llm_trw

28 ms·

391.
by llm_trw 2y ago
The price of LLMs has fallen 1,000 times in the last year for the same quality tokens. It's not clear if video models will follow the same trajectory.
392.
by llm_trw 2y ago
Yes, but that's 2 to 3 times slower than what happens if you let the code rot.
393.
by llm_trw 2y ago
You better not have a long running batch job and also have x4 the capacity you usually use so you can do your daily compute during the period when electricity is cheap.
394.
by llm_trw 2y ago
You either waste your time updating daily or you rewrite from scratch every 3 years. JS is what happens when you let the inmates run the asylum.
395.
by llm_trw 2y ago
It was France avoiding the Spanish Civil war.
396.
by llm_trw 2y ago
It's been at least one dozen years ago since the last time it happened. No one from that period is still in the exact same position as when it happened - some of them were promoted.
397.
by llm_trw 2y ago
It's breaking news about a topic that came out of the blue for everyone. There isn't the time or knowledge of wtf is going on to write that article.
398.
by llm_trw 2y ago
>Not just link in France. France is in the midst of a government fall, but this is something built into how they do democracy and so a new one is expected to form in a few months on the backs of their democratic process. The longest a Fr
399.
by llm_trw 2y ago
The implication is that a rather large chunk of the technology sector - e.g. everything Samsung makes - is now under a very wide blanket of uncertainty.
400.
by llm_trw 2y ago
>He's declaring his liberal opposition are sympathizers with North Korea. It sounds very Trump-ian, to be honest. Where in that single sentence did you find a reference to a coup?
401.
by llm_trw 2y ago
That's nice. But it has nothing to do with what OP said.
402.
by llm_trw 2y ago
It wasn't Trump that spend the last 4 years calling everyone who disagreed with him a Russian Troll.
403.
by llm_trw 2y ago
Yes, prohibitively.
404.
by llm_trw 2y ago
It's amazing that we now have subscription medicine.
405.
by llm_trw 2y ago
>Meanwhile the Democrats have to spend every term fixing the things the GOP broke Like NAFTA. When it comes to making people lives worse the democrats are just as capable as the republicans. It's just that when it comes to helping p
406.
by llm_trw 2y ago
I hope you enjoy the next 40 years, becuae they will be the same as the last 40 years with people like you making the same excuses election after election.
407.
by llm_trw 2y ago
>Democrats have not had a filibuster-proof majority in the Senate since before Reagan Republicans have not had a filibuster-proof majority in the Senate since 1907.
408.
by llm_trw 2y ago
If only a democrat majority had been elected at any point in the last 40 years that could have fixed this.
409.
by llm_trw 2y ago
At some point you have to blame the other people who kept it going for 50 years now.
410.
by llm_trw 2y ago
Like always scheme has a solution to the problem. Symbols that contain space separated word are enclosed in ||, |this is a symbol|, that while the naked space is still the symbol separator token. Now if we can just remove all other punctuat
411.
by llm_trw 2y ago
Again, that's also true for copy/pasting anything from stack overflow.
412.
by llm_trw 2y ago
No and it won't do you much good even if I did. The tools have broken again since then - thanks tensorflow data loaders - and my code only works against a version of python that's no longer supported in LTS Ubuntu/Debian10+.
413.
by llm_trw 2y ago
The question is: how is more likely to make a stupid mistake and not spot it? Me or an LLM? My money is on me.
414.
by llm_trw 2y ago
Are you familiar with LaTeX? I was writing a lot of rather complex augmented matrices. Instead of having to manually type them in with all the formatting done by hand I literally gave ChatGPT a python nested list and told it what to subscri
415.
by llm_trw 2y ago
I ended up training a bert on nothing but python for the embedding search. The results were crap. Then I used an llm to write a new docstring for each class/function definition in the training data and the results were better than stat
416.
by llm_trw 2y ago
I'd be looking for a new job. People like OP are the reason why there is so much skepticism about the new generation of AI. >I just copy pasted all the answers I found on stack overflow into our live trading database at it works lik
417.
by llm_trw 2y ago
The desperate need for clean nuclear power the world over.
418.
by llm_trw 2y ago
The computing machines that the Manhattan project used were famously as fast as the average secretary doing the same calculations. Yet the computer age still happened. Ask yourself why.
419.
by llm_trw 2y ago
The number one predictor of human wealth is energy consumption. Since the stagnation of the west in the 1980s we have seen the same stagnation in energy consumption there.
420.
by llm_trw 2y ago
I just wrote a multilinear algebra research paper using natural language to generate all the latex. This used to take me several hour of tedious mind numbing boiler plate. I finished in in 10 minutes today. People who aren't on board w

More ›