RE: syntax
For casual use, I kinda always liked the whole MATCH/AGAINST syntax for old school Innodb, though obviously things have changed a lot since those days. But it felt less like calling embedded functions and more like extending SQL’s grammar.
Regarding the rest, it seems like a reasonable approach at first tinker.
I really like the spirit! This sort of concept was one of the first things I wanted to explore with a spiritually adjacent project (cataclysm). It's been a bit, so I need to take time to update the python JIT generation now that we have much better coding models, but I'm a big fan of this sort of abstraction. https://github.com/Mattie/cataclysm
As most of us here can see, for many tasks now you don't really need to worry that you have the exact right syntax. I think you still need expert precision when it matters immensely, but we all develop tools, scripts, layers and the like that manual precision isn't necessary.
Don't think so, but there were some guesses on 3.5-turbo-- i.e. training a much smaller model on quality questions/answers from GPT-4. Same tactic worked again and again for other LLMs.
I'm definitely curious on the context window increase-- I'm having a hard time telling if it's 'real' vs a fast specially trained summarization prework step. That being said, it's been doing a rather solid job not losing info in that context window in my minor anecdotal use cases.
I've had to move to using Azure OpenAI service during business hours for the API-- much more stable unless the prompts stray into something a little odd and their API censorship blocks the calls.
Ah, really curious if it'll balk at the age old problem or just answer based on the code provided since it does inspect the nearby code to understand context for the generated code.
> It returns factually incorrect data, and it returns code with subtle but important errors if you ask it anything that's not regurgitated a thousand times in the training dataset.
To be fair, that's what pretty much every person does. The bar does seem pretty high if we need more than that (especially if not specifically trained on a topic). It's not a universally perfect expert servant, but I've been exploring the code generation of GPT4 in detail (i.e. via the 'cataclysm' module I just posted about). In 1 minute it can write functions as good as the average developer intern most of the time.
We're keeping score in a weird way if we're responding quickly with it needing to "code without subtle but important errors". Because that's the majority of human developers, too. I've been writing code for 30 years, and if you put a gun to my head, I would still have subtle but important flaws in every first typing of any complex generated code.
I'm not saying you're bashing it, by the way, I get your point, but I do worry a bit when the first response is citing that the SOTA models get things wrong in 0-shot situations without full context. That's describing all of us.
Now that it's released, it's time to experiment with what it can do.
from cataclysm import doom
# App gets the img file from the command line and saves it as a new file at half size with _half appended to the name
doom.resize_app()
Turned out to be all that's needed for a command-line file resize app (with PIL installed).
Yeah I wrote my own plunkylib (which I don't have great docs for yet) which is more about having the LLM and prompts in (nestable) yaml/txt rather than how so many people hard code those in their source. I do like some of the features in langchain, but it doesn't really fit my coding style.
Pretty sure there will be a thousand great libraries for this soon.
Here's what we need to be thinking instead of getting defensive or extra-optimistic:
If high-end security firms and fairly diligent government agencies were infiltrated, why would we think that smaller dev toolchain organisations not founded as security organisations will somehow be less likely to be targeted and become a vector for introducing supply chain attacks. Sophisticated attackers will go after any soft underbelly or pore they can find, and there's no reason not to believe they'd put significant effort into quietly abusing Jetbrains security just like they did with Solarwinds. I'm less worried about the "Russian" red scare mention other than it may give non-US organizations a few more opportunities to inject badness that we can't get visibility on.
Bottom-line is that it would be the holy grail and are we treating it as the high priority target it is? Having worked for dev tool companies in the past, I know they are a lot more worried about innovation than about their own internal processes.
I really would love it, especially if they add export mechanisms so I won't lose my data if they ever decide to die. Burned once already, unfortunately.
No, it's not a good thing beyond perhaps raising awareness of it and maybe temporarily increasing demand for it. The stability it had was a great sign and it was seeing some solid use for transactions (especially for virtual goods). The sudden increase is going to add risk to day to day transactions and thus make people shy again from using BTC for purchases.
Regarding the rest, it seems like a reasonable approach at first tinker.