismissing the entire field of AI just because LLMs are overhyped?
Re: Scaffolding to Superhuman: How Curriculum Learning Solved 2048 and Tetris
#18Earlier quoted context omitted.
> HN with a hard rule against AI posts. Greasemonkey / Tampermonkey / User Scripts with Array.from( document.querySelectorAll(".submission>.title") ).filter( e => e.innerText.includes("AI") ).map( e => e.parentElement.style.opacity = .1) Edit: WTH... how am I getting downvoted for suggesting an actual optional solution? Please clarify.
Notably this doesn't match the current thread.
Expand e.innerText.includes("AI") with an array of whatever terms you prefer.
Re: Scaffolding to Superhuman: How Curriculum Learning Solved 2048 and Tetris
#19I've always found curriculum learning incredibly hard to tune and calibrate reliably (even more so than many other RL approaches!).
Reward scales and horizon lengths may vary across tasks with different difficulty, effectively exploring policy space (keeping multimodal strategy distributions for exploration before overfitting on small problems), and catastrophic forgetting when mixing curriculum levels or when introducing them too late.
Does any reader/or the author have good heuristics for these? Or is it still so problem dependent that hyper parameter search for finding something that works in spite of these challenges is still the go to?
Re: Scaffolding to Superhuman: How Curriculum Learning Solved 2048 and Tetris
#20Earlier quoted context omitted.
> HN with a hard rule against AI posts. Greasemonkey / Tampermonkey / User Scripts with Array.from( document.querySelectorAll(".submission>.title") ).filter( e => e.innerText.includes("AI") ).map( e => e.parentElement.style.opacity = .1) Edit: WTH... how am I getting downvoted for suggesting an actual optional solution? Please clarify.
Notably this doesn't match the current thread.
Could always run the posts through a LLM to decide which are about AI :-p