mrintellectual

A lot of previous transformer-based models did not have great depth simply because "shallow" transformers already took forever to train.

I am concerned that we are approaching an age where only massive companies and research groups with tons of GPU resources will be able to train next-gen models.

Well put. It is easy to forget that the actions of governments do not necessarily represent the will of the people.

This goes over network construction very well, but I'd like to know about more about how modern compilers go about doing autograd.

I've heard great things about Vespa as well (I used to work at Yahoo). I unfortunately never got the chance to play around with it too much myself.

Looking forward to doing some at some point in the near future though.

As mentioned in the article, I recommend Milvus (https://milvus.io) - it's open source and cloud native with standalone versions available. Alternatively, if you're looking for an open-source solution for generating embeddings, I recommend (https://github.com/towhee-io/towhee).

Thanks for your feedback. I'm writing a post on vector indices and will throw it up this week.

There could be a personalized setting to show/hide Twitter links.

Hypothetically, you should report what happened and hire an attorney ASAP.

> You can pay yourself a “market salary” of $500,000/yr+. You can use the money as a piggy bank for flights, hotels, Uber Black, and other travel expenses.

This is just downright false. The founders alone do not decide C-suite compensation - that's reserved for the board of directors (which should be a combination of one or more founders plus other individuals selected by the investors). Travel expenses and other "piggy bank spending" would also have to be reported in quarterly or yearly financial reports to the shareholders.

If what this article says is true, the investors deserve just as much blame as the founder.

LinkedIn is good for networking and keeping up with existing connections, but the "social" aspect of it definitely feels awkward for me.

The shift to Meta is supposed to reset that growth, in a way. Not sure how it will turn out.

I'm curious as to why there was a dip in the number of new Steam releases in 2019...

I dislike the whole "reduced-hours" concept. It implies that this engineer would be working in an area that does not require cutting edge innovation and can get away with working 3 or 4 days per week while others work 5 or 6.

I think the job title should just be SDE; work times and other logistics can easily be clarified in the job description or during a phone screen.

I love electronics and electrical engineering in general. But this article explains perfectly why it is less accessible than software engineering - the former requires a individual components, soldering iron(s), a digital oscilloscope, etc. while all you need is access to a computer with internet for the latter.

Actions like this do nothing to repair the growing rift between conservative and liberal groups.

Makes sense, thanks. Appreciate the feedback.

Do you feel like this is because you have more than enough money now? Or is it because you now place greater emphasis on growth and a sense of fulfillment?

A friend of mine told me that those who did this are actually repeat offenders who kept getting released from jail due to a no-bail policy. Shows you the current state of the criminal justice system in LA.

Compensation stagnation may be a component, but it's likely only one of many factors. From what I can see, culture shift seems to be the primary driver of the Great Resignation.

Long COVID-like symptoms or not, the protection afforded by COVID vaccines still clearly outweighs the negatives. This holds true especially for at-risk populations.

The keyword here is _persistent_ as opposed to permanent. Significant bread consumption would lead to weight gain which would in turn also lead to a persistent but reversible impact on my overall health.

Keep it to a reasonable level, folks.

The office-to-WFH transition was fairly seamless for most engineering folk I know (outside of the psychological impact). On the other hand, many other roles such as product management and design benefit greatly from f2f meetings and whiteboarding sessions.

As a result, while I commend the overall increase in remote work, such roles should be thought out carefully as opposed to "let's offer remote work options just because everybody else is".

I am fairly certain this is a side effect of people sharing this link through messaging apps, thereby inflating Chrome (default browser for Android devices) and Safari (default browser for Apple devices) relative to Firefox.

I see what you did there.

I feel like the core premise of the article is wrong. Google's problem is not a lack of innovation, but the complete opposite - there's _too much_ of it. Great Google products come out very often but few get "long term" treatment. Part of it is engineer churn/turnover, but a bigger factor imo is culture.

> But scalable web services are, in 2022, a commodity available to all.

A quick aside - I feel most folks on HN can say that they have been, in some way, shape, or form, a part of this journey for the industry. Take pride in being part of a community that has made this a possibility.

This article takes clickbait to a new level.

There are also many (very smart) people who predict stagnation if the Nvidia deal succeeds.

Having worked at Intel before, I can safely say that it's difficult to boil down Intel's problems to a single issue (e.g. chip manufacturing). Stagnant company culture, limited growth opportunities within the company, and a fanatical focus on selling CPUs instead of innovating are all problems that Intel currently faces.

Having written a couple papers myself, I think it's entirely fair to release the code as is. Code written for research purposes is obviously not suitable for production but can still serve as a great tool for others to understand and build upon your work.