madenine

...and as a part of a deliberate effort to reduce the size and effectiveness of government agencies.

If you relocate a government agency HQ to an area that doesn't have any competitive jobs, you're making it more difficult for that agency to attract and retain tallent.

Just by moving the office in the first place you'll hemorrhage experienced personnel who don't want to move their lives across the country.

To proponents - that's a feature not a bug. Less effective regulation (and eventually deregulation) being the goal.

Point taken.

Still, as far as risk to RH is concerned - if this kicks off a change via their customers, regulators, or legal action the change is probably not in their favor.

RH makes money by selling data to investment firms, not off individual trades/fees.

If their customers - the firms that pay them, not app users - see the RH platform as a threat to their business they could pull the plug[0]. Or if this prompts regulators to examine RH and similar products.

The risk isn't in the outcomes of options/trades; its risk to their business model.

[0] RH's customer are actually market makers, who by and large will be profiting heavily off of this.

I don't think its the FBI's job to figure out who smashed up a Target or looted a Footlocker.

Its definitely the FBI's job to figure out who breached a federal building, let alone the seat of our legislative branch during an active session.

That being said... the FBI, DOJ, DHS, and state agencies definitely do shady stuff and track people via their devices at protests. I'm remembering surveillance planes over Baltimore that were traced back to feds.

is it a new environment, or is it an extended Jupyter Notebook? It looks like Jupyter Notebook to me. Why not Jupyter Lab?

Neither? It doesn't change the features of jupyter notebooks, and its not an improved/expanded UI like jupyter lab (you could use nbdev with jupyter lab). Its utilities and automation to make package/library development a better experience if jupyter is where you write your code.

From https://github.com/fastai/nbdev:

"nbdev is a library that allows you to develop a python library in Jupyter Notebooks, putting all your code, tests and documentation in one place."

No, its not really the same - but the parent is "planning to have a dabble with NLP and machine learning" so my guess is HuggingFace/SpaCy/PyTorch/Tensorflow/FastAI are more the consideration than touching CUDA directly.

If you want to dabble; use the cloud.

Colab is free. Colab Pro is $10/mo. GPU instances give you access to better hardware and don't lock up your machine for hours/days at a time.

Is there a common linear algebra benchmark?

Would love to see a non-Intel chip + OpenBLAS beating out Intel + MKL

Thinking of Lebron James who hypocritically spoke up about BLM and not about human rights abuses in China[1].

I don't understand this take.

Why does LeBron James have to address human rights in China before he, a Black American, can address human rights abuses against Black Americans in America?

https://www.youtube.com/watch?v=cMgarcFkXz4

Plane boarding is kinda fascinating.

I'm with you.

First real forays into FE that didn't involve slicing PSDs to tables were in Bootstrap and Foundation in ~2013. The responsive grid those frameworks delivered seemed magical at the time, but you got the sense that the benefits came on the back of some serious CSS wizardry and framework overhead.

The newer CSS features (grid, flexbox) seem so light and magical.

2018 video on Valve's use of deep learning for anti-cheat in CSGO https://www.youtube.com/watch?v=ObhK8lUfIlc

There might be something interesting here, but I’ll never know.

The real content is gated, and the blog/abstract rolls up to a sales pitch.

In short, yes.

HAVING COUNT(*) >= 2;

Back to 10%

most likely? my joke fails to pan out in reality

SELECT TransactionID, Amount, COUNT(*) FROM Transactions GROUP BY TransactionID, Amount

I'll take 10% of savings

I like to think of arxiv as the holding area for anything that COULD be peer reviewed.

Advantage is we get to see stuff now, not after it gets accepted at by a conference or journal. Disadvantage, we have to filter on our own.

Twitter and sites like arxiv sanity preserver are helpful for filtering.

Honestly, I was trying to convey the opposite - the gates are wide open and it’s never been easier to drive through.

I’m in favor of there being more and better resources to learn anything out there, but every time I see a deep learning 101 type material all I can think is “who is this for?”.

In ~July 2016 I was at a presentation by NVidia at GW in DC. They showed off how easy it was to build out and train a model using some of their tooling (Digits maybe?). After the demo they opened it up for questions and a grad student ‘asked’ “You just did in 10 mins with 30 lines of code what I worked on for an entire semester”.

That’s been the trajectory of the tools and increasing abstraction in this space. It’s just getting easier and easier to build models that work (which is great), and it gets easier and easier to do so without knowing more than an extremely high level overview of the math behind it all.

So while this looks like a great resource - who’s it for?

For jobs/problems that need you to have a thorough understanding of the math and theory behind the networks this isn’t going to cut it.

For jobs/problems that need you to get something working math or not - this likely isn’t necessary to get started.

So it’s for people that have been getting into DL but also haven’t bothered or needed to look up the math concepts?

When you go to the hospital, every machine, chemical, glove, needle has gone through an FDA approval process or has been manufactured to pass FDA standards.

There is no FDA or NHTSA for web technologies.

Now I really want pop music made by Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin

Goodfellow’s book on deep learning[0] is a good starter - the first chapters give a solid overview of ML theory as well. Elements of Statistical Learning is another.

[0]http://www.deeplearningbook.org/

Is this the Fortune500 version of challenging the person who beat you on Xbox Live to fight IRL?

On a different note, the biggest standout to me from the NYT piece was how the tax cut effected FedEx's outstanding tax liabilities. Doesn't this set up an scenario where the megacorps of our era can just kick their tax bill down the road until they get a favorable WH/Congress they can lobby?

Living in DC as a consultant, reliable Amtrak service meant I could make a 9:30AM meeting just about anywhere between DC and NY. 2-4 trips per month.

Those early morning / early evening trips are packed with other business people doing the same thing you're doing. Full train of people quietly working/reading, maybe talking to a colleague. Very civilized.

The few times I had to travel mid-day on the same routes? Shit shows of confused tourists, kids who don't want to sit still for 3 hours (don't blame 'em), and all manner of "what can go wrong will" (delays, issues with the heat/ac, issues with passengers, etc).

Its like there's two parallel dimensions of Amtrak out there.

I'd choose Amtrak 100% of the time for business travel if its an option, 0% of the time for pleasure.

Is there even a fight left?

Chrome was the newer, faster, better working option when the average user was acclimating to the web as a constant part of their lives. To a degree, Chrome became a status symbol - knowing to ditch IE and download chrome was a mark that you were minimally web and computer savvy.

For the average person to switch away from Chrome, FF would need to have an overwhelming advantage - and I'm not sure privacy is the place where the average user will register that advantage. With 100 other companies stealing and selling their data at any given time, why should they worry about their web browser?

probably. Also - lazy ad buys.

There's human effort involved in planning a micro targeting campaign. Way easier (and short-term cheaper) to do some broad strokes ads with a couple ok-ish proxies for your target market than pay people that know what they're doing. So your conversion rates are garbage and CAC higher than it could be, but your spending is on the action instead of front-loaded on salaries/agency fees.

depends what you’re doing. For querying large datasets? 100% with you.

For data cleaning, processing, analytics, ML on decently large datasets? Spark wins out

That's just like your opinion, man

Saw a tweet the other day. Walmart is a true behemouth; and they spend on R&D (anecdotally, have met some great data scientists from Walmart Labs).

For perspective:

Walmart makes ~58.7m an hour

Amazon / Macy’s / Best Buy / Sears / Target / Gap / Big Lots/ Kohl’s / J.C. Penney / Nordstrom / Dollar Tree / Barnes & Noble / Bed Bath & Beyond

... make ~57.3m/hr combined