Turing.jl[1] is quite usable and isn't slow[2].
[1] - https://turing.ml/dev/ [2] - https://arxiv.org/abs/2002.02702
365 karma · joined May 5, 2014
[1] - https://turing.ml/dev/ [2] - https://arxiv.org/abs/2002.02702
A few tips for coding screencasting:
For coding screencasts you should make sure that chroma subsampling isn't enabled when you're doing the encode of the video otherwise it may look worse with highly contrasting areas.
Decrease the encoding frame rate if its primarily going to be just a screencast without a lot of motion on the screen. At the expense longer mpeg segments (and more latency) you may want to increase the group-of-picture size.
There shouldn't be a reason that the encoded screencast looks worse then how it appeared to you unless the encoder you are using is just mashing the quality down.
The point estimates for mortality rate, R0 and the like all are subject to interpretation/manipulation. The health agencies have allegedly been subject to regulatory capture both in China and in the USA.
1. The communication of the CDC has changed markedly since the appointment of Pence to the head of the coordinating council. The messaging appears to be more controlled and filtered.
2. The frequent changing of the definition of what a "confirmed case" is in China leads me to completely disregard any published numbers much like other economic metrics published by China.
3. The restrictive testing conditions of the CDC likely has led to multiple cases not yet being discovered in the USA.
As a scientist I'd like to see independently reproduced results across other independent population samples. I think it is healthy to take a more Bayesian approach to these statistics, because you should let your beliefs be updated by additional data and factor in your prior assumptions.
Citing data from a single study from a Chinese health agency just isn't enough for me to change my beliefs completely, my prior knowledge is too strong when combined with everything else circulating in the political/regulatory sphere.
My personal distributions for what the values of mortality and R0 are both still have really fat long tails. But of course that's subject to change.
I lived through Hurricane Sandy in NYC. That was something. But I never felt the urge to go preemptively shopping before. Food has always been in abundance and regularly available a few blocks away 24/7. Shoutout to Westside Market in the UWS. Yesterday I became convinced I needed almost a months supply of non-perishables _right now_.
Tonight, there was no other customers shopping with urgency at my local Target as I was buying six spare toothbrushes with two bottles of NyQuil and two cans of saline nasal spray along with all of the food I could fit. All of the aisles were well stocked, mostly. They were less so once I went by. There was one lady with 7 cans of lysol spray, but we just winked at each other and hurried off. I think both of us were trying to not get any more shocked glances from the people just strolling trying decide which flavor of double stuffed Oreo's to buy this Saturday night. Deep down those people know what's going on, but aren't yet ready to take action.
Sure, I'm early and feel guilty because it was so easy to get everything right now. But also feels really good to have it stocked away in my pantry. If I don't use these supplies my local food bank will get everything I don't need.
I'd like to ask of you to consider doing the same! If all of this preparation is unnecessary, there are hungry people that will appreciate it later on. If the possibility to make something good come out of this possibly tragically bad situation motivates you take some time to get ready, I'm pleased.
The leanness could be increased during training by progressively trimming width/depth of weights, but I doubt if every model has this done.
To make it against whatever workplace "rules" there are seems overbearing and thought-controlling. Our personal beliefs and political experiences make up a lot of who we are, our individual perspective on the world and somewhat how we think.
How can you hold your organization to socially responsible policies and actions if you can't discuss what should be socially acceptable? Political beliefs form the ideas of what is the scope of the organization's responsibility.
In my recent experiences at a hedge fund and a very large investment bank, politics were discussed pretty openly in both environments (alas this was only in NYC). I found the discussions worthwhile and informative as it helped me understand my own beliefs more.
It's unwise to hold beliefs that you never have to justify. You should put in the intellectual work so that you can articulate and argue your own politics it will make sure your beliefs are well founded.
Keep the height restrictions inside of the city and continue to keep high rises near La Défense and serve them with efficient and effective mass transit.
I'm sure this trader has programmers working on this problem and putting in some type of limiting such that he doesn't sell off 3000 contracts in one go for a very cheap option close to expiration. 3000 contracts is nothing in SPY options but for a single name stock (not an ETF) it can be considered a sizable trade compared to the open interest.
Without proper auditing and disclosure of development processes and the new implementation of the diagnostic tests on chips (rather than disposable test media) it raises the question of how can anyone be sure exactly what and what not the machines are testing and recording? Slot machines in Las Vegas have their code more closely audited and monitored than this technology.
This situation just yells out for government regulation, peer review and reverse engineering.
Conspiracy hat engaged.
[1] - http://en.wikipedia.org/wiki/High-bandwidth_Digital_Content_...
It is still fun watch this switch go around the proverbial track, but I'm happy knowing that I'll never have to configure, build and test a switch of this complexity unless I really absolutely have to, with my largest caveat being AWS disappearing from the face of the earth.
Furthermore it's more shameful to release an update for the manual update users two days after the automatic update users get it. Get over yourselves already. This is already being exploited, push fixes out faster or atleast at the same time.
Steve Jobs was so right.
The Solar Physics group has had some great work published in the past, but having the Priscu[1] group getting results like this is also amazing. Kind of funny, that they're in the College of Agriculture and working in Antarctica.