I do not work, I ran everything according to the instructions, composer did not return errors, everything worked fine, the local server is running without errors, api key added his, trying to make any request, returns nothing, where can I find a solution to the problem? Here's a screenshot

Every time I submit a new question, as long as I don’t refresh the page or reopen the page, everything will be fine, otherwise the same question and answer will be repeated => Sorry, an error occurred, can you ask your question again?
Where am I going wrong, is it my cause?

#########################################
[root@ip-172-31-30-218 ~]# sqlite3 --version
3.7.17 2013-05-20 00:56:22 118a3b35693b134d56ebd780123b7fd6f1497668

[root@ip-172-31-30-218 ~]# php --version
PHP 8.1.14 (cli) (built: Jan 9 2023 18:48:38) (NTS)
Copyright (c) The PHP Group
Zend Engine v4.1.14, Copyright (c) Zend Technologies
with Zend OPcache v8.1.14, Copyright (c), by Zend Technologies`
#########################################

before_refresh:

after_refresh:

i want to change to davinci but it not working only gpt-3.5-turbo is working

On Line 8 of Api.php You are doing the following
$user_id = $_POST['user_id']; -> here there is no authentication check to the api.php endpoint to see if the user is logged in meaning You are suffering from 2 issues here lets start with the first problem you should authenticate the user before allowing them to make changes. But that does not stop the problem just restricts to your user base for anyone using this

The core problem here is by accepting the user id from the user and not session variables in the php only that the user does not control, You are allowing any user to just change or delete another users data or chat history

According to ChatGPT, you should be able to add personas, like this:

$opts = [
    'prompt' => $prompt,
    'temperature' => 0.5,
    "max_tokens" => 2048,
	'persona' => [
		'name' => 'Zumm',
		'emotions' => [
			'friendly',
			'helpful',
			'witty'
		],
		'voice' => 'gravelly'
	],
    "frequency_penalty" => 0,
    "presence_penalty" => 0.6,
    "stream" => true,
    "top_p" => 1,
    "stop" => [" Human:", " AI:"]
];

But when I do that, I stop getting responses. Any ideas why?

Bro this is terrific. By any chance, is it possible to use this chat as a chatbot that appears at the bottom of a website, customer service style? Do you have a hint as of how it might be able to implement it? Thanks anyway, your work is spectacular.

I am update it by git and try by downloading, noticed that now the key must be taken from the .env but it still do not work ScreenChat
But now in log the error to which it refers
Just in case, a screenshot of line 57 that I have in this file.

This model's maximum context length is 4097 tokens. However, your messages resulted in 192871 tokens. Please reduce the length of the messages.

only the first word of ai response is received and then eventSource stops receiving responses in several servers.

Something like Hello! How can I assist you today

Hello,
I don't want "object":"chat.completion.chunk"
How I can get it on one line?

Hey guys, I've just checked out the repository and I'm really pleased with it. It's great. Just a couple of suggestions from me;

• The chat window isn't auto-scrolling on new messages...so manual scrolling is necessary.
• The code assumes that the codebase sits within the root of our web server. In my case, I needed to update some files in order to make the application work.
• Would be nice to add an input field on startup that checks for the presence of an API key and only then load up the chat interface.
  It would be nice to form the code blocks generated by Open AI too.

I don't know if you guys accept PRs but I can certainly provide the code to cover said points in one if you accept those.

All in all, great work on this! Thank you guys.
Martin.

I came here from the openai-api project, which is awesome. I decided to use this as the "Hello World" of the api, and got it working. However, the results are quite lackluster using this project, as evidenced below:

ChatGPT gives vibrant answers to all the questions.

After sqlite3 writes about 20K data (~1.8k token), the event will return an error, the database cannot be updated, and the chat will not have any reply.

But when I manually delete a piece of data with more content, it works again until it "fills up" again.

Like some sort of throttling is being triggered, this issue exists making the chat history meaningless. Can you fix it?🤔

I'm trying to use davinci model, so I replaced model in $opts in event-stream.php:
$opts = [
'model' => 'davinci',
...
];

but now I get "Invalid URL (POST /v1/chat/completions)"

Could you point me in the right direction?

