duaoebg

I'm a ML engineer as well. There is almost nothing like it. I can go from having an idea to having the real world results in under an hour.

You're constantly having your biases thrown in your face in a way that many other professions do not.

These skills and others do genralize very well.

Lol, I do this. My customers would often try to run my software on dusty old servers and then complain that it's slow, so now I just send them the hardware for free.

The cynic in me believes this was ensure there was a ready supply of healthy men for war fighting.

Now that wars are mechanized and we have nuclear weapons this is less of an impetus.

There might be a different effect here as preditary health care in the US is unpalatable in the best of times.

I'm easily 10x a good programmer, probably more.

It's all I do, it's all I'll ever do. I'm not a balanced person and I'm ok with that.

What happened to capital letters?

I'm seeing a lot of posts on why the seasonal flu is worse.

This to me seems ridiculous by any reading of even the reported numbers.

Some cultures are win-lose, so to strike a deal you have to look like you're losing.

I've had women relay their complaints to me about Hawking. Apparently he liked to harass pretty young women by driving his chair into them.

It’s a class thing in the UK. I recommend anyone who is any good to jump on a plane to the Bay Area. Same goes for Australia. They refuse to pay competitively and then wonder why they can’t find good talent.

I wonder how long it will take until it becomes illegal to say that Epstein didn't commit suicide.

The noise looks very predictable, should be super simple to filter out

Training ResNet with 8IPUs is roughly the same as 8GPUs. I doubt it was Tensor Core optimized so I assume the GPUs can go way faster.

I'd love to spend some time and come up with Graphcore optimized models. But at this stage fused Tensor Core optimized ops are much more interesting to me.

The Tensorflow numbers are estimated. They may have estimated wrong.

They're part of an insider club, if you're an insider you'll know who they are.

These are the kind of people wear their graduation rings upside down so you'll feel the chunk of metal in the handshake.

Ah, you're the author. I missed that. Cool work by the way.

It's just a function of multi-targeting. When problem can be framed in a way that one of the outputs helps explain the other output I would call that component to be an "explanation component".

For one of my problems I know that the model would need to 'see' an orientation field in order to make its predictions, so instead of just asking for the predictions I also ask for an orientation field. The loss function has only a small weighting and is set against a very rough ground truth along with a total variation loss. This ends up producing a nicer orientation field than the ground truth.

Distillation would not meaningfully help make a model more transparent or explainable.

I imagine you're thinking about bottle necking on a really tight latent space. This would require a different architecture which would be more difficult to train and would probably suffer in accuracy.

Often it's better to re-frame the training as a multi-targeted problem with an explanation component.

When this first happened it seemed that the media was pushing the idea that be was killed as a crank conspiracy theory.

They failed and now a majority believe he was. Including me.

I think people underestimate the popular support that Maduro enjoys.

It's what happens in a democracy with a dominate population of poor underclass with a departing middle-class and a cleptocratic elite. The situation is surprisingly stable.

I see no humane resolution to this problem.

Ballmer is a buffoon. Even if we are to accept that MS was caged. I remember being at a launch party during an internship. Suddenly half the people ran out of the massive tent. I thought it must have been a fire, I ran out too, grabbed someone and asked why the mass exodus. Their reply was that Ballmer just went on stage. With that response I figured I have to see this. I went back in and recorded it. I was not disappointed, the performance was very cringe worthy.

I do hope MS becomes a nice place for women to work, as a white guy I hated it so hopefully in the future they can do better.

Defense spending is a perpetual state aid to Boeing

IQ numbers are normalized which should make it no surprise that histogram of IQ numbers form a normal distribution. It's in the definition. IQ numbers only indicate a relative position to the rest of the population. Basically (110IQ-100IQ) != (120IQ-110IQ).

I think the blog title use of 'solving' is misleading. As OpenAI didn't make the solving algorithm. It's a neat but limited use of transfer learning in an environment which requires heavy instrumentation. I could tell this immediately from the photo. What I didn't know was that the Rubix cube also had to be modified and that it had a high failure rate. While I'm sure this is state of the art in something it does seem overblown.

Way worse

Like many people I accept that there are unavoidable deaths in the name of progress.

What I don't like is that I cannot opt out of this experiment. I believe many of these deaths are needless and due to cost savings and sloppiness.

Tesla to me seems less like a center of excellence and more like a group of people who can put up with Elon. Presumably made easier by telling him what he wants to hear.

While the average driver may be bad, the average hour driven is by good drivers.

Plus I don't like the idea of a Tesla can crashing into me. Seems like something that could be weaponized.

It changes the equation enough to make it not worth attacking.

Also allows you to hold out long enough to import more weaponry from sympathetic regimes.

Plus drone based weapons are cheap and effective.

I don't see how these protests will be successful. I expect they will eventually lose and either fall in line or be replaced.

I can't help but wonder how things would be different if the Hong Kong population was armed like Americans.

It's a well known business strategy, run something important into the ground, get bailed out, rinse and repeat. Many companies are run badly on purpose.

No one forced Apple to ship production to China. It was well understood what was going to happen before they did it and they did it anyway. It's hard for me to have any sympathy for their current position. Business is ruthless and Apple knows it. Apple has a long history of crushing small suppliers to their own advantage so it's not like their hands are clean in all of this.

I don't mind being price gouged, I can afford it, but Apple no longer makes laptops suitable for me.

Employers have been threatening and actively trying to outsource my employment overseas my entire adult life. This drove a lot of people out of the software industry. I made the bet that they would be less successful than they planned. Now I gouge them for access to my rare skills - a problem they created. I could easily have been wrong and if that happened these companies would have not supported me out the goodness of their heart.

This thing must be crazy light.

I wonder how much of the advertised quietness will dissapear once it gets speced for range and passengers.

If they've solved a fundamental problem of noisy props I would be very impressed.