There are several free alternatives to OpenAI that use the same API; which would make it possible to substitute OpenAI for one of those models in this extension. At least on paper. There is an open issue on the github repository requesting something like that.
So, it's not as clear cut. The general approach of using LLMs for this is not a bad one; LLMs are pretty good at this stuff.
Yes, but the API at the end is providing the core functionality. Simply swapping out one LLM model for another - let alone by a different company altogether - will completely change the effectiveness and usefulness of the application.
The premier open weight models don't even comparatively perform well on the public benchmarks compared to frontier models. And that's assuming at least some degree of benchmark contamination for the open weight models.
While I don't think they're completely useless (though its close), calling them fantastic replacements feels like an egregious overstatement of their value.
EDIT: Also wanted to note that I think this becomes as much an expectations-setting exercise as it is evaluation on raw programming performance. Some people are incredibly impressed by the ability to assist in building simple web apps, others not so much. Experience will vary across that continuum.
Yeah, in my comparing deepseek coder 2 lite (the best coding model I can find that’ll run on my 4090) to Claud sonnet under aider…
Deep seek lite was essentially useless. Too slow and too low quality edits.
I’ve been programming for about 17 years, so the things I want aider to do are a little more specific than building simple web apps. Larger models are just better at it.
I can run the full deepseek coder model on some cloud and probably get very acceptable results, but then it’s no longer local.
One would hope, that since the problem these models are trying to solve is language modeling, they would eventually converge around similar capabilities
Their interfacing software __is__ open source; and, they're asking for your OpenAI api key to operate. I would expect / desire open source code if I were to use that, so I could be sure my api key was only being used for my work, so it's only my work that I'm paying for and it's not been stolen in some way.
My older brother who got me into coding learned to code in Assembly. He doesn't really consider most of my work writing in high level languages to be "coding". So maybe there's something here. But if I had to get into the underlying structure, I could. I do wonder whether the same can be said for people who just kludge together a bunch of APIs that produce magical result sets.
> But if I had to get into the underlying structure, I could.
How do you propose to get into the underlying structure of the OpenAPI API? Breach their network and steal their code and models? I don't understand what you're arguing.
> How do you propose to get into the underlying structure of the OpenAPI API?
The fact that you can’t is the point of the comment. You could get into the underlying structure of other things, like the C interpreter of a scripting language.
I think the relevant analogy here would be to run a local model. There are several tools to easily run local models for a local API. I run a 70b finetune with some tool use locally on our farm, and it is accessible to all users as a local openAI alternative. For most applications it is adequate and data stays on the campus area network.
A more accurate analogy would be, are you capable of finding and correcting errors in the model at the neural level if necessary? Do you have an accurate mental picture of how it performs its tasks, in a way that allows you to predictably control its output, if not actually modify it? If not, you're mostly smashing very expensive matchbox cars together, rather than doing anything resembling programming.
As an ancient imbedded system programmer, I feel your frustration… but I think that it’s misguided. LLMs are not “computers”. They are a statistics driven tool for navigating human written (and graphical) culture.
It just so happens to be that a lot of useful stuff is in that box, and LLMs are handy at bringing it out in context. Getting them to “think” is tricky, and it’s best to remember that what you are really doing is trying to get them to talk as if they were thinking.
It sure as heck isn’t programming lol.
Also, it’s useful to keep in mind that “hallucinations “ are not malfunctions. If you were to change parameters to eliminate hallucinations, you would lose the majority of the unusual usefulness of the tool, its ability to synthesise and recombine ideas in statistically plausible (but otherwise random) ways. It’s almost like imagination. People imagine goofy shit all the time too.
At any rate, using agentic scripting you can get it to follow a kind of plan, and it can get pretty close to an actual “train of thought”facsimile for some kinds of tasks.
There are some really solid use cases, actually, but I’d say mostly they aren’t the ones trying to get LLMs to replace higher level tasks. They are actually really good at doing rote menial things. The best LLMs apps are going to be the boring ones.
