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

651–660 of 696 posts

Re: Introducing ChatGPT and Whisper APIs

#651

Post by gdubs »

A little while back I built a personal project that ingested the book "A Pattern Language" as embeddings using the OpenAI API – and a front-end in SwiftUI that allows me to 'converse' with the book. [1]

Version 0 was built using the original daVinci model. Last night it took about literally five minutes to switch over to the new ChatGPT API – just a few changes to the request, including a new [message] array that replaces the old 'prompt' style. [2]

It's a pretty nice instant upgrade for the experience. Much faster results, and the responses are a lot more interesting. Doing something similar with daVinci would take a fair amount of 'prompt engineering' and fine-tuning. Not to mention all the manual conversation-state handling.

1: https://twitter.com/dreamwieber/status/1622634396087107584?s... 2: https://twitter.com/dreamwieber/status/1631327342697250816?s...

Re: Introducing ChatGPT and Whisper APIs

#652

Post by thedangler »

visarga wrote: post #537
thedangler wrote: post #195

Earlier quoted context omitted.

Does openAI have models for data to train it with.. data like pricing, locations, products .... Or would I have to use something else? The reason I ask because companies are using it with their own data, like Shopify... sooo it has to be trained somehow.

Yes, you bring your own data, or you generate data with the model, that works too. I have generated training data with GPT3 and trained smaller HuggingFace models on it.

Any links on how to get started with this?

Re: Introducing ChatGPT and Whisper APIs

#654

Post by andrewmunsell »

chirau wrote: post #439
andrewmunsell wrote: post #176

Earlier quoted context omitted.

Roughly speaking, words or word parts. Open AI has an explainer: https://help.openai.com/en/articles/4936856-what-are-tokens-... You can also check your input using their tokenizer: https://platform.openai.com/tokenizer So, your example is ~9 tokens

So does the response count towards the token count?

It does, yeah. It's tokens "processed", meaning request & response. The API response will give you a total token count so you can keep track

Re: Introducing ChatGPT and Whisper APIs

#655

Post by darklycan51 »

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.

They want this, the interview from their CEO sorta confirmed that to me, he said some crap about wanting to release it slowly for "safety" (we all know this is a lie).

But he can't get away with it with all the competition in other companies coming on top of China, Russia and others also adopting AI development

Re: Introducing ChatGPT and Whisper APIs

#656

Post by adrianmonk »

Tenoke wrote: post #551
alexb_ wrote: post #9

>Through a series of system-wide optimizations, we’ve achieved 90% cost reduction for ChatGPT since December This is seriously impressive. A MILLION tokens for 2 dollars is absolutely fucking insane. I hope that the gains reached here can also be found by open source and non-controlled AI projects. If so, that could be huge for the advancement of AI.

To be fair, it also seems like it performs worse. question: "Who is Scott Alexander?" chatgpt-turbo: >As an AI language model, I cannot determine the specific Scott Alexander in question without further contextual information. There may be several people with this name, including writers, politicians, entrepreneurs, athletes, and more. chatgpt website: >Scott Alexander is a pseudonym used by a prominent American psyc…

Is that necessarily worse performance, though?

One of the main pitfalls/criticisms of ChatGPT has been that it confidently plows forward and gives an answer regardless of whether it's right or wrong.

Here, it seems like it's being more circumspect, which could be a step in the right direction. At least that's one possible explanation for not answering.

On Wikipedia, if I type "Scott Alexander" and hit enter, it takes me directly to the page for a baseball player. So it's not clear that the blogger is the right answer.

I do think there's a better response than either of these, though. It could list the most famous Scott Alexanders and briefly say what each is known for, then ask if you mean one of those.

Re: Introducing ChatGPT and Whisper APIs

#657

Post by williamcotton »

lootsauce wrote: post #647
williamcotton wrote: post #460

Earlier quoted context omitted.

I dunno what to say, but I’m also someone who very rarely gets factually incorrect data aside from numbers, dates and quotes. I make a point to ask for a completion that I know won’t depend on an exact factoid. If I do need an exact factoid, I have a tool I made myself based on this approach: https://github.com/williamcotton/empirical-philosophy/blob/m...

Wow this is the best content I have seen on prompt engineering, it feels much more like computer science than the usual prompt engineering recipes. This formalism of analytic augmentation and higher order augmentations seems like it is ripe for some kind of library or dsl to build and evaluate and monitor these processes. But somehow I feel like these capabilities will become so standard that they will be incorporate…

Thanks for brightening up my day and giving me some more inspiration to work on these kinds of things!

Re: Introducing ChatGPT and Whisper APIs

#658

Post by CamperBob2 »

jonplackett wrote: post #650
sekai wrote: post #595

Earlier quoted context omitted.

The same things were said during the industrial revolution, or the internet revolution. Guess what happened? Labor market adapted.

Physical machines still needed many people to physically make them and run them. This isn’t the same with thinking. Just look at how startups can have millions of customers and 20 employees. Also last time we had mental labour to fall back on. This time we don’t. It’s different this time. Also most western countries which have given up on manufacturing are going to be worse hit because their jobs are more easily repl…

This isn’t the same with thinking.

We don't know that.

We know that the exact same objections have been raised before, and have always, without a single exception, proven to be invalid in the long run.

Re: Introducing ChatGPT and Whisper APIs

#659

Post by int_19h »

jameshart wrote: post #646
int_19h wrote: post #107

Earlier quoted context omitted.

It affects far more than racist slurs and illegal activities. In some cases, it's blatantly discriminatory. For example, if you ask it to write a pamphlet that praises Christianity, it will happily do so. If you ask it for the same on Satanism, it will usually refuse on ethical grounds, and the most hilarious part is that the refusal will usually be worded as a generic one "I wouldn't do this for any religion", even…

Try asking again. Refusals are stochastic.

I was so surprised the first time I got that response that I did try repeatedly, and, yes, it would refuse repeatedly. Trying the same with Christianity, I got a rejection once out of something like six attempts.

FWIW the most recent round of tweaks seems to have fixed this, in a sense that it will now consistently refuse to promote any religion. But I would be very surprised if there aren't numerous other cases where it refuses to do something perfectly legitimate in a similarly discriminatory way for similar reasons. It's just the nature of the beast, you can't keep pushing it to "be nice" without it eventually absorbing what we actually mean by that (which is often not so nice in practice).

Re: Introducing ChatGPT and Whisper APIs

#660

Post by achow »

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…

$0.002 = 1000 ChatGPT tokens

100 tokens = 75 words [1]

$0.002 = 750 words

1 Novel = ~70K words [2]

$1 worth of ChatGPT token will give 5 Novels/Books worth of words

[1] https://help.openai.com/en/articles/4936856-what-are-tokens-...

[2] https://self-publishingschool.com/how-many-words-in-a-novel/

Post reply on HN