rsfern

6 ms·

61.
by rsfern 7mo ago
Not a lot, but what is your point exactly? There are a lot of really Chinese scientists working in the US, and the ones who are postdocs and research scientists at NIST are apparently being pushed out at the end of this month. They’ve alrea
62.
by rsfern 7mo ago
This list of high risk countries is not new (with the exception of maybe Venezuela being recently added, I’m not sure). Researchers with these citizenships have faced extra security review before joining NIST for years, and last year the la
63.
by rsfern 9mo ago
Your first reply was insightful, but this one is not a thoughtful take. Power consumption and emissions are already increasing, and any regulatory changes in 2025 are not factored in to discussion of those numbers. It’s more interesting to
64.
by rsfern 11mo ago
I think this is not quite the right analogy. A better analogy is procedurally generated music, because that’s what model-generated music is. But just like with LLM code generation, the input to the program is natural language (or maybe mult
65.
by rsfern 1y ago
I think it’s just scale to the moon rhetoric, like “what if we used 100x more compute?”. Since the units are power and not energy, I’m going with 10 GW continuous load (for training? inference?) but I think it’s not exactly meant literally
66.
by rsfern 1y ago
I think this is just loose terminology, instead of squaring they should have said “multiply by the complex conjugate”, which is what you do to quantum mechanical wavefunctions to obtain real-valued probability amplitudes
67.
by rsfern 1y ago
This paper “were RNNs all we needed?” explores this hypothesis a bit, finding that some pre-transformer sequence models can match transformers when trained at appropriate scale. Though they did have to make some modifications to unlock more
68.
by rsfern 1y ago
Are there particular libraries that make your setup difficult? I just manually set the index and source following the docs (didn’t know about the auto backend feature) and pin a specific version if I really have to with `uv add “torch==2.4”
69.
by rsfern 1y ago
I don’t think so? the double descent phenomenon also occurs in linear models under the right conditions. My understanding of this is that when the effective model capacity is exactly equal to the information in the dataset, there is only o
70.
by rsfern 1y ago
This is already what the funding agencies do! The merit review process solicits outside expert assessment of the importance, feasibility, and potential impact (including economic development and societal impact) of the research, and the fun
71.
by rsfern 1y ago
This is all awesome, but a bit off topic for the thread which focuses on AI for science The disconnect here is that the cost of iteration is low and it’s relatively easy to verify the quality of a generated C program (does the compiler issu
72.
by rsfern 1y ago
“Kill the [model] for trying” kind of sounds like using reinforcement learning to get models to behave a certain way
73.
by rsfern 1y ago
The article in Soft Matter is open access: https://doi.org/10.1039/D4SM01490A They have some interesting analysis of the elastic deformation that happens during the rolling process (as opposed to the ball just falling
74.
by rsfern 1y ago
In the case that it’s a human posting sarcastically, wouldn’t that be a false positive?
75.
by rsfern 1y ago
This is sort of a misguided take IMO - as if LiDAR and other sensor streams are somehow not large scale data while video is The bitter lesson isn’t fundamentally even about data. the key ingredient is computation (which does scale with dat
76.
by rsfern 1y ago
Same - I’ve been using einops and jaxtyping together pretty extensively recently and it helps a lot for reading/writing multidimensional array code. Also array_api_compat, the API coverage isn’t perfect but it’s pretty satisfying to wr
77.
by rsfern 1y ago
I’ve served as a reviewer for a couple of NSF panels, and one of the things I really liked about the program I reviewed for is that a lot of the proposals included collaborations with local trade and vocational schools to involve and train
78.
by rsfern 1y ago
In the case of the two year old who was removed with their mother, they have US citizen family that the father is trying to transfer custody (it seems he doesn’t have legal status, but I haven’t seen a definitive source) PDF: https:/&
79.
by rsfern 1y ago
That’s a strawman argument that I would never advocate, and completely ignores my question. Alternatives include arranging legal custody for the child and to stay in the US with a relative (as one family was attempting), or finding a legal
80.
by rsfern 1y ago
One thing I don’t understand is how this is even a choice the parents have the legal right to make, assuming their US citizen children do not have passports (I don’t know if the answer to that is publicly known). Can a child legally be take
81.
by rsfern 1y ago
The habeas petition for VMS (the two year old) indicates the father (who was not detained at the time of the filing) transferred provisional custody rights to a US citizen relative, and that communications with the mother (who was removed a
82.
by rsfern 1y ago
No, that is a false dilemma. the right (and constitutional) thing to do is give all these people the due process and access to legal representation that they are entitled to, and work out a legal solution to all these conflicting concerns.
83.
by rsfern 1y ago
CHIPS incentive funding is way bigger than just Intel, so it’s a bit disingenuous to write off the whole program just because of one (or even several) high profile bad actor. We should have a nuanced discussion and fix the shortcomings of o
84.
by rsfern 1y ago
NSF has actually been experimenting with this sort of funding model for a couple years through its new (as of 2019) convergence accelerator program, which I think is awesome. It’s explicitly a multi-sector program where academics partner wi
85.
by rsfern 2y ago
I don’t disagree, but I think GP means someone who is not a political appointee, like the non-partisan national security officers whose role is basically exactly what GP discussed. I don’t know if it’s usual for NSC officers to be included
86.
by rsfern 2y ago
It’s been illegal for Feds to strike for a long time, regardless of union membership. The federal employees unions have been pretty active in coordinating legal action though - IIRC both main lawsuits leading to reinstatement of the probati
87.
by rsfern 2y ago
This calculation critically depends on a linearity assumption that doesn’t seem at all justified with the large tariff rate changes being thrown about They just estimate the derivative of the import volume wrt the tariff rate, then solve a
88.
by rsfern 2y ago
What is “standard Moore’s law”? People say that Moore’s law has been dead for 20 years because that’s when the industry hit the power wall and had to pivot from simply reducing transistor size to parallelism and these other engineering stra
89.
by rsfern 2y ago
So first I think I might need to apologize for some jargon collision - my background is mostly material simulation, and when I say “quantum simulation” I mostly mean using classical algorithms to solve the quantum mechanical wave equation d
90.
by rsfern 2y ago
GPGPU is definitely mainstream for large scale quantum and molecular simulation. Quantum computing might help speed up electronic structure calculations, but my impression is that it’s still in its infancy To give a sense of the scale of th

More ›