Definitely not the first to post on this but am completely floored by a 2008 paper published in the Journal of Infectious Diseases by Professor David Morens (Medical Epidemiologist), Dr Jeffery Taubenberger (Virologist), and Dr Anthony Fauci (Immunologist) at the NIAID. 1/11

WHEN GH0RI TOOK TO HIS HEELS WHILST FIGHTING A HINDU QUEEN

THE BATTLE OF KASAHRADA (Kayadara)
1178 AD

Strategy! Bravado! Glory!

(Thread)

The narrative set by left historians about one of their favorite poster boys "Gh0ri" is that he went unchallenged upon entering the Indian subcontinent with one king falling after another.(1)

They conveniently deleted this event. Why? An agenda aimed at making us feel weak and inferior against the invaders in an attempt to glorify them. Simple strategy: To destroy a civilization tamper their history.(2)

Gh0ri defeated Prithviraj Chauhan at the 2nd Battle of Tarain in 1192 and laid the foundation of Delhi Sultanate. What is not well known is the fact that 14 years before he won this battle, he was defeated by the queen of Gujarat "Naiki Devi"!(3)

In 1173 Md Gh0ri had conquered the Ghaznavids in Afghanistan. A ruthless, persistent and deceptive Gh0ri was successful in conducting massive raids deep inside Indian territory.(4)

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.

one of the things i like about Batman is that it's fundamentally about being orphaned, and being an orphan — how this guy stays tethered to the world in his own weird, abstract, extrajudicial way when he feels that his link to it has been violently severed

Batman does this by permanently recapitulating his own trauma: solving crimes for other people, protecting them. But that's actually just the beginning; he also recapitulates his own experience of orphanness by essentially raising several orphans

Dick is an orphan in the traditional sense: dead parents. Batman sees it happen. Jason, less so: missing dad, known mom who eventually dies. Tim, even less so: parents are around a while, and involved, before both dying. Carrie, even less than that: negligent parents.

Steph: evil father, who openly works against her. And then there's Damian, whose parents are both alive and yet both have identities that totally obliterate their role as his parents. He doesn't relate to them as a mother and father; he can't, and neither can they.

In all these different cases, Batman tries all these different ways of making orphanness for them not what it was for him: Either by trying to teach them the (arguably poor) coping skills early that he learned over a very long time, or by literally legally adopting them, etc.

Remembering the Bengali scientist Dr Subhash Mukhopadhyay- the Pioneer of IVF in India, the creator of India's first and world's second child using IVF on his birth anniversary. The Father of IVF in India who was rewarded with suicide was born on 16.01.1931in Hazaribagh.
#Thread

Dr. Subhash was the son of a doctor and studied at the National Medical College in Kolkata after completing his schooling. Fascinated by innovations in gynaecological surgery from his early days as a medical student, he completed his PhD in reproductive physiology +

from the University of Calcutta before going to the Edinburgh University in UK for a PhD in reproductive endocrinology. On his return to India in 1978, he started researching ovulation and spermatogenesis. Soon after, he teamed up with Sunit Mukherji, a cryobiologist +

and Saroj Kanti Bhattacharya, a gynaecologist, to work on a method of in-vitro fertilization for a patient (Bela Agarwal) with damaged fallopian tubes. +

Mukhopadhyay with a double PhDstarted work on IVF techniques in the mid 60s and created history by becoming the first physician in India and second in the world after British physicians Robert Edwards and Patrick Steptoe) to perform the In-vitro fertilisation. +