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

Introducing ChatGPT and Whisper APIs

openai.com

Post reply on HN
Print view View original

571–580 of 696 posts

Re: Introducing ChatGPT and Whisper APIs

#571

Post by wolverine876 »

nodemaker wrote: post #559
wolverine876 wrote: post #542

Earlier quoted context omitted.

It's a sort of philosophical idea - openness and free expression - taken to a logical and inhuman extreme. I cannot think of a situation where it is appropriate to say whatever I'm thinking. I think it would destroy trust, not least by demonstrating the unreliability of my judgment.

Here is a thought experiment for you. First think of the people you trust the most in this world, then imagine if they stopped speaking about whats on their mind with you. Would your trust in them increase or decrease?

They don't say nearly everything they think; nobody does.

Re: Introducing ChatGPT and Whisper APIs

#572

Post by pksebben »

ChildOfChaos wrote: post #239
samspenc wrote: post #207

Earlier quoted context omitted.

I thought Midjourney was better as well, until I saw some recent videos from Corridor Crew on Youtube. For those who don't know, this is a VFX studio in LA that tries to keep at the cutting-edge of video production techniques and posts content to their Youtube channel, and they have a massive number of followers and several viral videos. They recently created a full 7-minute anime using Stable Diffusion with their ow…

While this is cool this doesn't change my opinion at all. Each still image is still not that impressive. Good for them using the tech in a clever way but i don't find this that relevant.

those trees are part of a forest, you know.

the benefits of such fine grained control aren't a trick. it's why they were able to scrap together frames that don't jump all over the place (mostly).

the other benefit of such a broadly hacked upon model is that it grows in leaps and bounds.

All due respect to mid journey, but the stable diffusion hype is not just hype.

Re: Introducing ChatGPT and Whisper APIs

#573

Post by cavisne »

vishal0123 wrote: post #82
minimaxir wrote: post #2

> It is priced at $0.002 per 1k tokens, which is 10x cheaper than our existing GPT-3.5 models. This is a massive, massive deal. For context, the reason GPT-3 apps took off over the past few months before ChatGPT went viral is because a) text-davinci-003 was released and was a significant performance increase and b) the cost was cut from $0.06/1k tokens to $0.02/1k tokens, which made consumer applications feasible wit…

> I have no idea how OpenAI can make money on this. I did some quick calculation. We know the number of floating point operations per token for inference is approximately twice the number of parameters(175B). Assuming they use 16 bit floating point, and have 50% of peak efficiency, A100 could do 300 trillion flop/s(peak 624[0]). 1 hour of A100 gives openAI $0.002/ktok * (300,000/175/2/1000)ktok/sec * 3600=$6.1 back.…

Does openai actually specify the size of the model?

InstructGPT 2B outperformed gpt 3 175B, and chatgpt has a huge corpus of distilled prompt -> response data now.

I’m assuming most of these requests are being served from a much smaller model to justify the price.

OpenAI is fundamentally about training larger models, I doubt they want to be in the business of selling A100 capacity at cost when it could be used for training

Re: Introducing ChatGPT and Whisper APIs

#574

Post by pksebben »

practice9 wrote: post #104
Kurtz79 wrote: post #90

Earlier quoted context omitted.

Easier to use is often all that it takes. In Midjourney you get fantastic results just by using their discord and a text prompt. To get some similar results in Stable Diffusion you need to set it up, download the models, understand how the various moving parts work together, fiddle with the parameters, donwload specific models out of the hundreds (thousands?) available, iterate, iterate, iterate...

Sure, but if Midjourney outputs a low quality results for your prompt, they are going to be much more difficult to improve. It's a black box at this point. While with SD there can be multiple solutions for a single problem, but yeah, you have to develop your own workflow (which will inevitably break with new updates)

... along with like half the interfaces to pytorch, and whatever hacks you had to implement to get mps working.

But it's this kind of stuff that keeps me engaged. SD is truly a godsend to masochistic hacker types.

Re: Introducing ChatGPT and Whisper APIs

#575

Post by pksebben »

DoesntMatter22 wrote: post #421
OkGoDoIt wrote: post #124

Earlier quoted context omitted.

