phreeza

10 ms·

91.
by phreeza 2y ago
Mentioned in the article, this was considered too boring!
92.
by phreeza 2y ago
https://youtu.be/8Zd0clYzgj4?si=7T-NXAXfK0IxJ_kG Google lens identifies it without problems. Edit: auction link https://aukro.cz/top-stav-pragotron-preklapeci-hodiny-cpj-06...
93.
by phreeza 2y ago
I think so, yes.
94.
by phreeza 2y ago
A biologist once told me "enzymatic degradation of PET is like the fusion power of biotechnology".
95.
by phreeza 2y ago
I understood the comment as saying the fortran code maybe used some sort of inefficient numerical scheme and/or some inaccurate approximations? Doesn't seem completely outlandish that more modern methods could help there? Of cours
96.
Cyanobacteria could help sequester carbon from oceans and factories(wyss.harvard.edu)
1 points by phreeza 2y ago|0 comments
97.
by phreeza 2y ago
Do they not respect the crawl-delay directive in robots.txt?
98.
Thousands go to fake AI-invented Dublin Halloween parade(euronews.com)
1 points by phreeza 2y ago|1 comments
99.
by phreeza 2y ago
I gave it a quick spin and my initial impression is much worse than perplexity.
100.
by phreeza 2y ago
I guess this goes to show, nobody really has a moat in this game so far. Everyone is sprinting like crazy but I don't see anyone really gaining a sustainable edge that will push out competitors.
101.
by phreeza 2y ago
What do you do with it?
102.
by phreeza 2y ago
Ancient equivalent of clickbait?
103.
by phreeza 2y ago
By orthogonal-enough dimensions, do you mean vectors whose dot product is close to zero?
104.
by phreeza 2y ago
I haven't worked with LLMs enough to know this but I wonder: are they nonsensical in a truly random way or are they just nonsensical on a different axis in task space than normal humans, and we perhaps just haven't fully internali
105.
by phreeza 2y ago
One thing I have found myself wondering is why this didn't play out similarly with Google in the early days. I guess their secret sauce was just so good and so secret that neither established players nor copycat startups were able to r
106.
by phreeza 2y ago
Yea there are much better examples of quality history podcasts, that are non-rambling. E.g. Mike Duncan podcasts (Revolutions, History of Rome), or the Age of Napoleon podcast. But even those are really just very good digestions of various
107.
by phreeza 2y ago
It wasn't meant as a nitpick, I was curious. The other reply was actually interesting.
108.
The Mathematics of Dobble (2018)(petercollingridge.co.uk)
20 points by phreeza 2y ago|2 comments
109.
by phreeza 2y ago
Same reason they offer deals in the first place, price discrimination and yield management.
110.
by phreeza 2y ago
Did Cortez really land in Darien? I thought that was the site of the ill-fated Scottish attempt at setting up a colony.
111.
by phreeza 2y ago
It's a two edged sword though in my experience, because it can lead to self-exploitation. I found it much easier to separate work from private life in jobs where I didn't personally identify with the outcomes.
112.
by phreeza 2y ago
Yea I agree it shouldn't be on you as a customer, was more asking out of curiosity. I don't think Google scholar has this fully solved either, I've seen many misattributed papers there.
113.
by phreeza 2y ago
How would you propose making claiming possible without the risk of hijacking/misrepresentation?
114.
by phreeza 2y ago
Maybe they can launch with SpinLaunch.
115.
by phreeza 2y ago
I think all this ideological stuff is completely unrelated to the issue we started talking about. You can fine tune a large model on whatever data you want, and so you can also fine tune it on the most recent user inputs. You can do unsuper
116.
by phreeza 2y ago
The mechanism for fine-tuning and the original training is exactly the same (gradient descent on weights). The effects you describe are results of what exactly is used to fine tune on.
117.
by phreeza 2y ago
What about fine-tuning?
118.
by phreeza 2y ago
Numbers I found are that global SAF production in 2023 was 6e8 liters, which corresponds to about 4000 transatlantic flights. So not a huge amount but also not nothing. Edit: ah I missed the "electricity alone" part. That I don&#x
119.
Solar will get too cheap to connect to the power grid(climate.benjames.io)
105 points by phreeza 2y ago|149 comments
120.
by phreeza 2y ago
There is nothing stopping you in principle to keep learning on every interaction in an LLM. Practical benefits are likely small though.

More ›