jeffshek

Risky to say, considering deep learning by Hinton was also once called a failed experiment ...

Apple sent a survey asking once how macbook owners would feel if they removed it.

They believe the answer is going to be Bluetooth.

I was pleasantly surprised about the 64gb also. Same issue with Docker making me want to upgrade.

This has the largest battery possible under TSA regulations.

I built https://writeup.ai to help with that, but while it helps, it still feels like it's missing "something" at times.

If anyone's interested - I've written a guide to Anki and spaced repetition. I've used it for about 5-6 years and my deck is about 12,000+ cards. Much of it is based in software engineering.

https://senrigan.io/blog/everything-i-know-strategies-tips-a...

At Tesla's scale and priorities, they'd probably be less keen on using external cloud providers. Using TPUs at their scale would certainly require Google's AI consultants to supervise which isn't ideal for Tesla.

Not agreeing or disagreeing with their decisions, but if you have the resources, you can certainly design a custom chip that performs a specific type of task very well that beats other competitors. Nvidia's GPUs are have to be reasonably good at training across different NNs. You could have a chip that's exceptional good at training one/two specific types of tasks.

For most companies, this would be a bad idea. However, Tesla knows how to produce hardware.

Some potential reasons:

- TensorFlow is great at deployment, but not the easiest to code. PyTorch isn't frequently used in production until recently.

- If you have the resources for great AI engineers and researchers, your team will be good enough to build and deploy both frameworks.

- Preference toward the easier framework your tech leads prefer.

- Lots of new academic research is coming in PyTorch

- TensorFlow is undergoing a massive change from 1.1x to 2.0; if you choose TensorFlow, write on 1.1x just to then refactor to TF 2.0? Or write on TF 2.0 now and deal with all new edge cases? Or write in PyTorch (easier) but handle the more difficult deployment process.

- ML code quickly rots. Bad PyTorch code is just bad Python code. Bad TensorFlow code can be a nightmare to debug.

- PyTorch's eager execution makes coding NNs much easier to prototype and build.

IIRC, Zuck has debated running in the future. Having Facebook with that power would probably benefit.

After hours price : Tesla Stock is up 16.70%

+ Adjust earnings per share of $1.86 vs. expected losses of 42 cents per share (major reason)

Thank you for writing this. Regardless of what happens, I hope you find the time to create something with your woodworking and upload it somewhere for rest of us to see what you've been working on.

There's some harsher comments here, and it's one of my larger fears when posting anything to HN. The technical crowd can be unforgiving at times.

I loaded the site on desktop, I liked the visual arrangement a lot -- it's a clean/fresh aesthetic.

Did you write down all your bets? Not just the winners, but also the losers?

This is a bias that a lot of investors often make ...

That's a bit too much dislike toward Steve Ballmer. The enterprise market is completely owned by MSFT today (huge part due to him). Azure wouldn't be Azure if all the enterprise products/contracts already in place. Sure, they missed an opportunity on mobile, but so did BlackBerry.

The stock market (stock price) just didn't love MSFT despite strong financials during Ballmer's tenure.

https://www.macrotrends.net/stocks/charts/MSFT/microsoft/rev...

Hi Max :) Huge fan!

I'm adding this link on the article since I'm sure that's a bit more straight forward than my Docker mess lol.

https://github.com/huggingface/transformers (was) based on PyTorch, so they originally had lot of the models like gpt2-small. Because of that influence, PyTorch was probably going to win there.

The one con about TF Serve (TFX) is packaging the entire model into the container (so that ends up being 3gb?+). This was a couple of months ago, so I might be wrong by now ... It was an area I wasn't very confident I could do, TF Serve is really new, so there aren't many guides (and many were already out of date).

Regarding the diagram - yeah, it's SO much easier to visualize when you see it. It takes oddly a bit of time just to make boxes fit though. Huge fan of Traefik too :)

Most of my fine-tuning was on 355 (otherwise, 774 is too hard to train).

However, the default prompt on https://writeup.ai goes to 774.

The GPU with inference is helpful for much faster responses -- for instance, it's easy to run inference (CPU) if you're getting a one way shot of 200 words (no revisions), but if you're constantly changing and revising, then speed ends up mattering a lot for UX.

EDIT: Looked over your articles, I'm literally floored/amazed you're in high school and know this much. The whole world is your oyster.

It's just the writing style :)

Terse with a bit of snark and ecstasy jokes. Lots of references to startups.

It's definitely entertaining!

I'm still conflicted on the merit of lucid dreaming and problem-solving.

I find ZMA and Huperzine A can you get you lucid dreaming, but I don't know if it helps with problem-solving. IE. You'll still wake with the solution w/out dreaming.

Yep! It's been hard to ignore the other things people questioned about his attempt though ... I'm leaning toward trying biphasic, but still not sure yet.

Joshua Waitzkin has an interesting perspective on this. He will take an idea, keep it in his hold, and then make sure he lets go of the focus a few hours before bedtime. IE. Do something with family instead.

Subconscious brain still works on it, but still able to sleep well. Not sure if this would be applicable.

I think the different stages of sleep play a part here ... Salvadore Dali apparently liked to keep an object/pen in his hand that would drop and wake him up.

Still can't find enough real examples of polyphasic sleeping working, but love to know how well people of that retained content.

How is your blood pressure variability?

I went for 4-5 hours of sleep for a year; while resting blood pressure was normal, I found it would spike high easily.

Congrats, this was so impressive! Game graphics is an area I know nothing about. Everything just seemed magical.

Sure they can. They banned explosives (which I agree with) and you can probably do that with a few hundred dollars.

I've read a reasonable amount about this and listened to a bit of the podcasts they've done.

It feels a bit too early to tell. I don't believe many researchers will switch to Swift though.

This feels like it was written by GPT2-Small.

RenTech?

Cold showers have helped me. I've already had a reasonably good immune system though. I am on the weird side and probably take 2-5 cold showers a day though.

If I feel a little off, I'll take Vitamin C and Lysine as early preventive measures. The rest of my healthy habits are quite subpar. I exercise frequently, but I eat garbage and I sleep even worse. Averaging about once a year in being sick?

But strong machine > all?

IE. Chess / Go / Dota 2

I'm not quite sure if I really agree with the premise of this article. This isn't some type of Uber vs. Lyft analogy. WeWork dwarfs all the other coworking spaces. Customers aren't going to leave WeWork because they fear it going out of business (after-all that's the point of month-to-month). The turnover happens from unhappy customers.

This also seems to imagine given all their recent troubles that WeWork isn't going to dramatically improve operations. WeWork has never gone back to make things efficient since inception, so it wouldn't be a huge surprise if there's a lot they've left on the table. WeWork owns so many leases from landlords; the landlords might be strong-armed into concessions.

It's not unreasonable to expect a lot of their metrics are going to improve (and cut costs drastically down), especially with a new set of management.

I just signed up for WeWork in Lima, Peru this week. There's definitely customers. Everything is clean/organized. Although, the music is definitely a bit distracting ...