blackbear_

9 ms·

61.
by blackbear_ 1y ago
And somebody doesn't know that the Arctic and Antarctica are two different places...
62.
by blackbear_ 1y ago
> Seems like you would def want to compare the two groups, so what did this study actually do? You are right. Indeed, all vaccines and drugs undergo clinical trials testing for safety first, efficacy then, before being approved for sale
63.
by blackbear_ 1y ago
> There are a lot of seemingly valid problems pointed out with it. As I said, if the problems are indeed valid then a new analysis will show it. Otherwise, it's all speculations, and that is simply not enough. Science isn't bui
64.
by blackbear_ 1y ago
By far the most correct way to contest a scientific study is to provide a new scientific investigation that reaches different conclusions. If those claims are true then I'm sure it won't be difficult to conduct a proper scientific
65.
by blackbear_ 1y ago
Curious if you have found any resources to learn more about this way of thinking?
66.
by blackbear_ 1y ago
In principle this is possible, modulo scalability concerns: https://arxiv.org/pdf/2506.06039 Perhaps this will one day become a new post-training task
67.
by blackbear_ 1y ago
Good thought, that indeed works: https://arxiv.org/abs/2310.02226
68.
by blackbear_ 1y ago
No because the number of cars in each simulation not fixed. There are 100,000 simulations, but each simulation runs until a car slower than the first appear.
69.
by blackbear_ 1y ago
Not at all. Research that appears useful is going to be picked up by others, and if it's really a fraud it will be exposed eventually. That is the check you are looking for and indeed how we realized there's some fraud and reprodu
70.
by blackbear_ 1y ago
Most people cannot tell those apart...
71.
by blackbear_ 1y ago
Statistical significance is not connected to real-world relevance. With enough resources one could know the exact number without any margin of error. This number could however be the result of random fluctuations rather than systematic effe
72.
by blackbear_ 1y ago
The "history" you cite only goes back three years. Meanwhile, the BLS publishes the monthly corrections since 1979, and the average correction since 2003 is +9k between first and third estimates [1]. Moreover, do note that all pub
73.
by blackbear_ 1y ago
> did it get harder to estimate, or was the estimator just not up to the task? It got harder to estimate, because less people reply to the survey (see sibling comment). They are aware of this and they could improve their analysis and dat
74.
by blackbear_ 1y ago
That is true, but he still had the final call. Whether he ignored those concerns, or seriously engaged with them and decided to take the risk nonetheless, it's really unfair to make excuses and shift the blame now.
75.
by blackbear_ 1y ago
The revisions were -133k for June and -120k for May, and yes the are unusually large. Generally, it seems that estimation post-covid became less accurate, with five (if I'm counting correctly) revisions of 100k or more since 2022, excl
76.
by blackbear_ 1y ago
> If Trump knew a quarter ago that his tariffs were taking employment, he could have done something about it. Absolute bullshit, because everybody was saying this all along. "More than 5.6 million jobs could be at risk due to Trump’
77.
by blackbear_ 1y ago
Your point is moot becase the published data also includes confidence intervals, reflecting uncertainty in the estimation. The implications of this should be clear to anybody with high school numeracy levels. Interestingly enough, the curre
78.
by blackbear_ 1y ago
You have to compute attention between all pairs of tokens at each step, making the naive implementation O(N^3). This is optimized by caching the previous attention values, so that for each step you only need to compute attention between the
79.
by blackbear_ 1y ago
Afaik zuoyezi (坐月子) also works, the zuo meaning "to sit".
80.
by blackbear_ 1y ago
Congrats! How are your knees doing if I may ask and how did you take care of them? I've only been running for a couple of years and already feeling troubles brewing in.
81.
by blackbear_ 1y ago
> these types of things are unavoidable The floodings or the tragedies? > The full risk profile of our planet is impossible to determine Was this really necessary to avoid this specific tragedy?
82.
by blackbear_ 1y ago
The procedure you are describing could absolutely lead to overfitting, but you wouldn't know for sure until you test the prompt on an independent dataset.
83.
by blackbear_ 1y ago
Source for the timeline? Nowhere in the article is it said that the construction happened within weeks.
84.
by blackbear_ 1y ago
Windows. Like, all of it. (I know, not the answer you were looking for, but somebody had to bring it up)
85.
by blackbear_ 1y ago
Maybe controversial, but if you already know how to program then doing a small project with a LLM is the fastest and most efficient way. Ask it to implement something, then go read the documentation to understand what is going on. After you
86.
by blackbear_ 1y ago
> AI models will never obtain the ability to completely replace “low value work" Maybe, but this is not the meaning of replacement in this context and it need not hold for the "economic" reasoning to work. All that matters
87.
by blackbear_ 1y ago
> I read the paper before I made my original comment. That's good, now I'm wondering about the others in the thread. > 'Omics papers get away with a lot of hand waving Making assumptions and interpreting results is part
88.
by blackbear_ 1y ago
Only on the surface (promising a rise to 5% in 10 years is meaningless in this climate), and only because Trump is a lunatic. This week’s NATO summit will be all about placating Donald Trump https://www.economist.com/europe&
89.
by blackbear_ 1y ago
Yes they did, and published it all. Sometimes I can't believe how low discussions on HN can fall. Did really nobody in this thread bother to check this? Are we fine disparaging research solely based on the fact that they used a method
90.
by blackbear_ 1y ago
While the theoretical bottleneck is there, it is far less restrictive than what you are describing, because the number of almost orthogonal vectors grows exponentially with ambient dimensionality. And orthogonality is what matters to differ

More ›