What I relly dislike about these LLM is how verbose they get even for such a short, simple question. Is it really necessary to have such a lobg answer and who's going to read that one anyway?
Maybe it's me and may character but when human gets that verbose for a question that can be answered with "drive, you need the car" I would like to just walk away halfway through the answer to not having to hear all the universes history just to get an answer. /s
The verbosity is likely a result of the system prompt for the LLM telling it to be explanatory in its replies. If the system prompt was set to have the model output shortest final answers, you would likely get the result your way. But then for other questions you would lose benefitting from a deeper explanation. It's a design tradeoff, I believe.
My system prompt is default - "you are a helpful assistant". But that beyound the point though. You don't want too concise outputs as it would degrade the result, unless you are using a reasoning model.
Well, when I asked for a very long answer (prompt #2), the quality had dramatically improved. So yes, longer answer produces better result. At least with small LLMs I can run on my GPU locally.
I just read it as "we were stuck and nobody got something done at all" to "we at least got something deployed". Going from zero to something mathematically speaking a very large increase (like infinite?).
By that logic, and looking at the data [1] you should come to the conclusion that one should remove the death penalty, because Europe has less homicides compared to the US and does not have a death penalty. So there are actually less murders without death penalty.
Obviously, that conclusion is flawed, because the data also shows otherwise in some countries I guess - and just using the death penalty as data point is useless. There are way more factors in play that lead to murder.
An interesting thing I notice now is that people do not like companies that only post about outages if half the world have them ... and also not companies that also post about "minor issues", e.g.:
> During this time, workflows experienced an average delay of 49 seconds, and 4.7% of workflow runs failed to start within 5 minutes.
That's for sure not perfect, but there was also a 95% chance that if you have re-run the job, it will run and not fail to start. Another one is about notificatiosn being late. I'm sure all others do have similar issues people notice, but nobody writes about them. So a simple "to many incidents" does bot make the stats bad - only an unstable service the service.
Personal experience: Did some cloud stuff for SME, and later on started colocation. I think my learning curve for all the cloud-stuff was the same as for all the colocation stuff, except the cloud will not get you rid of firewalls, NAT, DHCP and all that stuff. Cloud isn't that much easier, it's just a little bit different. IMHO, the largest disadvantage of colocation is that it requires (sometimes) physical presence at a datacenter.
IANAL, but as fas as I know when you're importing it from China, you are subject to local laws (and may pay the fine for importing a ccopy of a trademarked product).
IMHO, this whole thing could be read with "human" instread of "agent" and would make the exact same amount of sense.
"We put 200 human in a room and gave them instructions how to build a browser. They coded for hours, resolving merge conflicts and producing code that did not build in the end without intervention of seniors []. We think, giving them better instructions leads to better results"
So they actually invented humans? And will it come down to either "managing humans" or "managing agents"? One of both will be more reliable, more predictable and more convenient to work with. And my guess is, it is not an agent...
> It is not yours anymore if you can't uninstall stuff.
But the article is literally instructions on how to uninstall it…also you can just uninstall Windows entirely.
Like I get where you’re coming from, but let’s not pretend that Windows PCs are iPhones now just because of Microsoft’s annoying dark patterns (patterns they’ve been following for years before Copilot came out).
There is really no need to split hairs here, Windows is definitely to be compared to iphones when it comes to how much you are in control.
The instructions described in the article will stop working with the next Windows update, as copilot usage is tied to middle manager #78's KPIs at Microsoft.
They could - and some of the 3rd party vendors did: There is a 1Password 7 and a 1Password 8. There was also a Things 1/2, which is now a Things 3. it usually works by creating a new app, and not updating the old one anymore.
Maybe it's me and may character but when human gets that verbose for a question that can be answered with "drive, you need the car" I would like to just walk away halfway through the answer to not having to hear all the universes history just to get an answer. /s
reply