Do you have any recently updated examples, blog posts, whatever showing that DALLE is worse than modern stable diffusion? I was still under the impression that DALLE was better (with better meaning the images are more likely to be what you asked for, more lifelike, more realistic, not necessarily artistically pleasing), with the downside of it being locked away and somewhat expensive. And my understanding is that sta…

Dalle 2 was great initially but the SD BLEW past it. I mean way way way past it. Dalle2 is like a Model T Ford and SD is a Fighter Jet. It's that different. Dalle-2 is dead already.

given what they're up to these days, I think it's understandable that they'd leave dalle to gather dust.

Re: Introducing ChatGPT and Whisper APIs

#576

Post by qwertox »

graderjs wrote: post #530
qwertox wrote: post #293

Earlier quoted context omitted.

whisper.cpp has no GPU support. Models below medium aren't that good, and medium and large are pretty CPU intensive. A minute of audio on medium can take anything between 15 and 90 seconds to transcribe, when using 8 cores, while the service transcribes on the large model in less than 7 seconds.

For English only: This is so wrong! I actually didn't find that much significant improvement between Medium to Large. In many cases Medium was actually better for English than large was. Where Large really excels in my experience is with super noisy or distorted audio. Same thing with Small to Medium. There's no much difference in text quality. For good quality audio, Small is all you need. Medium is not much more ac…

Last time I used it was last week ago, when I was comparing it to a docker install of Whisper/torch.

Then again, I'm using it on very noisy audio recorded with a lavalier microphone while riding a bike.

For transcription tests of recordings from an answering machine medium was more than enough, not sure about small.

Also, for the recordings during a bike ride, sometimes medium is better than large.

All this was used in German language.

I don't have a Mac, but your approach with using VAD is interesting. I'll see if I can preprocess my files.

Re: Introducing ChatGPT and Whisper APIs

#577

Post by Doctor_Fegg »

komali2 wrote: post #520
kcatskcolbdi wrote: post #461

Earlier quoted context omitted.

Sure it would be called those things and then nothing would come of it. If a country uses morally compromised methods to win a war history just calls it winning the war.

That seems to be changing. I've seen an uptick in criticism against the usa for unnecessarily (according to top military advisors, experts, generals etc at the time) dropping the atom bomb on Japan for example.

Absolutely. The bombing of Dresden has been viewed as a mistake - verging on a war crime - in Britain for the last 20 or so years.

Re: Introducing ChatGPT and Whisper APIs

#578

Post by pksebben »

coldtea wrote: post #140
anonylizard wrote: post #42

Earlier quoted context omitted.

You are like 2 months out of date. Stable diffusion now has a massive ecosystem around it (civitai/automatic1111), that when used well, completely crushes any competitors in terms of the images it produces. Midjourney is still competitive, but mostly because its easier to use. Dalle2 will get you laughed out of the room in any ai art discussion.

> You are like 2 months out of date. Dalle2 will get you laughed out of the room in any ai art discussion. So, the field is so immature than things change completely every few months?

that's the crazy bit - this field is at like, toddler levels of maturity, and it's already producing mind-blowing results.

Re: Introducing ChatGPT and Whisper APIs

#579

Post by Silverback_VII »

whatshisface wrote: post #284
freeqaz wrote: post #236

Earlier quoted context omitted.

It's also worth mentioning that, because Microsoft is an investor, they're likely getting these at cost or subsidized. OpenAI doesn't have to make money right away. They can lose a small bit of money per API request in exchange for market share (preventing others from disrupting them). As the cost of GPUs goes down, or they develop at ASIC or more efficient model, they can keep their pricing the same and then make mo…

Their new AI safety strategy is to slow the development of the technology by dumping, to lower the price too much to fund bootstrapped competitors.

I highly doubt it. OpenAI, Google and Meta are not the only ones who can implement these systems. The race for AGI is one for power and power is survival.

Re: Introducing ChatGPT and Whisper APIs

#580

Post by birracerveza »

Cantinflas wrote: post #53
anonylizard wrote: post #42

Earlier quoted context omitted.

You are like 2 months out of date. Stable diffusion now has a massive ecosystem around it (civitai/automatic1111), that when used well, completely crushes any competitors in terms of the images it produces. Midjourney is still competitive, but mostly because its easier to use. Dalle2 will get you laughed out of the room in any ai art discussion.

It's amazing that "being two months out of date" in AI means that you are already a dinosaur

Make it two weeks. I haven't paid attention for a second and stuff like Controlnet pops up and evolves into Multi-Controlnet and then into MultiDiffusion.
Post reply on HN