Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Stable Audio Open

stability.ai

Post reply on HN
Print view View original

61–70 of 137 posts

Re: Stable Audio Open

#61

Post by cthalupa »

kimixa wrote: post #58
Workaccount2 wrote: post #24

Earlier quoted context omitted.

The idea that AI trained on artist created content is theft is kind of ridiculous anyway. Transformers aren't large archives of data with needles and thread to sew together pieces. The whole argument is meant to stifle an existential threat, not to halt some illegal transgression. If they cared about the latter a simple copyright filter on the output of the models would be all that's needed.

I think the definition between "Lossy Compression" and "Trained AI" is... vague according to the current legal definitions. Or even "lossless" in some cases - as shown by people being able to get written articles output verbatim. While the extremes are obvious, there's a big stretch of gray in the middle. A similar issue occurs in non-AI art, the difference between inspiration and tracing/copying isn't well defined e…

Has anyone been able to actually get a verbatim copy of a written article? The NYT got a ~100 word fragment made up of multiple snippets of a ~15k word article, with the different snippets not even being in order. (The Times had to re-arrange the snippets to match the article after the fact)

I am simply not aware of anyone successfully doing this.

Re: Stable Audio Open

#62

Post by rfoo »

pa7x1 wrote: post #59
rfoo wrote: post #55

Earlier quoted context omitted.

Sure, friends also won't let friends skip the fact that circulating supply of ETH is now decreasing instead of increasing. Also, only ~30% tokens are staked. The 30% who chose to stake essentially tax the other 70% in use. Each of the validator do the same amount of work (ok, strictly speaking you get to do more when you have more ETH staked, but being a validator is cheap and does not cost significantly more energy…

> Sure, friends also won't let friends skip the fact that circulating supply of ETH is now decreasing instead of increasing. This changes absolutely nothing of the calculation. Furthermore, the change in circulating supply last year was of 0.07%. > Also, only ~30% tokens are staked. Correct. > The 30% who chose to stake essentially tax the other 70% in use. There is something called opportunity cost. With the existen…

> Participating in staking is fully permissionless, stakers are not taxing non-stakers. They are being remunerated for their work.

That's just a more polite way to say tax. Being permissionless is cool, but it's still tax in my dict.

> There is something called opportunity cost.

And, who is going to be able to have a larger percentage of their funds staked, a poor or a whale? You need a (mostly) fixed amount of liquidity to use the thing.

> Incorrect. A staker does proportionate amount of work to its stake.

Apologies, I edited my original reply which should answer this.

In short, I don't see anything preventing me to run 10000 validators with 32 ETH each with very similar cost to running just one. It's certainly not linear.

Re: Stable Audio Open

#63

Post by cthalupa »

QuantumGood wrote: post #38
Workaccount2 wrote: post #24

Earlier quoted context omitted.

The idea that AI trained on artist created content is theft is kind of ridiculous anyway. Transformers aren't large archives of data with needles and thread to sew together pieces. The whole argument is meant to stifle an existential threat, not to halt some illegal transgression. If they cared about the latter a simple copyright filter on the output of the models would be all that's needed.

NY Times v OpenAI and Microsoft says the opposite, that verbatim, large archives of NY Times articles were retrieved via API. This may or may not matter to how LLMs work, but "large archive" seems accurate, other than semantic arguments (e.g. "Compressed archive" may be semantically more accurate).

> NY Times v OpenAI and Microsoft says the opposite, that verbatim, large archives of NY Times articles were retrieved via API.

This does not match my understanding of the information available in the complaint. They might claim they were able to do this, but the complaint itself provides some specific examples that OpenAI and Microsoft discuss in a motion to dismiss... and I think the motion does a very strong job of dismantling that argument based on said examples.

https://fingfx.thomsonreuters.com/gfx/legaldocs/byvrkxbmgpe/...

Re: Stable Audio Open

#64

Post by washadjeffmad »

treesciencebot wrote: post #19

This looks like the one that got leaked a couple weeks ago, so i guess they decided its better to open source at this point after the leak [0]. [0]: https://x.com/cto_junior/status/1794632281593893326

It is. The model.ckpt from petra-hi-small matches the official HF repo.

SHA256: 6049ae92ec8362804cb4cb8a2845be93071439da2daff9997c285f8119d7ea40

Re: Stable Audio Open

#65

Post by GaggiX »

jrm4 wrote: post #57
Workaccount2 wrote: post #24

Earlier quoted context omitted.

The idea that AI trained on artist created content is theft is kind of ridiculous anyway. Transformers aren't large archives of data with needles and thread to sew together pieces. The whole argument is meant to stifle an existential threat, not to halt some illegal transgression. If they cared about the latter a simple copyright filter on the output of the models would be all that's needed.

I fail to see how the argument is ridiculous; and I'll bet that a jury would find the idea that "there is a copy inside" at least reasonable, especially if you start with the premise that "the machine is not a human being." What you're left with is a machine that produces "things that strongly resemble the original, that would not have been produced, had you not fed the original into the machine." The fact that there…

Having exact copies of the samples inside the model weights would be an extremely inefficient use of space, and also it would not generalize, unless it generated a copy so close to the original that it would violate copyright law if used, I wouldn't find it very reasonable to think that there is a memorized copy inside the model weights somewhere.

