537 karma · joined April 14, 2023
Having kept them before, they are genuinely about as hard to care for as goldfish but need bigger tanks and a little bit more cleaning.
Also super easy to breed, we let the spawn hatch once and ended up with about 70 larvae, they cannibalize quickly but 6 grew to full size and we sold them on very easily.
But here it looks more of a bottleneck from my (admittedly naive) understanding.
But I have been recently using it for some tooling and small servers on personal projects where I'd have used python before.
Frankly it's been a joy and I wish I'd started earlier. The concurrency primitives are great, and the static binaries make deployment easy (raspberry pi in this case).
Struggle to use anything other than python professionally, the need to settle on a common denominator trumps pretty much everything else.
AI agents like this are trying to recreate personal intimacy I guess, which does feel like it might be different somehow.
Feels like trying to imagine the societal impacts of the internet in the early 90s.
Getting very realistic / real world conversational training data for an ai would be hard. Only a subset of us appear on podcasts, radio or tv and probably all speak in a slightly artificial manner when we do.
Reddit is not even a great source a lot of the time, just a way better start than a ton of marketing sites. Some type of search product that enables you to filter out results from companies trying to sell something would be awesome and an easily achievable benefit of AI.
I imagine an agent with this kind of conversational capability but better. And then I imagine that coupled with a video model that presents a proper talking head that is pretty much indistinguishable from a real one
Then what if these personalities were highly configurable to your tastes, had a very long memory and were so cheap and available you could build an entire customised social circle from them.
This is a bizarre nightmare for us but will become the norm for future generations. As rigid and inflexible as it is, my kids even had a hard time understanding Alexa is not real at first, and they have come accustomed to this style of conversational interface from a young age, I think about that but in 30 years for their kids.
With that tome on my desk it honestly didn't take much work to be able to do a better job than Dreamweaver using notepad and writing HTML by hand.
I never got into the advanced Dreamweaver features, am sure I remember loads of stuff around managing complete sites with it, etc. But from a purely front end developers perspective it always seemed superfluous.
In 20+ years of employment I think I came across one developer who used it, and that was probably in 2005 or so!
Find it hard to imagine what kind of team would use it for development today.
They lay it out quite clearly I think, but things like libavcodec are probably one of the few types of project where the benefits of assembly outweigh the lack of portability.
I'm not sure rust or zig's support for SIMD would be the project's first complaint either. Likely more concerned with porting a 25 year old codebase to a new language first.
Clearing your apt cache and removing extraneous packages is peeing in the wind when faced with GB worth of shared objects.
Not something we'd encountered before but seems earlier than these changes are meant to come into effect.
We've cloned the base image into ECR now and are deriving from there. This is all for internal authenticated stuff though.
Claiming they have access to 5x this amount is not such a bold claim?
The hardware, tooling and time required to do a LoRa fine tune like this are extremely accessible.
Financially this is also not a big expense and I assume would have cost in the order of $100s of dollars in GPU rentals, possibly less if you ignore experimentation time.
So what is a barrier to entry here? The data? Well they didn't have that either so automatically generated a dataset of just 500 examples to achieve the task.
I'm sure they spent some time on that but again it doesn't sound an incredibly challenging task.
It's worth realising if you've not delved into fine tuning llms before. In terms of time, scale and financial costs there is a world of difference between building a product like this and building a base model.
For most of us, our tech work will be long forgotten and obsolete 20 years from now. At best it will have provided some small intangible advance - hopefully for the better.
But the people that built my house died before I was born, yet their work has a tangible ongoing impact to this day.
The people who built some European cathedrals lived over 800 years ago, yet that padstone laid by some nameless apprentice still holds an entire functional building in place.
I don't understand what you're weighing this against? A job that is literally saving lives maybe, or really leading in a field of science or technology?
Most of us don't have that though, even here on hacker news. Most of us are part of a larger effort that will progress just as well without us, our personal impact is marginal at best.
I've worked in tech for two decades for a company I deem "moral" and I feel I've had impact. But I could have fitted kitchens or made wedding cakes for that time and had just as positive an impact on the world and people I serve professionally. Hell, if I was a carpenter my work could probably outlast anything I've done in tech.