I think the argument is that stitching things together at a high level is not really coding. A bit of a no true scotsmen perspective. The example is that anything more abstract than assembly is not even true coding, let alone creating a wrapper layer around an LLM
Nah, Hunminjungeum (The initial blueprint), which initially designed for phonetic alphabet did support differentiating those phonemes.
It's just until early 1900s that a "nationalist" (SiKyeong Ju) decided to use the "phonetic alphabet" into a primary writing system and revamped the system to be focused on words to be identifiable, rather than following the pronunciation.
hangul now is just like equivalent of using a modded Phonetics alphabet that made each words identifiable. It's not representing the actual pronunciation, nor the actual pronunciation of the word. That's also the reason why the hanja (= kanji or "Chinese characters" or whatever your country calls it) was around until late 90s.
If you think Korean grammar is crazy, It's all thanks to him.
> hanja (= kanji or "Chinese characters" or whatever your country calls it)
(Also, for reference, we say "Chinese characters" because that's what every country calls them. Hanja is just the Korean reading of 漢字. Kanji is the Japanese reading of 漢字. 漢字 means "Chinese characters".)
I can't tell what you're saying "nah" about. The comment I responded to said that, to a Western ear, [pap̚] vs [bap̚] is a "very big difference" that isn't noted by the Korean writing system, because Korean cares more about aspiration [of stop consonants, presumably] than voicing. It was in English.
I observed that English shares that quality with Korean, and English speakers are not even able to hear the difference between [pap̚] and [bap̚], making this an odd choice to exhibit as a "very big difference" which will, by its lack of representation in the writing system, confuse English-speaking Westerners. It's a difference they can't perceive; why would they be confused over the fact that two identical sounds are both written the same way?
Well, That's only after late 90s. Before that, hanja (equivalent of kanji in Japanese) was used in news, official documents and etc. alongside the hangul. You need to know hanja to say out what it read
Actually that is mandated by Korea Financial Telecommunications & Clearing Institute (consisted of banks).
Looks like it's mandated just to play a ping-pong game of "who is the culprit of this data leak" that usually concludes into user's improper installation of "Security Software".
1. "Cross-Connect" fees when you connect 2 or more networks into any of your device (paid separately of ISP subscription)
2. Monopoly of "3 ISPs" that protected by law (No newcomers can join)
This all brings up to up-to 100 times expensive internet than Europe in Korea.
(not Residential, but for CPs. Residential internet services are cheaper, but if you operate any servers your service will be terminated.) fyi, I kinda estimate the twitch should pay upto 300K USD/mo. (if they got lucky and got "domestic business deal") to just make sure their service up and running
If you connect two or more telecom lines into your rack or any service, you need to pay "Cross-Connect" fees. This is paid separately even if you brought your network line by your own. I can't say how much price is it since it is mostly trade secret with the sales dept of the telecom. but I'd say its a lot.
Native Korean here, It's way worse here. It has been a decade since former telecom board members in the chief of Korea Communications Commission, which should prevent this in the first place.
The Netflix trial back in 2021 beforehand was due to SK Broadband (South Korean ISP) having lots of NTT bound traffic since they didn't join the Netflix's OpenConnect Program while requesting Netflix to co-locate in their server main content servers in (the SK Broadband). The results? Netflix found guilty for causing those traffics.
For your information, You can't establish an ISP in South Korea since the law forbids anyone other than "3 major telecom"s (SK Broadband, KT, LG U+). Regulations protecting telecoms and making sure CPs getting properly ripped off (Bonus point if you are not domestic business).
They were able to reduce the network price from $70m/yr to $12m/yr [1] (their revenue was around $220m/yr btw).
There is almost no IPv6 yet since we still have enough IPv4 addresses. I am not sure how it is implemented but p2p has been often used in Korean services since 20 years ago.
