One single successful company can't sustain the tech economy of a continent.
Even if they lost the commercial aviation crown after Boeing's flops, the US still has far more local champions with insane margins successful in other critical areas (military, web services, mobile, cloud, GPUs, AI, CPUs, etc) than the EU could ever dream of.
I'm sure working at Airbus or ASML is nice but those two companies can't employ all of EU's tech workers and fund EU's monetary deficits, especially since both are registered in the NL tax heaven.
I don't accept that example. The Aerospace industry is dominated by the Americans [0]. Challenging North America is a much harder proposition than challenge the EU. The EU doesn't have a competitive edge in that market and I doubt they'd stand up all that well if challenged, given that they were already challenged and didn't stand up well.
Europe of today is nothing like the Europe of the 1970s.
In the mid to late 20th century, Europe has a deep & diverse industrial base. Public infrastructure projects like high speed rail were world leading. Western Europe had a military industrial base genuinely competitive with the US.
I recently did a public tour of one of Airbus' major assembly lines. I remember the tour guide telling us, that their customers (the airlines) either have their own QA people on the line when their planes get assembled, or pay another company to do it for them.
I would like to phrase this as “trust and verify”, because the state of trust arises from being open to verifiability, contrary to common misconception that they are against each other.
Meanwhile Boeing just inspects itself, or at least they did before they fired most of their inspectors. It's really no surprise that they now have the build quality of the average aliexpress product.
I spent a very significant amount of my waking time working. This is one of the best opportunities to make friends, and IMHO it's also nicer to work with friends than with non-friends.
Most of the friends I made as an adult came from a work related setting.
Yep, my first thought was also to define friend. As the joke goes, friends help you move, real friends help you move bodies. Few people in a lifetime will be real friends, and it doesn't matter how you end up meeting them.
When you inject friendship in a work relationship, you are also injecting mechanics that are not work-related and can ultimately lead to problems in the work setting.
You can be pleasant around co-workers but being all out friends is dangerous.
fully agree with this. Plus, if you're worried about "promotion competitiveness" souring friendships with people in your team, you can always make friends with people in other departments and meet them during lunch. This doesnt have to be a hard binary rule
At least in Germany, the UK, and Spain it's not (only) contractors, but Airbus itself. IIRC there are around 50k employees in Germany and France each, 13k in Spain and the UK each.
I would have said there is no problem with your style (nothing brash/abrasive), but you used a lot of jargon, that people who are not very deep into LLMs (large language models) would not understand. Interests of hackernews visitors are very diverse, not everyone follows LLMs that closely.
This was my take exactly. I read the original and thought, "Wow, this sounds like really interesting stuff this poster us excited about. I wish I knew what the terms meant, though. I'll have to come back to this when I have more time and look up the terms."
I was pleasantly surprised to find a glossary immediately following, which tells me it wasn't the tone of the post, but the shorthand terminology that was unfamiliar to me that was my issue.
I think writing in "Ben's voice" is great. There are just going to be times when your audience needs a bit more context around your terminology, that's all.
I haven't read a lot of LLM papers, but I believe this is a rather weak paper low on details (note: not the results achieved of the LLM, but the paper itself). If it had landed on my desk for a review, I probably would have sent it back just based on that.
For example, they never really say how they trained the experts or which dataset they used.
It’s becoming pretty common, yeah. The two things you mentioned: training particulars and dataset mixture are also basically the only competitive advantage companies have. Since the code/architecture is trivial to reproduce, anyone with enough money can make a competing model “easily”.
OpenAI started this trend and cemented it with GPT4’s “technical report” which didn’t even specify the number of parameters in the model. They’ve been historically vague about their dataset for far longer than that though.
Exactly, same thought. Actually I would expect, that they trained each expert separately and later together, since you need to train the router network as well. I'm far from an expert in LLMs. But this would be interesting to know, especially how different training setups influence the performance.
