[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/sci/ - Science & Math

Search:


View post   

>> No.11510520 [View]
File: 68 KB, 588x900, say_who_you_are.jpg [View same] [iqdb] [saucenao] [google]
11510520

>>11510422
436343733263745843532213564343 * 34347552343246574632355457685325 + 17^3 = 14987339217924131880581509431668276350703333802430930734371388

I manually verified it and I'm sure nobody wrote it down before me.

>> No.11509346 [View]
File: 68 KB, 588x900, say_who_you_are.jpg [View same] [iqdb] [saucenao] [google]
11509346

I think I always knew it was a machine. The Profs will tell you early on about how heartless and comptetitive the publish and perish culture is.

I don't agree that the output is irrelevant, even if lots of papers may be for any one person. I often google and find out that somebody studied in detail some nieche linear algebra topic and did calculations and put things into frameworks - those are certainly not notable problems, but when you searcha round and find somebody published the work on it, it's great.

>>11509059
If anything, the lockdown shows me I'm not productive when just sitting alone.

>>11509141
The practical problems in machine learning is bringing the hardware together to run the software. Software improvements are naturally done in incremental theoretical steps, because the field is old enough for the progress being hardware related. There's no fun - at least if you're theoretically minded - to cut things off here and there to make things work with the hardware you have or with the power capactity you have.
I rather read papers about the theoretical application of path integrals to relearning in weight-space than try out this and that tool and write nasty emails with the chip producers etc.

>>11509156
"actual value" is a meme under capitalism. What's valuable? Making people live to 100 instead of 85 via fighting cancer? Make machines faster so that the entertainment products can be engineered in higher quantity? There's no progress, if progress means maximizing profit, if profit means competitive redistribution of the exchange value surrogate. The "value" progress argument is made under the premise that technological change is an attained advantage per se. It's to say a child drawing an image is useless because it doesn't change the digits stored on the bank server associated with your name. The latter is part of a usefull ethics, so enhancing some metric within that is also not useful. Not as useful as the fun some guy has when doing math.

Navigation
View posts[+24][+48][+96]