Aaron

Finally got my @USDS Challenge Coin, from two administrators at once no less! Thanks @mattcutts and @mikeydickerson! Proud to serve.

@mattcutts Thanks for your service Matt! I really enjoyed and learned a lot working with you. Time for a well deserved vacation!

At the end of April, I'm stepping down as Administrator of the U.S. Digital Service. It's such an incredible group of people, and I'm grateful for my time working alongside them. More info: medium.com/the-u-s-digita…

@Ana_M_Medina How did you even decide what to order though?? Their menu is a book.

A sad moment

@emaland I've had great success with piece of masking tape over the top.

@Rob_Hoffman @RobinhoodApp Dang and I just got a coupon for 50% off BBBY stock in the mail.

@emaland This isn't the first time I've heard this. Maybe there's something to it.

@thosjleeper Neal Stephenson's Anathem, which is about a thousand pages. I bought it used for a great price too, so it's also the best dollar per page ratio of any book I own.

@jasonshevrin NoChiSoShaw

@WilliamLiDC I used to pronounce it like #2 until I saw your poll and realized #3 is so much fancier

@djbaskin I love dialup!

This is our business plan 💙

@john_overholt @OddLetters @BadLegalTakes 😂

@theshawwn @gwern I have had good results normalizing around the mean/std of the specific training dataset during GAN training. I have specifically tested this and ended up with better overall results versus doing something like normalizing around -1/1. Worth a try!

@gwern I’m also not convinced it matters. It’s just a linear transform. On the other hand, the biggan zero bug was a linear transform too, so perhaps it’s worth testing.

Idea: It's common for imagenet classifiers to subtract the mean and std of the entire imagenet dataset. But we don't do this for BigGAN training. Why not? It seems like the discriminator might benefit, since it's a classifier. Of course, this is specific to imagenet.

@jaredcunha @froidotdev Fine, a little more than 400 hours 😉

@ChelseaParlett A mathsk!

New favorite mask (I’m trying to smile with my eyes in these but I failed 👀)

@konklone COVID_Relief_Bill_redline_FINAL_FINAL_v1v2v3_reallyfinal_final_last_edits_for_real.docx

@kathytpham This is really awful, I'm so sorry Kathy.

Interstate, pandemic time death storytime: My dad died unexpectedly last week. He was a Vietnamese refugee who was a truck driver for 32 years. He continued driving during the pandemic. They found him over the steering wheel on the side of the road in Hayes County, TX 1/n.

@cdebotton @cassidoo The blues are great for getting people to mute you on zoom.

@DanCrenshawTX @BadLegalTakes

Prosecuting people for holiday gatherings inside their own home violates the 4th Amendment - protection from unreasonable search & seizure. Elected leaders who violate the Constitution should be held accountable, & law enforcement should disobey unconstitutional orders.

@karpathy This series was fantastic and continued to get better right to the end.

nice! I rarely watch tv shows but I binged through this one (helped by strong nostaliga for times in the high school chess club)

@kateconger It's a loophole in the coronavirus's terms and conditions.