Happy 2021, everyone!

To kick off the new year right, here is a thread of people that I love to follow. Check them out and follow them if you can. If we follow each other and you want to be added, please let me know.

Love,
Rata

JWALAMUKHI TRIPURASUNDARI, UTHANAHALLI, MYSURU (KAR)
#bharatmandir

Jwalamukhi Tripursundari is believed to be the sister of Devi Chamundeshwari, the presiding deity of Mysuru
It is also that she represents the flaming mouth of Jalandhara, the demon whom Shiva crushed to death.

When Devi Chamundeswari was fighting Mahishasura, she observed that from every drop of blood that spilled out of Mahishasura, another demon was born. So she created Tripurasundari to ensure that no blood dripped from Mahishasura.

Tripura Sundari means beauty of the three worlds.
The murti was installed during Wodeyar rule. An utsava murti made of pancha loha is kept in the sanctum sanctorum.
Another shrine here is dedicated to Siddeshwara / Shiva surrounded by a beautiful wooden Mantapa.
@GunduHuDuGa

Important paper from Google on large batch optimization. They do impressively careful experiments measuring # iterations needed to achieve target validation error at various batch sizes. The main "surprise" is the lack of surprises. [thread]

https://t.co/7QIx5CFdfJ

The paper is a good example of lots of elements of good experimental design. They validate their metric by showing lots of variants give consistent results. They tune hyperparamters separately for each condition, check that optimum isn't at the endpoints, and measure sensitivity.

They have separate experiments where the hold fixed # iterations and # epochs, which (as they explain) measure very different things. They avoid confounds, such as batch norm's artificial dependence between batch size and regularization strength.

When the experiments are done carefully enough, the results are remarkably consistent between different datasets and architectures. Qualitatively, MNIST behaves just like ImageNet.

Importantly, they don't find any evidence for a "sharp/flat optima" effect whereby better optimization leads to worse final results. They have a good discussion of experimental artifacts/confounds in past papers where such effects were reported.

1. Now that Stan Lee is being eulogized far and wide, it's important to remember how marginal Stan Lee and his collaborators (Jack Kirby and Steve Ditko) were in the 1950s and 1960s.

2. Marvel Comics (before that Atlas) was just a cog in the machine of a bottom pulp publisher run by Martin Goodman, the husband of one of Lee's cousins. It was the lowest of the low in the publishing world.

3. Now Mario Puzo (not yet the author of the Godfather) shared offices with Stan Lee in the 1950s and 1960s. Puzo wrote for garrish men's adventure magazines and, like Lee, dreamed of writing a novel & breaking out. But Puzo looked down on Lee.

4. Flo Steinberg, 1960s secretary at Marvel: "They were always making jokes about us. They'd come in and giggle. mario Puzo would look in and would see us all working on his way to the office and say, 'Work faster, little elves. Christmas is coming.'"

5. When JFK was killed, the whole office of Magazine Management was stunned and quiet. Except Lee. He continued working. "He was still working on the comic books," Puzo said. "Like that was the most important thing in the world."

If you're a college student who wants to pay off all your loans... here's an idea for you:

It takes ~$5k to start and will make ~$30k a year of profit

It won't get you rich, it'll pay for college.

(..and will teach them more about business than they’ll learn in school)

The idea isn’t sexy. But it works:

**Plastic Moving Bin Rentals**

Everyone knows moving sucks. The worst part is being stuck with 400 cardboard boxes at the end of your move.

Instead - @benmlevy told me about Gorilla Bins. They dropoff plastic moving bins, & pickem up later.

Let’s break it down. Our goal is to pay for college (~$30k profit per year).

To do that, I think we’ll need $50k in revenue.

The avg move (a 2 bedroom apartment) is $180

To hit our goal, we need do ~5.5 moves per week

COSTS?

How many bins do we need to buy up front?

5.5 moves a week x 50 bins = you need 275 bins. Let’s round up to 300 to be safe.

Ok let’s hop over to @uline to see how much these bins cost

Bingo. Bins cost about $15 a piece.

$15 x 300 bins = $4,500 initial startup costs.

Time to beg your parents or a rich aunt to lend you money to get the biz rolling (...or just take $5k of student loan money and invest it here…)