HN user

Dawny33

676 karma

mesa name Daani :)

Posts88
Comments49
View on HN
www.youtube.com 5y ago

PyTorch Performance Tuning Guide [video]

Dawny33
3pts0
ray.readthedocs.io 6y ago

Tune: A Scalable Hyperparameter Tuning Library

Dawny33
2pts0
joschu.net 6y ago

An Opinionated Guide to ML Research

Dawny33
2pts0
xojoc.pw 6y ago

Why I'm focusing only on some programming languages

Dawny33
5pts1
cstheory.stackexchange.com 7y ago

List of theoretical CS workshops and conferences

Dawny33
1pts0
techcrunch.com 7y ago

Happy 10th anniversary, Android

Dawny33
2pts0
techcrunch.com 7y ago

Uber reports Q2 losses of $404M, up 32 percent from Q1

Dawny33
3pts1
asatarin.github.io 8y ago

Testing distributed systems

Dawny33
127pts2
twitter.com 8y ago

Solomon Hykes leaves Docker

Dawny33
1pts0
james-iry.blogspot.com 8y ago

A Brief, Incomplete, and Mostly Wrong History of Programming Languages (2009)

Dawny33
3pts0
archive.bridgesmathart.org 8y ago

Hypocycloid Juggling Patterns [pdf]

Dawny33
2pts0
namc.in 8y ago

16 year old app developer: Plagiarist or Prodigy?

Dawny33
243pts148
www.hindustantimes.com 8y ago

ISRO (India) launches 31 satellites in one go

Dawny33
23pts4
www.business-standard.com 8y ago

ISRO (India) launches 100th Satellite

Dawny33
1pts0
eng.uber.com 8y ago

Year in Review: 2017 Highlights from Uber Open Source

Dawny33
2pts0
createmomo.github.io 8y ago

NER with a CRF on top of a Bi-LSTM

Dawny33
2pts0
thenewstack.io 8y ago

Solomon Hykes Changes Roles at Docker

Dawny33
1pts0
segment.com 8y ago

2017 Segment Open Fellows

Dawny33
2pts0
www.technologyreview.com 8y ago

The financial world wants to open AI’s black boxes

Dawny33
1pts0
unsupervisedmethods.com 8y ago

Learning Machine Learning and NLP from 187 Quora Questions

Dawny33
4pts0
www.pagerduty.com 8y ago

Failure Fridays: Four Years On

Dawny33
1pts0
malaikannan.io 9y ago

Why Use Activation Functions in Deep Learning?

Dawny33
4pts0
github.com 9y ago

Tools in Data Science Course Material – GitHub

Dawny33
2pts0
speakerdeck.com 9y ago

Understanding Serverless Architectures

Dawny33
2pts0
redmonk.com 9y ago

How AWS Cloud is demolishing the cult of youth

Dawny33
5pts0
dataversioncontrol.com 9y ago

Introducing Data Version Control – Iterative Machine Learning

Dawny33
2pts1
blog.serverfault.com 9y ago

Introducing DnsControl – “DNS as Code”

Dawny33
9pts0
unix.stackexchange.com 9y ago

Why printf is better than echo

Dawny33
2pts0
unix.stackexchange.com 9y ago

Pros and cons of Vim and Emacs

Dawny33
3pts0
www.nextplatform.com 9y ago

Scaling Deep Learning on an 18,000 GPU Supercomputer

Dawny33
3pts0

* What do you think India will look like in 2026? *

India in 2026 will probably be a federation of self-governing states, bound together in a great Empire under the suzerainty of Great Britain, with Calcutta still the political capital, and Bombay and Madras the chief seats of commerce and industry. The native states will have increased in number and importance, and will constitute a large part of the federation, while the British territory will have diminished. The entire population of the Empire will be at least 320,000,000. The languages spoken will be English, and various vernaculars. The chief religions will be Hinduism, Mohammedanism, and Christianity. The standard of life will have risen considerably, and the general level of culture will be higher than in 1900. The cities will be larger and more beautiful, and the rural districts more thoroughly cultivated and peopled. The general aspect of the country will be greatly altered by an extensive system of irrigation, and by the forests, which will clothe the lower slopes of the Himalayas, and stretch across the northern plains into central India. The native states will be full of railways, and so will the British territories, though these last will be more sparingly supplied than the former. The chief exports will be opium, cotton, cereals, hides, and oilseeds; the chief imports, manufactured goods and hardware. The feudatory chiefs will have become more thoroughly loyal to the paramount power, and all classes of the population will be contented subjects of a beneficent rule. (W. S. Lilly, in the _Fortnightly Review_.)

DuckDuckGo Traffic 7 years ago

As long as the search results don't need to be extremely localized, I found DDG much better and less obnoxious for a year.

Still can't get away from "Pizzerias around me" style queries on G + not expecting DDG to take these up anytime soon, as they are probably like 10% of my search queries per day anyways.

Submit a deep learning project. If your application isn't accepted at this stage, we'll tell you why.

Why only 'Deep Learning'? When you call yourself an 'Online Community for AI Devs', you need to consider projects from any domain in AI.

Deep Learning is just a sub-part of ML, to be precise, neural nets.

Please stop throwing keywords

We do monitor too. We made a Master-Worker architecture which scales according to the data traffic. (Talked abt it in Velocity Conf. London this year)

One way is we monitor the Ansible tasks which make this happen. Apart from that, there is no way we can do it.

+ trying to make it happen by using something like a Prometheus+Grafana server would defeat the entire purpose of going serverless

grouping a bundle of functions together behind an API gateway, you’ve created a microservice

Always beats me why people don't warn the readers about the almost impossible nature of logging and monitoring such systems :/