With 20 years experience you'd get through our semi-automated processes and get an interview if the skill sets and employment history matched - regardless of a university level experience.
Are you sure it's the degree that's holding you back?
I'm not advising you lie, as it's fraud, but just adding that over here it would be incredibly rare to ask for proof of academic qualifications from 20 years ago when recruiting such an experienced hire.
I myself have around 20 years experience, I have a degree but god knows where the certificate is and additionally the university who issued it no longer exists.
Although my degree is adjacent somewhat it is not in CS. I've never been asked to show it to anyone, never asked a hire for theirs either.
Maybe with 50+ years there's a limited amount left to learn, but with my 25 I'm still picking things up regularly.
So it's usually more junior and less experienced developers who ask "how do you do that" - sometimes with wonder, sometimes genuine confusion.
My answer is never "it's simple" or "it's just" but always something along the lines of "it's a culmination of using the same tools consistently for many years", usually followed by "hey have you seen this trick too".
My answer to the fairly common question of "how do you do it" is always the same though.
I always relay that it's taken an evolving 20 year career to learn all that voodoo, I never imply it's simple to gain that knowledge nor do I encourage others to change their working patterns to match mine. I do like showing off a few tricks though!
One thing I do always mention though is the longevity of vim + the command line. I went through a period of swapping IDEs like most junior devs, now I have a setup that I feel can last me till retirement and I can just continually invest in learning to use the same tools better
Being new to all this home automation stuff I was quite intrigued how they worked though. They're exposed on the WiFi network over a really simple UDP based protocol which led me down a rabbit hole of writing a little go client to mess about with them, took a few evenings.
Not saying Xiaomi bulbs would be quite as simple to write an integration for, but they might have been. It's kind of fun seeing how people have reverse engineered all these custom protocols.
I was a web developer 25 years ago and for the majority of projects we only made static sites, all templated and minified. My skills are somewhat "of that time".
Recently I was asked to build a site to demonstrate some new software. It consisted of over 10k pages that once built would rarely, if ever be updated.
I just scripted and templated the generation of every single page hosted it in S3. This may sound ridiculous in this day and age but it takes a few minutes to rebuild and update the entire site.
Guess my point is I don't really find duplicating things to be a pain in the ass
I keep raising this but it's never prioritized as it means taking my time from developing our products
Essentially my manager then loses dev time to reduce a bill that due to institutional accounting practices they never actually see!
I suspect a lot of the huge AWS customers just eat this because it's so hard to mitigate.
If a business has hundreds of AWS accounts it becomes very hard to track, and if each account can only shave a few hundred dollars a month off their individual bill then there's very little impetus to actually do the work for the individual teams. Despite that possibly adding up substantial savings for the overall business.
Using AWS for smaller personal projects will always be more expensive and probably less fun.
On the other hand I recently had to run an ML model over hundreds of thousands of media files. I used AWS to launch 100s of GPUs using spot instances and complete the job in a few hours, then just turned it all off and moved on. It cost a few hundred dollars total.
In my mind it's at this kind of scale AWS really makes sense.
It's so slow and needs much trial and error to get a decent cut, I feel a more expensive machine would be much more versatile but cost 10x for the machine and consumables as you say.
With the similarly priced 3d printer I feel like I've actually saved a lot of money and hassle printing things. I can quickly download a model and print a perfectly useable "thing" in no time, often with no mistakes. Prusa slicer and a larger $5 nozzle has made things even better, even though I have an off brand 6 year old printer.
Lots of hobbies can get expensive though, especially when high quality tools are involved. I'd actually argue 3d printing is an exception to that.
Decent power tools and consumables rack up and don't even look at the price of hand tools from the likes of Lie Nielsen.
I make models for my 3d printer just a few times a year. I find it much easier to get back into openscad than remember my way around a gui or wrestle with more advanced tools.
There's nothing I've not managed to make with openscad yet, though the things I have made are embarrassingly simple (yet highly functional).
I identify with this. Have full time job, some disposable income, a few hobbies but no real time for them.
This has led to a lot of gear acquisition, buying tools or equipment that are beyond my level of expertise and I don't have time to make the most of.
I guess there's worse things to burn money on but does feel like if I had more time to actually do hobbies I'd discover I can get by with far fewer gadgets.