extasia

They also made this mistake in the guardian article, I emailed them to correct it!

Just bought the Fukuyana book, it sounds fascinating!

I can personally recommend deepinfra. Been using it the past few days and it's served me very well.

A short blogpost on quitting YT addiction

BERT is ~100m parameters. GPT3.5 is 1000x bigger. A Bert sized decoder model would likely generate nonsense

For most applications BERTs output is used to fine tune additional NN layers, eg for text classification

Lifting heavy objects and returning them to their original place. (lifting weights!)

Because DL is 90% Matrix multiplication?

Research code has different values. If you throw away 90% of the code you used after a week you'd code differently too!

The tricky part is that 10% of your code will become the basis of your thesis / postdoc;)

This will probably be something that the reviewers bring up. Arxiv papers haven't been peer reviewed yet!

Drop some commitments. If you're pursuing a PhD there's no reason to have a job and side hustles, focus on your research, a PhD isn't something to half ass. Drop one of the other two. That's my 2c!

Of course knowledge was persisted in written form, but it was not industrialised. I would imagine progress is dominated by a small number of individuals, if information is dispersed across society broadly, your chances of those capable individuals having access to the knowledge requisite to advance the field is much higher.

How many geniuses did not advance science because they were ignorant of the current state of the art?

Well for one there was no printing press via which to persist knowledge and translate it cheaply through time and space.

Imagine what an individual could achieve technologically today, if they were unable to access any materials or knowledge that was accrued over the history of our species. The result would probably be a subset of the things you describe as originating from before 250 years ago, but they would certainly not be splitting atoms or mapping the human genome.

You're describing commercial tech companies in a capitalist economy. If you were familiar with other industries I'm sure you'd see similar themes, it's just you're exposed to techs evils by your proximity to the field (my assumption ofc!).

To add some hope, work backwards and think of a field that needs technology. Do those companies seem so bad as well?

Teachers getting 120k per year? Where is this happening?

The sea of fertility tetraology by Mishima. The Sailor who went to sea is one of his shorter works if you're looking to give him a read with less commitment.

What's Active Record and why is it an anti pattern?

I had two straps break on me too. You can get a replacement for a reasonable price on amazon.

40USD seems a little steep. How much do these things cost to produce?

How much would GPT cost?

Casio F91W watch. Cost about 10gbp.

What's your expected usage, volume wise?

The cognitive dissonance is dripping off of these sorts of articles. I used to read 500+ Guardian articles per year but they're basically ragebait at this point unfortunately..

Thanks, yeah! Empathy is wonderful:)

Might be a regional thing but "my friend" sounds very condescending / patronising to me! I'd personally feel quite upset if somebody began a critique of my behaviour with that.

I guess "buddy" or "mate" might work though!

I agree with your last point. How do we fix this problem though?

I think it's being used endearingly, regardless of the person's gender.

I see your point though and can understand how it may make you feel as a transwoman.

How might one achieve the same effect with different phrasing?

To be contrarian, one could argue that "make it work" includes building in a performant way from the start. In my mind "make it fast" is about optimization!

Use a local LLM like LLaMa 2.

Try new things. Simple over everything is the key to production python code. Too many layers of abstraction is a massive footgun, especially if youre passing objects around a lot, due to dynamic typing etc.

Constantly circle back and refactor. Ruthlessly.

Same as in any lang, focus on data structures and algorithms (business logic is algorithmic) rather than the implementation, at least to start with. Get it working (usually easy in python, it's so dynamic!), make it good, make it fast. A good data model makes coding a joy. If your data structures suck every little thing will accrue friction. This is a red flag that you chose the wrong data model.

Source: 1000s of python hours in engineering and research roles

I find his stuff is oriented more to beginners. I'd recommend mCoding instead perhaps.

All of Hettingers python talks are pretty great.