Having multiple services talking directly to the same database is without a doubt an anti pattern, for >99.99 cases so that is why it is considered one. Functionality aside it increases complexity for managing throughout and makes security very difficult.
LLMs constantly show most promise for work that really has very little value add, and I wonder given just economics if this will guide behavior towards not even bothering with its applications. Starting with cover letters, and then towards arbitrary content generation. Some companies already started doing this, if a meeting is not really productive you don't need to go or can feel free to leave. Not sure how many people would even read follow up notes, if it's free I guess another thing people won't use.
Kind of disagree with this, which seems outdated given that it becomes easier and easier to implement best practices. Setting up CI/CD for a basic MVP is trivial, making it super easy to contribute makes iterations faster than just working on "product fit"
They don't have luxurious insides, but the overall design I find is still nice. Seems odd to spend resources and thought into making really neat interior design with dashboard and screen and then default to the cheapest materials for it.
... You seriously just tell people to deploy a stack with your IAM role that "has the least permissions possible" good luck lol. Another thin wrapper around open-ai that doesn't even do the heavy lifting (deploying a secure and trusted authenticated role to access resources that can be easily audited etc)
Again, just a trivial wrapper over open ai, which already does this very well. But seems to not have any real benefits other than not needing to move mouse. You don't need to send data to OpenAI, and in a business you wouldn't want to do that. LLMs already generate synthetic data for graphing that you just copy/paste in and use. It should instead be passing in types and header maybe.
The big thing for graphs is formatting, the ideal solution IMO is one that can JIT code gen parts that generate real code for formatting so you can reuse and tweak, I don't need to pay to send request to increase the marker size by one pixel. Even if running locally that's waste of compute
What do you mean separate authorization logic? There are many layers to auth and usually they act as interceptors in request that go very fast. If you have blanket permissions to list, you are able to list resources you have access to... that's trivial. However `Blog` resources might have explicit deny policies on them as well, so yes those are also evaluated. Not sure how else you'd expect it to work sans caching like current state of resources and access.
Amazing how people can be elevated to such high positions without any acumen or business knack. Almost all amazon devices have been a failure in terms of generating profit. Somehow this guy got to keep his job despite all that, instead of trying to nab someone at apple. That's why the layoffs in Alexa were so disruptive, Alexa never made any money, they always poured money into incredibly stupid ideas, and overnight decided they didn't want to do that anymore, despite it being the M.O. for like past 10 years. No idea why he didn't leave like Bezos or the other guy who headed retail right as the pandemic challenges arose.
CAPTCHAs are supposed to tell humans and robots apart, if robots are all doing better than humans, maybe we should flip the acceptance criteria to make sure you are not performing task at superhuman level, (until we train bots to do this). On an unrelated note, I have found captchas that don't even work and reprompt me all the time, I wonder if there is some naive filtering behavior they are applying.
Having some marine background the first sentence confused me, I didn't realize people use the crab mascot for rust so much to use evolutionary terms. But there is already a term for the "opposite" of carcinization, it's just decarcinization. This analogy can come full circle if the next variant of this language uses the coconut crab or a hermit crab as its mascot.
I find this type of writing compelling, so if you write a blog I think you would get people to read it. One thing I've seen after the fact is the schism between media and evidence which adds an extra dimension here with the air traffic controller's story. The post mortem investigation usually takes a long time, and by then the media has completely moved on. Reminds me of the Boeing MCAS incidents, which interestingly I had a few airline industry friends, and internally according to them almost incidents were due inadequate training of pilots, but no one would say that since there is a nationalistic aspect to it.
not really. Companies pay more in expensive cities where they locate offices because they need to attract local talent, and they have to compete against other companies. Many national companies have cost of living adjustments baked into their salary bands because of this.
This behavior of legal loopholes to discount workers should be made illegal. The definition of sub-contractor has got to change and/or have laws/policies in place that actually make sense in a sub contractor relationship. These are shell companies that spin up and exclusively deliver for amazon, they aren't gig workers delivering when they want, they have strict quotas.
I think people miss this point, Vegas is a spectacle. If they are going to do something it has to be an attraction not just a rote tool that serves a purpose. They literally want it to be a place you have to go to because you can't experience it anywhere else.
Using this as a basis for urban planning and study is bizarre. Zermatt is basically a resort, yes there are full time residences but dominated by winter ski population, (and it's served by very good Swiss rail services). As it's in a valley, it's quite small and physically restricted. The fact they banned cars is really not that interesting as a study for other potential places.
exactly, the most frustrating type of behavior for any colleague (esp jrs) is those who just default to consistently ask for help instead of diving deep. That behavior is costly in slowing down others. People need to learn how to learn, otherwise they will just be handicapping team if they cannot function independently. At the same time, navigating large proprietary or new knowledge sources can be tedious and sometimes impossible, but I have found that starting from the ask first mentality to be a positive feedback loop for never diving deep.
as an analogy, consider that renovating often is much more difficult and complex than new construction. As you build up you're making some irreversible moves, i.e. you build a door or a set of trusses, now you can't move a large piece through it anymore.
this is just PR newswire hype, calling it foundation model is a stretch. People have been training nets on satellite imagery for a long time, I don't even see anything here that would need a foundation model, most of them are vanilla CNNs that work relatively well, the problem has mainly been the data pipeline of different gsds and nadir of images which makes layering difficult.
That's not an LLM problem, there's a physical divide between a physical stores operation and whatever its representation online is. Google tries to do this by tracking people and activity in locations to see if people are at a place or something but reality is there is not a clean interface to that other than contacting someone at the location.
I didn't even read article, and my point was entirely fulfilled by your response and other comments here. There are plenty of places where things are sub optimal but the amount of attention received by this is disproportionate. All I am saying is an observation
HN seems to be chock full of people who have vendettas against so called "elite" schools probably because they didn't get in. Which is weird since there also is large number of people who were self taught and didn't even go to school and are very successful. These schools are private businesses, forget private universities being some about some bastion of knowledge and discovery of truth, their primary concern is funding ongoing operations. There are plenty of schools that can provide a similar education if that is what you are going for.
Exactly, I never said it was a good thing that housing is an investment, only that that is the current state of affairs, and to alter that would have ramifications that would cause huge problems. Best thing we can do is stabilize what we currently have and figure a way forward that doesn't worsen the problem, but we cannot go back.
Many people have mortgages on their home, being underwater on a mortgage (i.e. the LTV spikes because you are now borrowing more than the home is worth) would set off a bank/debt crisis, that is what I meant by catastrophic.
I think at this point that would be catastrophic, the average persons most valuable asset usually is their home (for better or for worse). A all encompassing reversal would economically cataclysmic.
I would think we need better incentives for new market entrants (first time home buyers) and penalties for n-private home owners who rent out.
they are already doing that with XLA, google has TPUs, amazon has tranium/inferentia. common interface in future that you basically just cast model to `.toDevice` of an enum of accelerated computing types seems to be the goal.
how can you finetune gpt-4 or 3.5? The shell/wrapper around a request to open AI is not any worthy piece of software, It'd be more impressive to vend those services. Otherwise it's basically just a few scripts.
This article was also from a week after he bought it.
The most obvious conclusion is that the number of bots on the platform has skyrocketed as they removed probably most the people working on bot prevention right as chatgpt and the like took off. You must use all outside indicators as proxy for twitters certainly declining usage, ad revenue is definitely down, price per ads probably will drop as the impressions metrics will now have thousands of bots that don't translate to CTR, and outside searches for twitter probably have gone down as well.