🤖Unlock the Power of ChatGPT3 API with AutoHotkey!🔥
How to use ChatGPT3 API with AutoHotkey: require different parameters for web and API calls to get desired results.
- 00:00 🤖 Using AutoHotkey with Chat GPT and Open AI requires different parameters for web and API calls.
- We’re showing how to use AutoHotkey with Chat GPT and Open AI, and there are some important differences between the parameters used on the web version and the API calls.
- 01:37 🔑 Get an API token to submit input and output, based on character count.
- You need an API token to submit input and output, and the formula for calculating tokens is based on characters.
- 02:47 🤔 N answers are returned per request, with a limit of 4,000 tokens.
- There is a limit of 4,000 tokens per request and the number of answers it gives back is N.
- 03:26 🤔 Play with the API library to get the desired results, but be aware of the temperature’s effect.
- Lowering the temperature between 0 and 1 allows for factual answers, while a higher temperature of 1 allows for more creative answers.
- Play with the API library to get the desired results, but be aware of the odd behavior between temperature and something else.
- 05:09 🤔 ChatGPT can’t store history of its actions.
- GPT cannot send two values at the same time, so there is no history of what it does.
- 06:08 🤔 The rules of making a TLDR are to keep the most important events and ideas, be concise, and start with an appropriate emoji.
- Every prompt is independent and context must be manually coded into the input box.
- 07:19 💰 Paying for calls, but API should still work even when Chat GPT is busy and we’re separated from the free version.
- We are paying money for every call and should be able to use the API even when chat GPT is bogged down and busy, and we are likely separated from the free version.
- 08:35 🔥 Get 50% off our new course to learn how to use ChatGPT3 API with AutoHotkey for easier tasks!
- Check out our version 2 course on sale now for half off to learn how to use the amazing tool Chat GPT to make everything simpler.
How to use ChatGPT3 API with AutoHotkey
(3) 🤖Unlock the Power of ChatGPT3 API with AutoHotkey!🔥 – YouTube
(00:00) hey everyone it’s Joe and azaleas here from the automator today we got a really fun topic we’re blending Auto hotkey and also using it with chat GPT and open AI uh in this example we’re going to show you and also give you a link to get the r code for using the open ai’s API calls so this is simply you know browser call is technically an API call but this is using the web service API call and there’s some really important things to realize about it so make sure you stick around and watch the whole thing because
(00:28) there’s some gotchas at the end that we want to make sure you understand yeah that’s right and so let me show you real quick what I’ve got here so if we go ahead and type give me five words here we should get an output from chat gbt with the with this list of parameters and I think this is something that we might want to mention a little bit these parameters here are not the same as when you are in chat ai.
(01:08) com so you might get some answers from chat GPT that are not the same that you would get if you are on the web version of it so you will have to play a little bit with those guys on their website um on the documentation they really mentioned what temperature means uh what top P it is what was what is that on the end so I’m just gonna give you a very high level overview about them the maximum tokens has to do with the length of the answer and what you give it and you have to be very careful with that one because
(01:48) um as this is leveraging your token your change GPT token you might get charged if you use too many tokens right so I would rather have it in lower values if you don’t don’t want to get charged too much um but in certain situations if you want the thing to answer with 200 words you might need to bump up your maximum tokens here and let’s back up a step here just slightly so a you need to have your token you need to create an account you know and give it a credit card or something because there are cost costs
(02:24) involved unlike if there’s a browser with account the next one is once you’ve given it your token um I’m sorry you’re well it is your API token uh when you submit stuff and that’s where he’s saying there’s both the input and the output count against you yes what’s really crazy is the formula for calculating tokens it’s not like every certain characters yeah the good news is it’s actually so cheap right now at least like you you know it’s it’s you don’t really have to
(02:56) worry too much but I I still agree with what Jose is saying so you don’t want to put in a ten thousand and then suddenly realize later like hey wait a minute you know Where’d I get this bill but we’ve yet to have like any sort of a real bill that’s for real and they do have still a limit there’s 4 000 tokens per per request I think it is this is the limit that’s a kind of like a hard hard stop even if you don’t put a limit to yourself so um the N is the number of answers that it might give back right now this
(03:25) interface is just gonna display one of them so this is not really uh an option that is going to be too much flexibility because I only have one output here if I had several windows I could give you different answers right the temperature this is the one that is a little bit interesting because uh uh a 100 is kind of like like giving it freedom to be creative so it it it might come up with new stuff but sometimes you don’t want that sometimes you want it to be factual so you would lower the temperature I think
(03:59) that’s how it works and uh it is between zero and one so it’s 0.2 is a temperature of 0 of 20 and so on the same happens here with the top P so this is uh it’s just a top percent and from the answers that it generates is going to give you the top percentage and usually 0.
