pilotneko

He just ate the cost. It’s five dollars.

Honestly, discoverability for these kind of guidelines is very poor. I work in Healthcare (administrative/financial) and even with domain knowledge, it can be hard to find things on CMS’ website. For example, try and find the list of inpatient only procedures for 2023.

This seems…wrong? Doesn’t YouTube Music pay $0.008 per stream? According to this calculator, it’s much less? Is there some kind of nonlinear payout schedule for YouTube Music?

There are built-in styles. Personally, i find plt.style.use(‘ggplot’) at the top of my scripts to make a big difference.

Maybe kludge it with a dithered transparency mask?

I would suggest ImageJ, which is a Java-based image processing program developed at the National Institutes of Health (NIH). It has a very low memory footprint, built-in support for most DICOM file formats, a ton of extensions, and (of course) cross-platform support.

As an added bonus, you can run it as a portable app.

I think there might be some session leakage. I typed “A pig with a bowler hat.” and the model returned a picture of a half moon.

Well, Tennessee literally just split the 5th Congressional District, represented by U.S. Rep. Jim Cooper, D-Nashville, among three different outlying districts. Literally makes it 3X harder for a Democrat to represent Nashville.

https://www.tennessean.com/story/news/2022/01/12/republican-...

Austria is less Germanic than you think.

Nice work. You might be interested in this visualization of the first million prime numbers from UMAP: https://umap-learn.readthedocs.io/en/latest/exploratory_anal...

For-profit hospitals do not improve care. There is simply no way to remove money from patient care pathways and improve outcomes.

What course are you taking? Imagenet is only 150 GB, and Common Crawl is only 320 TB.

Big data is a moving target, but I’m comfortable defining it as data too large to fit in memory. Obviously, you can always get a bigger node, my rule is thumb is that if you need generators, you are working with big data.

For what it’s worth, I live in a no-income tax state (Tennessee). I didn’t move here, I just grew up here. We have some of the lowest property taxes in the country for metropolitan areas: I pay around $2,500/yr in property taxes for my home within the city limits of Nashville.

The downside is that the city has very little income for infrastructure improvement/upkeep and education.

It’s a Makerbot cupcake, which is pretty old by this point (~10 years). One of the first “affordable“ 3D printers.

https://makerbot.fandom.com/wiki/Cupcake

Strange to see dimensionality reduction as the root node? I’m having a hard time reading this diagram.

Very affordable too. I think we paid $100 for a perpetual license.

I know it’s not the point you were making, but McMinnville is rather close to the Arnold Engineering Development Complex, which is a pretty large Air Force research facility (massive wind tunnels). There are likely a least a few bright kids at those schools that are being negatively affected by this ban.

+1 for How to Read a Paper by Keshav. I have been using this method for the entirety of my academic and industry career. Now that I run a Data Science team, I have all my new hires read this as part of onboarding.

I'll just note that this pairs wonderfully with the Whitesides paper on Writing a Paper (https://gmwgroup.harvard.edu/files/gmwgroup/files/895.pdf), which is also required reading for my team members.

Well, that’s the thing, you can’t make a generic version of these drugs because you can’t fully characterize the structure. They are just way too big and complicated, which is scary because a single difference in a glycosolation site can induce an extreme immunogenic response.

Best the FDA can do is allow biosimilars and subject them to significantly more testing, which increases the cost 100x (compared to a traditional generic).

https://www.pfizer.com/sites/default/files/investors/financi...

Just noting that -mab therapeutics are not synthesized directly, they are produced in cloned cell lines. It’s practically impossible for one pharma company to replicate another companies -mab, since they will never have access to the exact cell line used to manufacture the original. Also, Rituximab hasn’t been fully characterized, hence the discovery of new uses. There is no such thing as generic when it comes to -mab therapeutics.

Point is, the cost to create Ocrelizumab may be legitimately higher than Rituximab.

In short, the author used yolo5 to count cars from webcam streams locations along highways in San Juan. I think it would have been easier to do background subtraction and count vehicles using standard vision techniques (e.g., threshold, watershed, count blobs), given that they had access to a fixed view security feed.

If you have less than 250 engineers, this guide is not for you? Strange.

Edit: Turns out I read it wrong, apologies. This guide is for companies with less than 250 engineers.

Pretty sure it’s DistilBERT, not DistilBART: https://huggingface.co/transformers/model_doc/distilbert.htm...

The problem has expanded. The service dashboard is now showing increased error rates in N. Virginia for Amazon DynamoDB, Amazon Elastic Kubernetes Service, and Amazon Elastic Load Balancing. Although the service health dashboard is not reporting it, this is affecting Fargate tasks as well.

Small consolation, but I believe GINA should prevent that behavior. https://www.hhs.gov/hipaa/for-professionals/special-topics/g...

It is being used in the reformer architecture to relate the dot product attention layer. This enables much longer input sequences and reduces computational complexity from O(n^2) to O(nlog(n)), where n in the length of the sequence.

https://huggingface.co/transformers/model_doc/reformer.html

Healthcare insurance companies and for-profit hospitals divert money from patient care. How is that supposed to improve patient outcomes?

Also, why would you side with insurance company and just assume their lower pricing is correct? You know they add overhead, and they are raking in premiums. The relationship between the payer and provider is adversarial, with the payer holding the stick.

The concern wasn't about ads, it was about patient privacy. You guys basically put together a search engine for medical encounters, allowing physicians to access medical records they shouldn't be able to: https://www.wsj.com/articles/behind-googles-project-nighting...

Even if we don’t worry about ads, Google doesn’t exactly have a good track record with patient privacy. Look at what you guys did with Ascension, allowing physicians to access medical records from other hospitals within the system based on similarity.

To be clear, just because something represents a common-sense workflow improvement or might be useful to the physician doesn't mean you can just go and do it. While it might be a good idea to pull similar cases (or counterfactual examples) from your vast set of patient records, it doesn't mean it's LEGAL. Medical records are not like case briefs, you can't just crack open someone's record without proper consent.

Might seem like a braindead point, but the website doesn’t actually state what MVP stands for. For sales and marketing, MVP might mean Most Valuable Player. For most developers, MVP means minimally viable product. Unfortunate name collision that could completely shift the tone of this article...