that's right bsky.app/profile/iid....
It's because the construction poses "egg" as a mass noun, which egg is typically not in English, so it sounds strange to the ear and also suggests that the concept of egg has a Platonic quality that clashes with the mundanity of actual eggs
it hadnt occurred to me why "egg" is funnier/better but this is probably why. replaces "[some specific set of] eggs" with "[the conceptual quality of/existence of] egg"
March 22, 2025 at 5:34 PM
that's right bsky.app/profile/iid....
February 19, 2025 at 12:20 AM
December 13, 2024 at 1:22 AM
cursed dataset: will train if it's not IID with respect to one characteristic not found in the data itself. will not train if it is IID. effect goes away at low batch sizes and gets worse at very high batch sizes.
December 8, 2025 at 6:51 AM
cursed dataset: will train if it's not IID with respect to one characteristic not found in the data itself. will not train if it is IID. effect goes away at low batch sizes and gets worse at very high batch sizes.
tdiz親子・子に大人気iidくん
お礼言う相手がよくわかってない子も可愛くて好き
お礼言う相手がよくわかってない子も可愛くて好き
March 4, 2025 at 2:31 PM
tdiz親子・子に大人気iidくん
お礼言う相手がよくわかってない子も可愛くて好き
お礼言う相手がよくわかってない子も可愛くて好き
like, i am not currently privy to frontier lab discussions on this stuff the way i was last year, but i think that basic issues like "mesa-optimization falls directly out of the math of sequential training on non-IID data with weight-pinning regularization" needs attention
bsky.app/profile/theo...
bsky.app/profile/theo...
when a model is trained, it is trained on data X, subject to mathematical conditions {Y, Z, [...]}. trivially, a model trained on a particular corpus and then fine-tuned on subsequent data subject to a KL-div penalty from the reference distribution can have a basin of attraction predicted by neither
September 25, 2026 at 2:43 PM
like, i am not currently privy to frontier lab discussions on this stuff the way i was last year, but i think that basic issues like "mesa-optimization falls directly out of the math of sequential training on non-IID data with weight-pinning regularization" needs attention
bsky.app/profile/theo...
bsky.app/profile/theo...
I made my first contribution to RenderDoc! It's tiny, but it's something. :)
github.com/baldurk/rend...
github.com/baldurk/rend...
Add missing IID checks for creating ID3D12CommandQueue1 by TheRealMJP · Pull Request #3759 · baldurk/renderdoc
Description
Calls to ID3D12Device::CreateCommandQueue and ID3D12Device9::CreateCommandQueue1 were failing with E_NOINTERFACE when passing the IID of ID3D12CommandQueue1. The 1 version of the interf...
github.com
January 13, 2026 at 5:40 AM
I made my first contribution to RenderDoc! It's tiny, but it's something. :)
github.com/baldurk/rend...
github.com/baldurk/rend...
#rstats #econsky PSA: The next release of `fixest` will include some important changes (plus cool new features).
E.g. Fixed-effects regs will now default to 'iid' SEs rather than clustered. github.com/lrberge/fixe...
You can install and test drive the dev version from R-universe; see the README.
E.g. Fixed-effects regs will now default to 'iid' SEs rather than clustered. github.com/lrberge/fixe...
You can install and test drive the dev version from R-universe; see the README.
July 18, 2025 at 3:36 PM
#rstats #econsky PSA: The next release of `fixest` will include some important changes (plus cool new features).
E.g. Fixed-effects regs will now default to 'iid' SEs rather than clustered. github.com/lrberge/fixe...
You can install and test drive the dev version from R-universe; see the README.
E.g. Fixed-effects regs will now default to 'iid' SEs rather than clustered. github.com/lrberge/fixe...
You can install and test drive the dev version from R-universe; see the README.
People have been debating this for like a week and it’s literally just gender. it’s assumed to be something urban, single women do with other urban single women. It’s extremely simple
I think the Very Online Leftist hatred of brunch is really about brunch being basic; in the same way that they want to insist that you are not protesting right. They're a particularly petty branch of hipster, basically.
April 8, 2025 at 11:39 PM
Anyway long commuting thread but the Take is this: Bluesky is a platform built by decentralists who have acquired a user base of extreme centralists and it’s funny
December 12, 2024 at 1:25 PM
#EcksteinsMonthOfMusic
#MusicChallenge
#MusicSky
Magic
The Charlatans ~ Good Witch, Bad Witch
youtu.be/IID-WZ1kaT8?...
#MusicChallenge
#MusicSky
Magic
The Charlatans ~ Good Witch, Bad Witch
youtu.be/IID-WZ1kaT8?...
Good Witch, Bad Witch 2
YouTube video by The Charlatans - Topic
youtu.be
September 20, 2026 at 6:39 AM
#EcksteinsMonthOfMusic
#MusicChallenge
#MusicSky
Magic
The Charlatans ~ Good Witch, Bad Witch
youtu.be/IID-WZ1kaT8?...
#MusicChallenge
#MusicSky
Magic
The Charlatans ~ Good Witch, Bad Witch
youtu.be/IID-WZ1kaT8?...
April 8, 2024 at 11:06 PM
yeah, sure. IID is 'independently and identically distributed,' meaning that every individual sample is drawn from the same distribution and has an equal probability of occurrence when sampled. in practice, the IID constraint is typically violated somewhat during even pretraining.
July 24, 2025 at 1:33 AM
yeah, sure. IID is 'independently and identically distributed,' meaning that every individual sample is drawn from the same distribution and has an equal probability of occurrence when sampled. in practice, the IID constraint is typically violated somewhat during even pretraining.
Ik zou zeggen: dit is Hollands, niet Nederlands. Brabanders doen dat iid over het algemeen anders.
December 10, 2024 at 10:36 AM
Ik zou zeggen: dit is Hollands, niet Nederlands. Brabanders doen dat iid over het algemeen anders.
Just gutting. The IID is itself an attempt to mitigate the economic harm of a DUI charge/conviction relative to the logical sentence — taking away the driving privilege.
The California ACLU is staunchly opposed to the ignition interlock device program and is a big reason that it has not been expanded. Despite the fact that California heavily subsidizes the cost for low-income drunk drivers, the ACLU says that the program is a form of "racialized wealth extraction."
December 12, 2025 at 8:32 PM
Just gutting. The IID is itself an attempt to mitigate the economic harm of a DUI charge/conviction relative to the logical sentence — taking away the driving privilege.
in the 2000s iid draw muscular werewolves covered in blood ripping off dogtags and screaming in the woods
i should do that again
i should do that again
February 5, 2025 at 6:21 PM
in the 2000s iid draw muscular werewolves covered in blood ripping off dogtags and screaming in the woods
i should do that again
i should do that again
February 18, 2025 at 1:32 AM