Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
tedivm
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
24 ms
·
331.
▲
by
tedivm
3y ago
Not only that but they actually provide real support. I once complained on Twitter that I was having trouble with an EKS issue, and the product manager for EKS reached out to me directly, put me in contact with one of their engineers, and t
332.
▲
by
tedivm
3y ago
People really need to understand that Google is a garbage company that can't be relied on for anything. I've spent years as a consultant, and several more in the startup world, and I'd rather host all my content in us-east-1
333.
▲
by
tedivm
3y ago
If it was a free app sure, but in the paid category those numbers drop quite a bit.
334.
▲
by
tedivm
3y ago
My wife went through this for about a year (pre covid). She went from being extremely active to sleeping for 16 hours a day and still being exhausted. She got tested for just about everything, and all they could find were signs that her bod
335.
▲
Show HN: Easy Multiprocessing in Python with QuasiQueue
(github.com)
2 points
by
tedivm
3y ago
|
1 comments
336.
▲
by
tedivm
3y ago
I built QuasiQueue to make it easier to create multiprocess projects in Python. I found dealing with the multiprocess library and queue to be a bit difficult by itself, when it really shouldn't be. With this library you define "wr
337.
▲
by
tedivm
3y ago
That's just about the silliest interpretation of my comment that could have been made.
338.
▲
by
tedivm
3y ago
One of my former coworkers at Rad AI always used to joke that most problems people try to solve with AI could be solved better with simple linear regression, and I really do think that input was one of the reasons why that company is doing
339.
▲
by
tedivm
3y ago
The people pushing RTO have to ignore a huge amount of data in order to make their point.
340.
▲
by
tedivm
3y ago
> Have you worked in startups though? Organizations of 1-20 people where the whole company doesn't know who its customers are, why they would buy the product, or how to make money? That sounds like an amazingly poorly run startup wi
341.
▲
by
tedivm
3y ago
I think it depends on how much you expect your model to be used, and how quickly the model needs to react. The higher either of those becomes the more likely you'll want to specialize. If you expect your model to be used a lot, and you
342.
▲
by
tedivm
3y ago
I guess a question to ask is "What is GPT-4". Is it the algorithm, the weights, the data, or a combination of them all? To put it another way, the researchers at Rad AI consumed every paper that was out there including very cuttin
343.
▲
by
tedivm
3y ago
Was it actually a Chrome bug though? It only happened when the Grammerly desktop app was installed. My guess is grammerly is doing something sketchy.
344.
▲
by
tedivm
3y ago
This was something we thought about a bit at Rad AI. I think one of the problems this paper ignores isn't whether you can get a general purpose model to beat a special purpose one, but whether it's really worth it (outside of the
345.
▲
by
tedivm
3y ago
In December of 2012 AWS admitted that four files were lost: two completely deleted and two more truncated.
346.
▲
by
tedivm
3y ago
Their contract with OpenAI gives them unlimited access to the data sets, the tooling (which is all on Azure and was built out with the help of Azure engineers), the models and their weights, and basically everything around it.
347.
▲
by
tedivm
3y ago
It really depends on what you're researching. Rad AI started with only 4m investment and used that to make cutting edge LLMs that are now in use by something like half the radiologists in the US. Frankly putting some cost pressure on r
348.
▲
by
tedivm
3y ago
> It seems like HN is melting down. Almost literally- this is the slowest I've seen this site, and the number of errors are pretty high. I imagine the entire tech industry is here right now. You can almost smell the melting servers.
349.
▲
by
tedivm
3y ago
I'm assuming it's a combination of researchers, data scientists, mlops engineers, and developers. There are a lot of different areas of expertise that come into building these models.
350.
▲
by
tedivm
3y ago
This was handled so very, very poorly. Frankly it's looking like Microsoft is going to come out of this better than anyone, especially if they end up getting almost 500 new AI staff out of it (staff that already function well as a team
351.
▲
by
tedivm
3y ago
She didn't get booted from the company, but they did find a new interim CEO (the former twitch CEO).
352.
▲
by
tedivm
3y ago
https://github.com/opentofu/opentofu/blob/b8e2fad8e24cdd98c5...
353.
▲
by
tedivm
3y ago
No, it's a superset of Terraform. The developers of OpenTofu have been adding every new feature in, and are committing to version parity. They've already got the new testing framework built out for example, as well as the changes
354.
▲
by
tedivm
3y ago
Even in that blog post, where you get so close, you do this- > We want to be able to continue to develop Sentry and Codecov as single source Open Source projects You're stilling calling it "Open Source", but are calling it
355.
▲
by
tedivm
3y ago
OpenTofu is a superset of Terraform. They maintain complete compatibility with Terraform while adding additional features on top of it. They have more dedicated developers on the project than Hashicorp does, and they are building out a fair
356.
▲
by
tedivm
3y ago
Except those same companies then run into hidden limits that exist on Terraform Cloud and hadn't been previously mentioned. Honestly, even bigger companies are better off going with Spacelift as it can scale extremely well, has reasona
357.
▲
by
tedivm
3y ago
That's some real deflection. While the overall thread was certainly about CodeCov, the questions I asked your head of open source where very clear and not tied to that specific case, and their responses were as well. I do admit it'
358.
▲
by
tedivm
3y ago
> But if you believe pursuing additional licensing mechanisms means we're destroying OSS, there is probably little I can do to disavow you of that notion. Speaking of disingenuous, every time the sentry people come in here to talk i
359.
▲
by
tedivm
3y ago
That's their goal. The head of sentry's open source policy has openly stated that he, and sentry as a whole, want to change the definition of open source so that it has additional limits that make it easier to profit off of and co
360.
▲
by
tedivm
3y ago
Sorry, fixed it.
More ›