I made my own ChatGPT...

[@dominikkaminski4600]

Great work man! Keep on doing!

[@unconv]

Thank you! I will :)

[@shawnmckenzie_offcial9598]

wow!! Superb!! This really come handy, Thanks man! really appreacite this content!! Kudos! By the way, I need your kind assitantce. Is there anyway that we could modified the output so that the word could appear slowly one-by-one just like what we experience in ChatGPT instead come out all at once? Thanks in advance! Again, love you works. Thumbs up from me.

[@unconv]

Thanks! You need to use Server Sent Events (SSE) to do that. I might make a follow-up video adding that functionality

[@shawnmckenzie_offcial9598]

@@unconv Thanks Man!

[@shawnmckenzie_offcial9598]

@@unconv Currently we are using the ChatGPT 3.5 Turbo. Now there is ChatGPT 4...how we can change or use the chatGPT4 model?

[@unconv]

@@shawnmckenzie_offcial9598 If you have access to the GPT-4 API, you should be able to just change the model from 'gpt-3.5-turbo' to 'gpt-4' in the message.php file. I don't have access to it yet, as it is in Limited Beta and I'm still on the waitlist.

[@shawnmckenzie_offcial9598]

@@unconv Thanks Brother!

[@SBWebDev]

I'm thinking how many subscribers this dude going to have in 3 months. Very good content. Keep it up

[@unconv]

Thank you!!

[@ovrkl]

Top content right here

[@unconv]

Thank you!

[@LOLlol-kw5fo]

Hey I need to see what user is asking and run a function which will give a custom response to the user how to do that??

[@unconv]

Check out my latest video (and 2 upcoming videos)

[@LOLlol-kw5fo]

​@@unconvwhich one??

[@LOLlol-kw5fo]

​@@unconvplease respond

[@richmedialabs]

Thanks for the video and code! I have been playing around with it and notice that the reply from Chat WTF often has the first word twice. Like "SureSure Here's an inspirational quote to keep you motivated..." Do you have any idea why that's happening?

[@unconv]

Is this happening with the latest version? If you can reproduce it, would be good if you could look at the Network tab of Chrome dev tools and see the EventStream tab of the request to "message.php?chat_id=XXX" and see if the first word repeats there too or not.

[@richmedialabs]

The version is from a few days ago. I’ll have to check the specifics. And I’ll check the eventstream.

[@richmedialabs]

@@unconv yes the first word is repeted in the EventStream. it doesn't happen all of the time, so not sure what is causing it. i'll test more, thanks.

[@all-in-one7730]

what code is included on vendor/autoload.php file?

[@unconv]

It is the composer autoloader for the orhanerday/open-ai package

[@unconv]

Search for how to install composer (or ask ChatGPT) and then you need to run "composer install" from the command line in the project folder. This will create the vendor/autoload.php file

[@mulamulelimphaphuli8301]

Thank you.

[@mathew9893]

So once I import my API key from open AI is this good to go or is there anything else I need to do

[@unconv]

You need to run composer install to install the OpenAI client and the Markdown formatter also, but that should be it

[@alirezadaneshmayeh5248]

"Thank you very much, the tutorials were excellent, and the project was really good. I would appreciate it if you could guide me. I want to upload some PDF files into my database so that when a user asks a question, it retrieves the answer from within those files. Could you please advise on which files to use and what changes I should make to make it work correctly?"

[@WebkooIndonesia]

Error now, can you check it? please

[@WebkooIndonesia]

VM387:1 Uncaught (in promise) SyntaxError: Unexpected token '

[@unconv]

Check this GitHub issue: github.com/unconv/chat-wtf/issues/2

[@FaheemShah-i7n]

it takes to much time to send back response how can we reduce response time

[@unconv]

Sometimes the OpenAI API is just slow :( Also, this chatbot only responds after the full response has been generated, so it seems slower than ChatGPT, which starts outputting the response right away. I might update it in the future to use the streaming version so it seems faster. You can also change the token limit, so that the response will be shorter (which might not be what you want, though)

[@unconv]

I just posted a video where I upgrade the code to use the streaming version of the API. It's a lot faster to respond now!

[@bestoftatebros8901]

So how does it work? Do I just upload the files from github to my hosting and add my API key?

[@unconv]

And you need to install the composer packages

[@focusxmedia2014]

can i reach out. i need to get this working. I added my api key and its not working. Im willing to contribute

[@unconv]

You might have whitespace or a newline in the api_key.txt file. I forgot to trim those in the original version. Also you need to run "composer install" to install the dependencies before you can use the code. I made some last minute organizing of folders before I published the code on GitHub and I forgot to move the composer files to the php folder. I have fixed these now in the repository. If it still doesn't work, check your PHP error log or check the HTTP response of message.php from Chrome dev tools.

[@saebifar]

its fantastic :)

[@unconv]

Thank you! You should check out the video where I update it to the new ChatGPT API