Read and retweet. One of thr best investors in India who has made multibaggers from various sectors such as seeds, Financials, Auto Ancs, Real estate, Chemicals, Pharma,Technical textiles, Animal health etc. Uniquest and most realistic+diverse approach in my view @SamitVartak

Andhakasuravadha Samharamurti -#story #Thread
Sculpture of Bhagwan Shiva as the destroyer of the Asura Andhaka - Cave 29,Ellora caves

Asura Andhaka hangs on to a trishula held by 6-armed Shiva.

Once Parvathi & Shiva were relaxing in the pleasant gardens of Mount Mandhara.

Bhagwan Shiva sat down there to enjoy the happiness.Parvati closed the eyes of Shiva playfully & darkness covered the whole universe.Parvati was perplexed & her palms were wet with perspiration.A few drops of her sweat mixed with sweat from Shiva’s body& fell to the ground.

From this sweat,a fierce creature was born,it started to roar.Child was born blind as Shiva's eyes were closed so was named as Andaka.Andha means darkness&also ignorance.Around the same time Asura Hiranyaksha was performing rigorous tapas Shiva for a boon of powerful son.

Bhagwan Shiva was pleased by tapas of Hiranyaksha & blessed him with darshan.Hiranyaksha asked for a powerful son,Shiva told him that he cannot have his own son,but can adopt a blind child who would become very powerful later.Shiva gave child Andaka to Hiranyaksha.

Hiranyaksha returned to his kingdom with Andaka.Later after the vadh of Hiranyaksha, Andaka is said to have lived with his cousins.He was ridiculed,so dejected Andaka went to forest.He did severe tapas for over many years towards Brahma Dev. Brahma dev appeared before him.

This is Kuinini ‘Nini’ Manumua, the woman who was ultimately displaced by inclusion of Laurel Hubbard.

She’s 21, and it would have been her first Olympics.

Thread: *Post-Liberal reading list*
1/ Religion and the Rise of Capitalism - R.H. Tawney
Unto This Last - John Ruskin
Liberty, Equality, Fraternity - James Fitzjames Stephen
The Great Transformation - Karl Polanyi
The Quest for Community - Robert Nisbet

2/ Family and Civilisation - Carle C. Zimmerman
Science, Politics and Gnosticism - Eric Voegelin
Liberty Before Liberalism - Quentin Skinner
The Great Debate - Yuval Levin
Lincoln and the Politics of Christian Love - Grant Havers
Where We Are - Roger Scruton

3/ The Humane Vision of Wendell Berry - Various
The Unsettling of America - Wendell Berry
Moral Matters - Mark Dooley
Red Tory - Phillip Blond
Blue Labour - Eds. Ian Geary and Adrian Pabst
Together for the Common Good - Eds. Nicholas Sagovsky & Peter McGrail

4/ The Politics of Virtue - John Milbank & Adrian Pabst
Localism in a Mass Age - Eds. Mark T. Mitchell & Jason Peters
The Antimodern Condition - Peter King
The Principle of Duty - David Selbourne
Nostalgia - Anthony Esolen
A Common Human Ground - Claes G. Ryn

5/ The Politics of Gratitude - Mark T. Mitchell
A Time to Build - Yuval Levin
Remaking One Nation - Nick Timothy
Despised - Paul Embery
Defending Identity - Natan Sharansky
Between Kin and Cosmopolis - Nigel Biggar
The Virtue of Nationalism - Yoram Hazony

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.