rsfern

22 ms·

301.
Neurolibre: A preprint server for reproducible data analyses(neurolibre.org)
1 points by rsfern 4y ago|0 comments
302.
by rsfern 4y ago
I think WSL is the official workaround There is also https://github.com/cloudhan/jax-windows-builder
303.
by rsfern 4y ago
I don’t think it has to be that precise, it just needs to be two tones with a slight frequency difference so you get interference coming and going to make a beat frequency. But i guess your brain does this “in software” since the sound does
304.
by rsfern 4y ago
Thanks for elaborating! > And running/testing the approaches and models is often still quite cumbersome due to different versions and so on. I definitely agree with this. I don’t know that I have any particular insight on infrastruc
305.
by rsfern 4y ago
Currently isn’t this is mostly also GitHub? with some functionality to load artifacts like model checkpoints from git LFS, figshare, public S3 bucket, or similar. I’m not super familiar sure huggingface, but that it seems similar to torchvi
306.
by rsfern 4y ago
Their NeurIps paper is all about scalable training/inference in ML systems that combine deep learning for perception and symbolic logic for e.g. visual question answering pdf: https://www.cis.upenn.edu/~mhnaik/pape
307.
by rsfern 4y ago
I saw it on Sunday, I think this is second chance pool in action There’s a PyTorch example here: https://scallop-lang.github.io/tutorial.html#section-10
308.
by rsfern 4y ago
Awesome, thanks! I think I found a full preprint on the authors site (pdf) https://www.cs.utexas.edu/~swarat/pubs/PGL-049-Plain.pdf
309.
by rsfern 4y ago
The sum-of-NIST-digits CNN+logic example was really interesting: https://scallop-lang.github.io/tutorial.html#section-10 Anybody familiar with the space have pointers to good introductory material on this field? Or terms to
310.
by rsfern 4y ago
Why? I haven’t seen a silver bullet yet, but I always learn something new
311.
by rsfern 4y ago
I’m curious if they set out to stabilize this particular sulfur allotrope, or if they had more of a serendipitous “hmmm… that’s weird” data analysis moment. It’s really interesting because there could be so many ways to stabilize interestin
312.
by rsfern 4y ago
DOI link: https://doi.org/10.1038/s42004-022-00626-2 The micrographs of the nanofiber composite electrode are cool!
313.
by rsfern 4y ago
To add some complementary points to what others have posted, poster sessions are great for undergraduate students and junior grad students to cut their teeth on presenting research in a lower pressure environment. Plus networking — the post
314.
by rsfern 4y ago
Yes (materials science) - my absolute favorite poster sessions are at Gordon research conference in my sub field, but the big meetings still do the arena style poster sessions, mostly with conventional printed posters The worst is trying to
315.
by rsfern 4y ago
That’s not how I understand the idea of anti-racism, at least from what I’ve read by people like Ibram X Kendi; I’m not really sure what you’re responding to about fundamentalism Kendi frames it like this: it’s a contradiction to say “I’m n
316.
by rsfern 5y ago
Your comment is glossing over the fact that high energy photons cause damage through a different mechanism not available to lower energy photons. A higher dose of lower energy photons may cause damage still, but not in the same way Ionizin
317.
by rsfern 5y ago
You don’t strictly need an institutional affiliation, but it makes things easier for sure. Arxiv has an endorsement process you can go through to post without an affiliation https://arxiv.org/help/endorsement
318.
by rsfern 5y ago
This is cool. I’m wondering what the responsive part is — the example seems like it’s just latex2html output basically? I think it would be super cool to have more https://distill.pub style papers, with like interactive plots in
319.
by rsfern 5y ago
To be fair, a lot of these news sites instaplay videos and make it really annoying to actually read the text. Why reward that kind of behavior with ad revenue?
320.
by rsfern 5y ago
Cool! I hadn’t seen the series before. Seems like it predates the book actually, which I wasn’t really expecting I haven’t found a streaming link yet, but these (related?) clips look pretty interesting https://www.bbc.co.uk/
321.
by rsfern 5y ago
Materials Science: Stuff Matters, Mark Miodownik https://archive.org/details/StuffMatters Very accessible and fun to read, and the book is structured around introducing a lot of fundamental materials science concepts i
322.
by rsfern 5y ago
> You want to beat Python? Build a type system that can capture dimensional analysis. Warning: it won't be easy. Curious about your thoughts on pint and Unitful.jl — pint doesn’t really go all the way to a full type system, and Unit
323.
by rsfern 5y ago
If Russia is responsible for the missing materials it could easily be part of a propaganda campaign. This specific scenario has been discussed a lot in the last few weeks
324.
by rsfern 5y ago
I agree with most of what you wrote here, but the article discusses two separate labs, one of which was looted and is now missing some radioactive material. There’s not much detail as to specific materials and quantities, and it seems like
325.
by rsfern 5y ago
The “communications” branding has more to do with the prestige level and expectation of faster-paced short research notes. Usually opinion pieces are explicitly filed under a “perspective” track (vs review and original research track). The
326.
by rsfern 5y ago
Good, we have some insane district maps in Maryland. I don’t understand why we can’t use more fair processes for deciding on district boundaries. A bunch of states have absurd district maps (Texas stands out in particular on the other side
327.
by rsfern 5y ago
Why not link to the Times article on your second point? It’s more nuanced than you are making it seem. They’re just reporting that some of the emails in the email dump are authentic, not confirming the entire affair https://www.n
328.
by rsfern 5y ago
You linked to an editorial in the wall street journal in another thread, but it seems like you didn’t actually read the times piece. The times confirmed that some of the emails in the email dump from the alleged laptop are authentic, that’s
329.
by rsfern 5y ago
I agree that it’s important to understand the risks and ride accordingly. I do wish the US would prioritize cyclists and pedestrian safety more though > engineers have spent a lot of time trying to minimize the amount of energy a human a
330.
by rsfern 5y ago
Another commenter linked a nature article with the top 100 papers by citation, and they start by pointing out that a bunch of Nobel-winning work didn’t make the list I think the Nobel list would be a different sort than your 25 must read li

More ›