[HN Gopher] Magicoder: Source Code Is All You Need
___________________________________________________________________
Magicoder: Source Code Is All You Need
Author : tosh
Score : 54 points
Date : 2023-12-05 20:46 UTC (2 hours ago)
(HTM) web link (arxiv.org)
(TXT) w3m dump (arxiv.org)
| yawnxyz wrote:
| Github: https://github.com/ise-uiuc/magicoder
| thefak111 wrote:
| I can't fucking take it anymore.
| quickthrower2 wrote:
| Cola wars?
| kevindamm wrote:
| whack-a-mole with coder wars
| mnky9800n wrote:
| Dude you could be Harry Kim on the holodeck creating whatever
| you can think of. What's wrong with that?
| lainga wrote:
| We have to go through the Eugenics Wars first. I am NOT
| wearing that awful quilted whatever-it-was tunic the soldiers
| apparently wore.
| gumballindie wrote:
| But how if ai bros keep stealing ip to build their little
| models.
| stillwithit wrote:
| Take the laws that enrich them allow away from them.
|
| They're just people. Not divine mandates. There is zero
| real obligation to serve contemporary socio-political and
| economic norms.
|
| Not really finding any of this progress in computing
| shocking though. Unix and that model wastes a lot of
| resources dealing with strings.
|
| What a shock we reduce the amount of fluff to just enough
| symbolic logic to instigate appropriate electron state to
| solve a problem, rather than brag about a new DSL to
| tokenize, parse and logic to template in yet more asinine
| "file formats" ...once you remove the chimps chasing
| Shakespeare, superfluous state doing nothing but propping
| up a venture backed coder boot camp grads jobs we find
| super powerful software.
|
| Shocking, I'm shocked.
| yawnxyz wrote:
| Every time one of these come out I hope it DOES take all my dev
| jobs so I can focus on things that don't require writing code.
| Every time they fall really short.
| alfalfasprout wrote:
| alas all of the benchmarks are based around totally self
| contained problems.
| intelVISA wrote:
| I don't think it's possible in the strictest sense: beyond jr
| level none of us really 'generate code' as our main value
| add.
|
| That said, I think ML advances may possibly usher in the next
| gen of low-code tools that may liberate a large portion of
| web devs from being human LLMs.
| marmakoide wrote:
| I don't see this creating new algorithms (as in, not in the
| training corpus), but maybe giving the kind of answer you would
| expect from Stack Overflow, without all the social fluff around
| it (comments, badges and so on).
|
| The day one of these find new algorithms to solve problems with
| better complexity or simpler code that state of the art, I'll
| wake up. When I give a LLM a computational geometry problem,
| it's exactly like a student trying to bullshit his/her way
| through an exam without any actual deep understanding.
|
| For example, I ask for an algorithm to compute Laguerre Voronoi
| diagrams (usually not available in books or code examples), and
| I get answers for plain Voronoi diagrams, because it's what you
| will find in many books and code samples. Generating boring but
| necessary code, in moderation, is a win.
| femiagbabiaka wrote:
| ?
| Reubend wrote:
| I'm guessing this model was made to be small simply to keep costs
| low and make sure that they could feasibly train the model with
| the amount of time/effort they had. But to some extent I'm left
| wondering whether this technique would continue to be fruitful
| when scaled up to a huge model and with a bigger initial training
| set.
| gumballindie wrote:
| Stealing licensed code.
| fizx wrote:
| This looks like a llama2 finetune, so the dataset (inclusive of
| llama2) isn't fully open as claimed, and I'd still have to accept
| the Facebook and possibly OpenAI licenses.
|
| Let alone that clearly the base model was built on non-source-
| code, so their premise doesn't hold.
|
| Disappointing.
___________________________________________________________________
(page generated 2023-12-05 23:00 UTC)