i'm pretty excited to have access to the system prompt, that's going to be huge i think.
Introducing ChatGPT and Whisper APIs
431–440 of 696 posts
Re: Introducing ChatGPT and Whisper APIs
#432Earlier quoted context omitted.
This is explained in the OpenAI docs. There is a chat completion API and you pass in the prior messages from both the user and the assistant.
I used the same trick with the previous GPT3 API (da-vinci) and it worked well, I'd pass as one big prompt: User: hello (previous prompt) Bot: hi (previous response) User: who are you? (new prompt) Bot: (here it continues conversation) I wonder how the new ChatGPT API differs, other than the fact that it's structured (you use JSON to represent the conversation memory separately instead of one large prompt). I guess I…
Judging by this[0] the new structured format is immune to "injections":
[0] https://github.com/openai/openai-python/blob/main/chatml.md
Re: Introducing ChatGPT and Whisper APIs
#433Earlier quoted context omitted.
It is so massive that I can't help but think about what happened with Google Maps API a few years ago where they had extremely low pricing for years then hiked the price by 1400% once enough people were locked into applications based on that API.
That's exactly what's going to happen. Low prices now, wait until your business becomes dependent on it, then jack it up to whatever you need it to be.
Hard to see how businesses would become dependent on it, if it takes off then someone will run similarly capable models at a reasonable profit margin.
Re: Introducing ChatGPT and Whisper APIs
#434Earlier quoted context omitted.
FWIW, AssemblyAI has great trasncript quality in my experience, and they support streaming: https://www.assemblyai.com/docs/walkthroughs#realtime-stream...
We're using AssemblyAI too, and I agree that their transcription quality is good. But as soon as Whisper supports world-level timestamps, I think we'll seriously consider switching as the price difference is large ($0.36 per hour vs $0.9 per hour).
Both of those prices strike me as quite high, given that Whisper can be run relatively quickly on commodity hardware. It's not like the bandwidth is significant either, it's just audio.
Re: Introducing ChatGPT and Whisper APIs
#435Earlier 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You are like 2 months out of date.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.
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…
> 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),
“Artistically pleasing” is often what people ask for.
> with the downside of it being locked away and somewhat expensive.
Those are enormous downsides. Even if DALL-E was better in some broadly relevant ways in the base model, SD’s free (gratis, at least) availability means the SD ecosystem has finetuned models (whether checkpoints or ancillary things like TIs, hypernetworks, LORAs, etc.) adapted to... lots of different purposes, and you can mix and match these to create your own models for your own specific purposes.
A web interface backed by strictly the base SD model (of any version) might lose to the same over DALL-E for uses where the set of tools in the SD ecosystem do not.
Re: Introducing ChatGPT and Whisper APIs
#436Earlier quoted context omitted.
I am scared for all people working service jobs. Why? Because they're no longer doomed to eke out a meaningless existence doing a robot's job badly?
It’s better than having no job isn’t it?
No. The only way to a post-scarcity economy is by continuing to harness technology whenever and wherever possible.
There has been, and will be, no point in time at which the optimal strategy is "Stop" or "Wait" or "What about my job?"
Re: Introducing ChatGPT and Whisper APIs
#437Earlier quoted context omitted.
I checked the price of a A100, and its costs 15k? Is that right?
And $2.25 per hour on 1 year reservation means 8,760 hours x 2.25 = $19,710 rent for the year. Not a bad yield for the provider at all, but makes sense given overheads and ROI expected.
Cost of power usage is marginal compared to that too:
300W per A100 * 8766 hours per year * $0.12 per kWh = $316 to power an A100 for a year
Re: Introducing ChatGPT and Whisper APIs
#438Earlier quoted context omitted.
> I have no idea how OpenAI can make money on this. This has to be a loss-leader to lock out competitors before they even get off the ground. The worst thing that can happen to OpenAI+ChatGPT right now is what happened to DallE 2, a competitor comes up with an alternative (even worse if it's free/open like Stable Diffusion) and completely undercuts them. Especially with Meta's new Llama models outperforming GPT-3, it…
I thought it was Midjourney who stole their thunder. Stable Diffusion is free but it's much harder to get good results with it. Midjourney on the other hand spits out art with a very satisfying style.
Check out OpenJourney. It's Stable Diffusion trained on Midjourney style images. I had great results with it.
Re: Introducing ChatGPT and Whisper APIs
#439ELI5 What is a token? Is it a word, question, letter, what? If I ask a question like... What is the capital of Canada? And it responds with 'Ottawa', how many tokens have I used there and how are they calculated?
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?
Re: Introducing ChatGPT and Whisper APIs
#440Question, Can I give openAI some data for it to process so I can use it to my own advantage. Say I want to train it on specific topic of information I've gathered over the years. Can I some how give it that data and then I can use the API to get back out data in a chat or some other forms of questions? I'm not too familiar with how it works.
You can do this to an extent via fine-tuning, but you will need to do so via one of the other GPT-3 models rather than the ChatGPT API model (`GPT-3.5-turbo`). The latter is not available for fine-tuning.
I wonder if it would be possible to simply fine tune another GPT-3 model, have ChatGPT form the query based on input into ChatGPT's API, then pass that response into a call to the GPT-3 model, and take the response from that, pass it into ChatGPT, and ask for it to parse it and make sense of it.