What I intend to do? I'd like to train a model with custom questions/answer and then use it in the chat. And I read that fine-tuning is available only for these models: davinci, curie, babbage, and ada

Thanks

After I run the program, I get an error:

Warning: SQLite3::prepare(): Unable to prepare statement: 1, no such table: chat_history in /home/php/app/api.php on line 12

Fatal error: Uncaught Error: Call to a member function bindValue() on bool in /home/php/app/api.php:13 Stack trace: #0 {main} thrown in /home/php/app/api.php on line 13

unable to get local issuer certificate in C:\Users\TECHQUE\Desktop\ChatGPT\vendor\orhanerday\open-ai\src\OpenAi.php:962
Stack trace:
#0 C:\Users\TECHQUE\Desktop\ChatGPT\vendor\orhanerday\open-ai\src\OpenAi.php(229): Orhanerday\OpenAi\OpenAi->sendRequest()
#1 C:\Users\TECHQUE\Desktop\ChatGPT\event-stream.php(51): Orhanerday\OpenAi\OpenAi->chat()
#2 {main}
thrown in C:\Users\TECHQUE\Desktop\ChatGPT\vendor\orhanerday\open-ai\src\OpenAi.php on line 962

Hi,

First of all, big thanks for such a wonderful collection of codes.

I am really interested to know how I can extract images(image URLs) from a webpage by inputting the web page URL.

Thanks.

when i Log::info $data like this

$this->openAi->chat([
                'model' => 'gpt-3.5-turbo-16k',
                'messages' => $temp,
                'temperature' => 1.0,
                'max_tokens' => 3000,
                'frequency_penalty' => 0,
                'presence_penalty' => 0,
                'stream' => true
            ], function ($curl_info, $data) use ($gptArticle) {
                Log::info($data);

Sometimes it is correct：

[2023-10-12 10:05:02] local.INFO: data: {"id":"chatcmpl-88fLfcuOH33GOKQ49iq2vXAuzCHZQ","object":"chat.completion.chunk","created":1697076259,"model":"gpt-3.5-turbo-16k-0613","choices":[{"index":0,"delta":{"content":" air"},"finish_reason":null}]}


[2023-10-12 10:05:02] local.INFO: data: {"id":"chatcmpl-88fLfcuOH33GOKQ49iq2vXAuzCHZQ","object":"chat.completion.chunk","created":1697076259,"model":"gpt-3.5-turbo-16k-0613","choices":[{"index":0,"delta":{"content":" compressor"},"finish_reason":null}]}


[2023-10-12 10:05:02] local.INFO: data: {"id":"chatcmpl-88fLfcuOH33GOKQ49iq2vXAuzCHZQ","object":"chat.completion.chunk","created":1697076259,"model":"gpt-3.5-turbo-16k-0613","choices":[{"index":0,"delta":{"content":" guarantees"},"finish_reason":null}]}


[2023-10-12 10:05:02] local.INFO: data: {"id":"chatcmpl-88fLfcuOH33GOKQ49iq2vXAuzCHZQ","object":"chat.completion.chunk","created":1697076259,"model":"gpt-3.5-turbo-16k-0613","choices":[{"index":0,"delta":{"content":" the"},"finish_reason":null}]}


[2023-10-12 10:05:02] local.INFO: data: {"id":"chatcmpl-88fLfcuOH33GOKQ49iq2vXAuzCHZQ","object":"chat.completion.chunk","created":1697076259,"model":"gpt-3.5-turbo-16k-0613","choices":[{"index":0,"delta":{"content":" highest"},"finish_reason":null}]}


[2023-10-12 10:05:02] local.INFO: data: {"id":"chatcmpl-88fLfcuOH33GOKQ49iq2vXAuzCHZQ","object":"chat.completion.chunk","created":1697076259,"model":"gpt-3.5-turbo-16k-0613","choices":[{"index":0,"delta":{"content":" standard"},"finish_reason":null}]}

