THREAD: DC Math > 1 + 1 = 2 or is it?

Kudos to the FBI on their quick action trying to locate the person responsible for the 2 pipe bombs in DC. Whew! What a close call, right?

My question is: What happened to the THIRD?

ADI SHANKAR AND SRI TOTAKACHARYA

Adi Shankaracharya chose his disciple Totakacharya to be the first head of the Math at Joshimath.
The Narsingh Mandir (Mandir of Sri Narasimha) in Joshimath is one of the 108 Divyadesams and also the home of Badrinath during the winter months.

Totakacharya was earlier a boy named ‘Giri’ and a devout disciple of Adi Shankaracharya. Unlike others, he was not known to be an expert in Shastras but his dedication to his Guru was complete and he used to listen to His discourses with utmost faith and devotion.
@SriRamya21

One morning as Adi Shankara and his disciples were getting ready for the discourse, Giri was delayed. The other disciples urged their Guru to commence the lesson on Bramhsutra Bhashya telling him that Giri would anyway not be able to comprehend its deep meaning.
@RajeAiyer

Adi Shankar responded “नायमात्मा प्रवचनेन लभ्यो न मेधया” – Atman cannot be attained through discourses or mere intellect.
He,then, awakened the supreme knowledge in Giri.
Giri spontaneously composed: श्री शङ्करदेशिकाष्टकम् extolling his Guru’s grace.

Adi Shankaracharya granted sanyasa to Giri and gave him the name Totakacharya.
The story of how Giri became Totakacharya is a great example of the importance of faith and devotion in your Guru and not the mere knowledge of the Shastras.
@DetheEsha

BANASHANKARI AMMA, NEAR BADAMI (KAR)
#bharatmandir #navratri2021

Originally built by Chalukyas in 7th century. Devi Banashankari was their family deity. This image of Banashankari Amma was installed by Jagadekamalla I in 630 AD.

Later renovations were done by Marathas.

Once, a demon Durgamasur was troubling the people of this area. Answering the prayers of Devas, Ma Parvati emerged from the yajna fire as Banashankari and killed Durgamasur.
This area has rich forests of coconut, betel nut and other trees.
@GunduHuDuGa

It is said that during a severe famine, Devi provided vegetables and fruits to the people of this area.
Hence, she is also worshipped here as Shakambhari.
In the annual festival called Banashankari jatre, a Rath yatra takes place and Devi moves around the city in the chariot.

After 17 years and more than 300 races, it could be @alo_oficial's last F1 race on Sunday 😭 He's given us a lot over the years... including some superb GIFs!

Here we have a rare example of a GIF you can actually hear. TOMA. 💪

The Spanish Samurai

Never give up. 💪 #GraciasFernando

Winning is everything. #GraciasFernando

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.