I haven't been harping on it because it at least contributes to taking it seriously, but this is part of why I shudder every time they call the halls of Congress "sacred" and talk about them being "desecrated".

Did you know?

When Saudi Arabia's King visited India in 1955, PM Nehru ordered all the Hindu Temples and Idols of Hindu Deities that came in Saudi King's way to be covered up.
Our temples are the architectural marvels in their own league.

Each of our temples tells the story of our rich history and cultural heritage. Was Nehru that ashamed of his own nation's heritage that he had to cover them up or there was some other reason? The answer would be best known to him only.

PM Nehru invited the then Saudi King Saud bin Abdul Aziz to India who accepted the invitation and planned a 17 day long official visit to India in 1955. Saudi King had brought a large delegation with him and he had to visit various Indian cities like Mumbai, Hyderabad, Agra,

Aligarh, Shimla, Mysore and Varanasi. However, when Saudi King reached Varanasi, Nehru ordered to cover all the temples and Idols with curtains that came in the way of the Royal Convoy. During his stay in Varanasi, Govt buildings were painted with the words "Kalma Tayyiba".

'Kalma Tayyiba' means 'There is not God but only Allah and Mohammed is his messenger'. In addition to it, Pooja and Aartis were also stopped at temples. I can call it the appeasement politics of the highest order and Nehru was indeed the greatest appeaser of all times.

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.

Great thread full of potential https://t.co/jA8h08M48z ideas! 💡

📸 Social disposable camera app

Create a digital "roll" with friends. Take pictures from the roll. You can't see them until: the roll is used up, a time limit expires, or you get to a specific location.

🎙 Asynch interview platform

Send a guest a video or audio clip asking a question. Guest replies when able with same medium. Repeat. When done, share produced video/podcast.

✅ 24 Hour Startup Checklist/Overlay

A web app to track the progress of your project: ideation, naming, shipping code, launching.

Page can also be a dynamic browser overlay for live streaming.

Have tools for polls, surveys, sharing, launching to various outlets, etc.

👩‍⚖️ Healthcare "Lawyers"

If health insurance screws you over, no one has your back but you. Instead, can pay a small monthly fee to a service to take care of BS like this for you should issues arise.

Ever notice how the top niches are centered around health, wealth, and relationships?

There's a deep-rooted psychological (and biological) reason why.

And today, I'm gonna to explain all of it so you can start profiting off of this knowledge 💰💰

*Now, this thread is going to be long and in depth. Make sure to favorite the top tweet now so you can come back to it later.

So. Down to the very core of every living being (humans, dolphins, bacteria, aliens), there are only two drivers:
1. To survive
2. To reproduce

Right off the bat, we know that sex is a primary driver to human behavior.

This would also mean that each gender has its own insecurities regarding sex.

Men have a subconscious fear that their woman will leave them for someone more successful

Women have a fear that their man will leave them for someone younger & better-looking

We can conclude that men are attracted to looks, and women are attracted to status.

This is human nature down to its very core.

Crazy, blue-haired feminists will try to argue and say otherwise.

But, 50 years of feminism DOES NOT OVERRIDE millions of years of evolutionary biology.

Here's a quick visualization exercise so you can see how true all this is...