I found Amplify excellent to get up and running quickly. I’d highly recommend it for anyone without a well-oiled CICD setup who wants to quickly get a website up to test out an idea.
Unfortunately, I quickly hit the limits of its configurability (particularly with Cloudfront) and had to move off it within a few months.
As someone who works in applied ML but tries to stay heavily on top of research, I think it’s this. I am always stunned when I try to recruit people, talk with folks at conferences, or even look for jobs myself how few people care about commercialization. I’ve seen this attitude grow a lot in corporate R&D in recent years too as they pull in more academics with promises of independence from the “business grind”.
I'm not sure that it is just lack of interest in commercialisation, though that does exist.
I think that the academy and industry work in different scale.
In the industry low risk short term project have high value.
The academy aims for novelty which usually means high risk and longer terms.
Companies that aim to high risk longer term are usually either start up that that project is their core or large companies that can afford failing 9 out of 10 projects and have the benefit form the others to return the investment in 5 years.
Be careful -- If it turns out that those rejected end up with the $10k, YC might start making it a policy of rejecting companies and accepting them on their second application :P
If you want to use the p3 instances with Tesla V100's, you'll need to use CUDA9 instead of CUDA8, and unfortunately none of Amazon's AMI's have the latest PyTorch that support CUDA9. The easiest way to use them is get one of the Conda Deep Learning AMI's and then follow the latest PyTorch install instructions, which seems to work pretty seamlessly.
The problem with digital data is that it can be replicated at no cost and with no trace, so there wouldn't be any way for Ashley Madison to ever be sure the leak was contained. They could pay $10MM for a copy of the data... and two months later it can anonymously show up on Tor. Not to mention the hackers would need to come forward to collect the $10MM, which might land them in a jail cell before they can collect.
All bitcoin transactions are public record, and even if you tumble you are setting yourself for getting caught by repeatedly doing the same actions. Besides, the value of the information decreases over time, meaning you're going to be making the large payments up front.
> even if you tumble you are setting yourself for getting caught by repeatedly doing the same actions.
By doing what? Making payments to an address?
> Besides, the value of the information decreases over time, meaning you're going to be making the large payments up front.
Any lumpsum can be structured as an equivalent series of regular payments. Whatever the present value of not having the data released is, it can be turned into a set of finite or indefinite payments, which last 1 year or 5 years or 10 years or however long the company feels the harm will be non-neglible. And by doing this, you still get the anti-defecting incentive of a stream rather than one-off payment.
Those transactions become permanently public, and can be used against you in court if at any time in the infinite future you are tied to that address, or tied to any addresses pulling bitcoins out of that account (which are also publicly recorded actions). Compared to cash, which is not a matter of public record, bitcoin is not a "Perfectly Adequate" solution.
Without contracts or signals, you can't make monthly payments on a depreciating asset/service because it becomes an auction. The hackers are rationally incentivized to maximize their returns, and that includes selling the data to the highest bidder before the price depreciates. That means that Ashley Madison would need to beat the highest bidder each month, without knowing if the other bidders are even real.
You apparently don't. Payments do not in themselves offer any risk whatsoever when the payments are made to a brand-new address (which is so obvious and basic a point of secure Bitcoin use I didn't need to mention it); these payments to the blackmailer can then be mixed in numerous ways at varying costs and levels of efficacy - Coinjoin, a large DNM, a web wallet, an exchange (either to just gradually withdraw, or sell outright into another cryptocoin & then sell that cryptocoin at a different exchange, particularly one of the cryptocoins with built-in anonymity features), sent to stealth addresses, etc etc. The pseudonymity of this does not affect the mechanism of turning a lump sum into a stream of payments: the payments can be safely made with reasonable levels of transaction costs, and thus the blackmailed and blackmailers can transact safely and keep the data private or not, with neither side being ripped off for more than one payment.
> The hackers are rationally incentivized to maximize their returns, and that includes selling the data to the highest bidder before the price depreciates. That means that Ashley Madison would need to beat the highest bidder each month, without knowing if the other bidders are even real.
If the hackers are bluffing, then AM can always call their bluff. If AM is not confident they are bluffing, then they will simply pay up to the value of keeping the data secret. Where's the problem here? Again, why does a stream of payments fail where lump sums would succeed?
I'm a big fan of the BetterSnapTool available on the App store. It costs $1.99, but it works with all widows and has drag-to-snap, shortcuts keys, etc.
I'm still surprised Apple hasn't just incorporated this into OSX.
BetterTouchTool is free, and strangely incorporates the same window snapping functionality as it's sister: BetterSnapTool. I've always been a little confused by the paid download for BetterSnapTool... Unless I'm missing something you can get all the same functionality and more (mouse and trackpad configuration etc) with BTT.
This is largely just an argument against bad presentations - PowerPoint or otherwise. I agree that PowerPoint is not the right medium for many things (e.g. technical reports), but a lot of these critiques are also the strengths of PowerPoint. The lack of space forces the presenter to slim down their argument to the essentials. That shouldn't encourage burying important information in sub-bullets, but to the contrary, encourage highlighting key risks/concerns.
The downside of technical reports and white papers is that it's far too easy for people to blather on without purpose, and it's far too hard for me as the audience to review. I can skim through a PowerPoint in minutes and get a good sense for whether it's worth my time. It's much harder to do that with a long free-form text. I also think the presenter is more important than the medium in most cases: I bet the same person making those awful NSA slides would write an awful white paper (and vice-versa).
Long story short, people have a really hard time concisely and precisely expressing their arguments, and we all need constant practice to get better. Everyone should also have several options in their presentation tool kit and be able to use the right one for a given task.
I agree, but I'm not sure I would have said that without first getting clobbered by the AI. The list of really hard things for a human seems to be growing these days.
Love this package. I was using PyBrain for a recent project, but I had an awful time getting it to work and found the performance equally awful. After stumbling upon Keras, I was up in running in minutes, and the huge performance boost of Theano allowed me to actually finish on time. Thanks for putting this together!
Unfortunately, I quickly hit the limits of its configurability (particularly with Cloudfront) and had to move off it within a few months.