(04:24) 5 is okay if you want kind of like the most accurate answer it would be the top one you know the top ten you can do the top ten percent and so on so that you play with this to get the answers that you’re hoping for I wouldn’t use the stream here because that’s where it starts typing out the stuff I have it set up I but the problem the the point is I I let it here so that you know it is available as an option in the API library that we created so you take a look at it and you play a little bit with it and see what you can come up
(04:58) with you from memory there was something to me odd I don’t think they explained it very well if I remember correctly between the temperature and the what was the top p is that what you’re saying yes that yeah that yeah but not the other right exactly that’s what they said like do not change both at the same time but it is not you cannot send those two values I send them both at the same time in the code but they they they seem to only understand one of both so yeah in any case I think this is the one
(05:38) can we go back to the the GUI yes so here and let’s demonstrate it so um say please do this again please do this again now in chat GPT in the browser yeah um we would know what that would do right right but here he just answered sure because unfortunately and I have no idea why and maybe at some point we can figure out how there’s a better way to do it but there’s no history there’s no context they’re independent of each other unfortunately each input you are sending right now is independent of the previous
(06:16) one there is no context now there is a way on because if you take a look at the code I’m sending a um uh Json string with some information there is a way to set up the context in the Json string the problem is that I cannot set it up because for you to set up the context you need to get the previous um ID of the conversation and that’s the part that I haven’t really created yet but that is part of the the issue there is no context automatically you actually have to code it yourself that’s what happens so that’s something
(06:59) that you have to take into consideration when you’re using this every prompt you’re doing is an individual prompt and everything that you need as a context should be in the input box right there in the last really big gotcha which which we’re not 100 sure on when you did one test but we’re not 100 sure is theoretically to me and I know from what I’ve read too we’re paying money for every call it shouldn’t have a problem when the servers are really bogged down and busy you should be able to do this
(07:29) but one time when we went where we were using the browser and then we’re like oh it’s not running oh we’ll just use our ChatGPT3 API with AutoHotkey it also didn’t work so we’re not 100 sure if this will or won’t work when chat GPT is bogged down and busy right it probably means that we are probably separated from the free version somehow but sometimes there might be a thing with channel gbt that all of it is down so yeah that’s what I was thinking as well yeah I’m thinking that probably what we got was that we’re growing like
(08:07) crazy service was down for everybody but we are somehow separated from the free version that probably um we will not have some issues that are free person would have yeah so let us know what you think are you using chat GPT are you trying to use their API I’ll put a URL up with a lot of training on API calls which we have a lot of webinars and videos separate videos and stuff on them also this tool we’re giving you it is in version two so we did Just Launch our you know painlessly transition to V2 this is a great time to
(08:43) start learning V2 so uh in this code it’s in written in version two so you know you might want to consider our course I’ll put a URL up here it’s on sale right now for a half off so you might want to check it out but if you learned something here please like the video really helps us out we get a lot more views we’re the largest Auto hockey channel by far I noticed the other day we have over 1,400 videos and we push out two to three a week and um also don’t forget to like And subscribe because if you don’t subscribe you may
(09:12) miss out on some of our cool stuff that we’re doing later at some point we’ll you know go into this a bit more make this a little more robust to do some other stuff but ChatGPT3 API with AutoHotkey is just a crazy amazing tool that is really um making it so everything we do is so much simpler there you go that’s right thanks everyone cheers