Re: Stable Audio Open

#66

Post by hecanjog »

Uehreka wrote: post #11

> The new model was trained on audio data from FreeSound and the Free Music Archive. This allowed us to create an open audio model while respecting creator rights. This feels like the “Ethereum merge moment” for AI art. Now that there exists a prominent example with the big ethical obstacle (Proof of Work in the case of Ethereum, nonconsensual data-gathering in the case of generative AI) removed, we can actually have…

I also highly doubt anyone who signed agreements to have their music included in the Free Music Archive would have been OK with this. The particular type of license was important to contributors and there's a difference between allowing for rebroadcast without paying royalties and allowing for derivative works... I don't really care to argue the point, but it's why there were so many different types of licenses for the original FMA. This just glosses over all that.

Re: Stable Audio Open

#67

Post by neuralRiot »

Workaccount2 wrote: post #37
JohnKemeny wrote: post #29

Earlier quoted context omitted.

I think you should read the case material for NY Times v OpenAI and Microsoft. It literally says that within ChatGPT is stored, verbatim, large archives of NY Times articles and that they were able to retrieve them through their API.

..which makes no sense. It is either an argument of ignorance or of purposeful deceit. There is no coherent data corpus (compressed or not) in ChatGPT. What is stored are weights that create a string of tokens that can recreate excerpts data that it was trained on, with some imperfect level of accuracy. Which I agree is problematic, and OpenAI doesn't have the right to disseminate that. But that doesn't mean OpenAI d…

> It's not illegal for me to read an NYT article and write my own summary of the article's contents on my blog. This has been true forever and has forever been a staple in new content creation.

It’s not that clear-cut. It falls into the “Fair use doctrine”The cose 107 of the US copyright law states that the resolutiodepends on>

> (1) the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes; (2) the nature of the copyrighted work; (3) the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and (4) the effect of the use upon the potential market for or value of the copyrighted work.

Another thing we need to consider is that the law was redacted with the human mind limitations as a unconcious factor, (i.e not many people would be able to recite War and peace verbatim from memory). This just brings up the fact that copyright law needs a complete re-think.

Re: Stable Audio Open

#68

Post by kimixa »

cthalupa wrote: post #61
kimixa wrote: post #58

Earlier quoted context omitted.

I think the definition between "Lossy Compression" and "Trained AI" is... vague according to the current legal definitions. Or even "lossless" in some cases - as shown by people being able to get written articles output verbatim. While the extremes are obvious, there's a big stretch of gray in the middle. A similar issue occurs in non-AI art, the difference between inspiration and tracing/copying isn't well defined e…

Has anyone been able to actually get a verbatim copy of a written article? The NYT got a ~100 word fragment made up of multiple snippets of a ~15k word article, with the different snippets not even being in order. (The Times had to re-arrange the snippets to match the article after the fact) I am simply not aware of anyone successfully doing this.

The amount of content required to call it a "Copy" is also a gray area.

Same with the idea of "prompting" and the amount required to generate that copywritten output - again there's the extremes of "The prompt includes copywritten information" to "Vague description".

Arguably some of the same issues exist outside AI, just it's accessibility, scale, and lack of a "Legal Individual" on one side complicates things. For example, if I describe Micky Mouse sufficiently accurately to an artist they reproduce it to the degree it's considered copyright infringement, is it me or the artist that did the infringement? Then what if the artist /had/ seen the previously copywritten artwork, but still produced the same output from that same detailed prompt?

Re: Stable Audio Open

#69

Post by TaylorAlexander »

Uehreka wrote: post #11

> The new model was trained on audio data from FreeSound and the Free Music Archive. This allowed us to create an open audio model while respecting creator rights. This feels like the “Ethereum merge moment” for AI art. Now that there exists a prominent example with the big ethical obstacle (Proof of Work in the case of Ethereum, nonconsensual data-gathering in the case of generative AI) removed, we can actually have…

I’m so happy to see this! I’ve been saying for a while, if they focused on sample efficiency and building large public datasets, including encouraging Twitter and other social media sites to add image license options and also encouraging people to add alt text (which would also help the vision impaired!), they really could build the models they want while also respecting creatives, thus avoiding pissing a bunch of people off. It’s nice to see Stability step up and actually train on open data!

Re: Stable Audio Open

#70

Post by immibis »

Workaccount2 wrote: post #24
Uehreka wrote: post #11

> The new model was trained on audio data from FreeSound and the Free Music Archive. This allowed us to create an open audio model while respecting creator rights. This feels like the “Ethereum merge moment” for AI art. Now that there exists a prominent example with the big ethical obstacle (Proof of Work in the case of Ethereum, nonconsensual data-gathering in the case of generative AI) removed, we can actually have…

The idea that AI trained on artist created content is theft is kind of ridiculous anyway. Transformers aren't large archives of data with needles and thread to sew together pieces. The whole argument is meant to stifle an existential threat, not to halt some illegal transgression. If they cared about the latter a simple copyright filter on the output of the models would be all that's needed.

What's good for the goose is good for the gander. It may or may not be like theft, but either way, if one of us trained an AI on Hollywood movies, you best believe we'd get sued for eleventy billion dollars and lose. It's only fair that we hold corporations to the same standard.
Post reply on HN