Sometimes it is trouble：

[2023-10-12 09:45:10] local.INFO: data: {"id":"chatcmpl-88f382uaVhRbGxjRK8xW2MsNW4832","object":"chat.completion.chunk","created":1697075110,"model":"gpt-3.5-turbo-16k-0613","c  
[2023-10-12 09:45:11] local.INFO: hoices":[{"index":0,"delta":{"role":"assistant","content":""},"finish_reason":null}]}

data: {"id":"chatcmpl-88f382uaVhRbGxjRK8xW2MsNW4832","object":"chat.completion.chunk","created":1697075110,"model":"gpt-3.5-turbo
[2023-10-12 09:45:11] local.INFO: -16k-0613","choices":[{"index":0,"delta":{"content":"#"},"finish_reason":null}]}

data: {"id":"chatcmpl-88f382uaVhRbGxjRK8xW2MsNW4832","object":"chat.completion.chunk","created":1697075110,"model":"gpt-3.5-turbo-16k
[2023-10-12 09:45:11] local.INFO: -0613","choices":[{"index":0,"delta":{"content":" The"},"finish_reason":null}]}

data: {"id":"chatcmpl-88f382uaVhRbGxjRK8xW2MsNW4832","object":"chat.completion.chunk","created":1697075110,"model":"gpt-3.5-turbo-16  
[2023-10-12 09:45:11] local.INFO: k-0613","choices":[{"index":0,"delta":{"content":" Power"},"finish_reason":null}]}

data: {"id":"chatcmpl-88f382uaVhRbGxjRK8xW2MsNW4832","object":"chat.completion.chunk","created":1697075110,"model":"gpt-3.5-turbo-16k-061
[2023-10-12 09:45:11] local.INFO: 3","choices":[{"index":0,"delta":{"content":"house"},"finish_reason":null}]}

data: {"id":"chatcmpl-88f382uaVhRbGxjRK8xW2MsNW4832","object":"chat.completion.chunk","created":1697075110,"model":"gpt-3.5-turbo-16k  
[2023-10-12 09:45:11] local.INFO: -0613","choices":[{"index":0,"delta":{"content":" of"},"finish_reason":null}]}

data: {"id":"chatcmpl-88f382uaVhRbGxjRK8xW2MsNW4832","object":"chat.completion.chunk","created":1697075110,"model":"gpt-3.5-turbo-16k
[2023-10-12 09:45:11] local.INFO: -0613","choices":[{"index":0,"delta":{"content":" Efficiency"},"finish_reason":null}]}

data: {"id":"chatcmpl-88f382uaVhRbGxjRK8xW2MsNW4832","object":"chat.completion.chunk","created":1697075110,"model":"gpt-3.5-  
[2023-10-12 09:45:11] local.INFO: turbo-16k-0613","choices":[{"index":0,"delta":{"content":":"},"finish_reason":null}]}

data: {"id":"chatcmpl-88f382uaVhRbGxjRK8xW2MsNW4832","object":"chat.completion.chunk","created":1697075110,"model":"gpt-3.5-turbo
[2023-10-12 09:45:11] local.INFO: -16k-0613","choices":[{"index":0,"delta":{"content":" Un"},"finish_reason":null}]}

Very confused

Hello,

I was looking for an alternative (with same algorithm (OpenAI API), as the ChatGPT main live chat website are literally freezing, not able to provide the rest of answer, as "typing effect" they use or i don't know, make the answer not complete. As i see, this app using the same effect, is there a way to just remove the effect ? As it look to be a huge problem for so many peoples (if not all) using chatgpt API, having answer from bot which was cropped for unknown reason (typing animation i guess as i tried another app which was not real time (so, no typing animation), and never ever had uncompleted answer.

Is there a way to remove typing effect from chatgpt bot ?

thanks

Hi

Thanks for this nice streaming feature ;-)

Does the docker use GPT3 or 4 and how do we switch please?

Thanks