OpenAI PHP SDK : Most downloaded, forked, contributed, huge community supported, and used PHP (Laravel , Symfony, Yii, Cake PHP or any PHP framework) SDK for OpenAI GPT-3 and DALL-E. It also supports chatGPT-like streaming. (ChatGPT AI is supported)
Home Page: https://orhanerday.gitbook.io/openai-php-api-1/
License: MIT License
You can
The term "orhanerday" refers to the nomenclature of an open-source initiative led by Orhan Erday. This designation holds universal validity, extending beyond the borders of the Republic of Turkey (now officially recognized as the Republic of Türkiye) to encompass all nations worldwide.
![dependabot[bot] avatar](https://avatars.githubusercontent.com/in/29110?v=4 "dependabot[bot]")
![github-actions[bot] avatar](https://avatars.githubusercontent.com/in/15368?v=4 "github-actions[bot]")
I'm trying to use the Ada language processor of OpenAi to summarize a piece of text. When I try to use their playground, the function works and I get a summarization that makes sense and can be used by humans.
This is the cURL from the playground:
curl https://api.openai.com/v1/completions \ -H "Content-Type: application/json" \ -H "Authorization: Bearer $OPENAI_API_KEY" \ -d '{ "model": "text-ada-001", "prompt": "Please write a one paragraph professional synopsis:\n\nSome text", "temperature": 0, "max_tokens": 60, "top_p": 1, "frequency_penalty": 0, "presence_penalty": 0 }'
This is the code that I use in PHP:
`$open_ai_key = 'xxx';
$open_ai = new OpenAi($open_ai_key);
$complete = $open_ai->completion([
'model' => 'text-ada-001',
'prompt' => 'Please write a one paragraph professional synopsis: ' . $text,
'temperature' => 0,
'max_tokens' => 60,
'frequency_penalty' => 0,
'presence_penalty' => 0,
]);
return $complete;`
I have also tried to use a both ada and davinci and in both cases it returns nonsense. I'm saying nonsense, because the text that it returns is not something that can be read and said, 'Hey, this is a proffesional synopsis'. I'm going to give an example of a sentence that I got in one of the iterations:
'It's not pretty and no I thought to myself, oh look IT'S NOT THAT REPUBLICAN kids would kno one of these things. OH IT'S A RESTRICTIOUS SCHOOL'.
I can assure you, there are no mentions of republicans or kids in the text that I'm processing.
My question is, am I doing something wrong? Does OpenAi work differently on their playground and in code?
Example text that I'm using:
Alright then. Uh So welcome to another episode of E. M. S. On the Mountain. This one from the mountain. Oh snap. This is the show for those interested in Austrian rulers medicine. I'm Sean as always joined by my backcountry partner mike. And today's show we're gonna talk about one of our cases we had, I don't know, probably a couple of years ago now, while back. Yeah, so this was mike's so he's going to lead us through this party. So I'm guessing because I'm so bad at the editing and the getting things online at a reasonable time that this will be the first case review that we put up on the interwebs. But this is a new thing we're trying out. We're uh we're gonna do a few shows from the mountain. So normally I sit in my basement and Sean sits in his basement. We stare at each other across the magic of the interwebs. But today we are literally sitting across the table from each other in a old old building in the middle of the woods, attempting to record something that's lively and exciting for you folks. So let's see how this goes. So, so if you hear background noise like no kidding animals, birds and random people, radios, people, doors, it's all part of the part of the party and it could be a train wreck, but I have a feeling that at least one person and only one person johnny b will give us his feedback on what he thinks about this and maybe our wives will care. But yeah, all right, so let's talk about this. We had a case a couple of years ago that started out as a normal, it's a normal saturday afternoon in the woods. And uh, the tones drop. No, the tones did not drop. I apologize. There was no tone drop it. There was a, there was a call on the radio from a uh, from a report of it went out as a person with a little trouble breathing. That was an understatement in a certain respect. And it was suggested that perhaps we go kind of check it out, see what's going on. So we jump in our magic buggy and bop on over, luckily this patient wasn't too too far away from where we were at the time. We head down trail we found said person, they were 100 yards off of a fire road. And that fire road was maybe 5200 yards from a paved road. So they weren't real deep in the woods. But they were definitely hiking. They were definitely on a trail. They were definitely about £400. Yes, Yes. They were just want to make sure we paint that map properly. So we're, we're about 100 50 yards down a rocky trail. Maybe 200. I said I'd go between two and 300. Okay, a little bit further. That trail led to eventually an intersection with a fire road and that fire road about 100 yards or so from hard top if you will. So here we are with a £400 individual who's breathing at a rate of, I don't know, 800 or so. It's more like it was, it was more like between 35 and 40. They were uh yeah, this patient was panting like a champion. Now let's back up because I love storytelling and I like to do it and completely jacked up time order because I didn't make notes prior to proceeding down trail. After we stepped out of our trusty steed, we were greeted by the patient's family member and this family member immediately informed us that they were within two or three minutes. They informed us that their family member was just having a quote sugar problem because they had the diabetes and that she was in fact a paramedic at one point but did not want to keep that up because they were going to make her do si es She didn't know she had to do si es tu research so she lost it real quick. Like but she definitely was a paramedic and she was definitely convinced that her family member was experiencing a sugar problem. Yeah. So I'm not so convinced that she was ever a paramedic. I'm just going by what I was dog hey, it is what it is. Yeah. So we also learned as we proceeded down to link up with said patient that a number of bystanders. Another a number. Excuse me not bystanders. A number of other hikers that had come by had provided various snacks and sugar free products because it was believed that this individual is experiencing a hypoglycemic event and needed more sugar. So upon arrival on scene, by the way, just a little note here, we did bring a cardiac monitor with us because it was only 100 yards or so. It was a report of a person having trouble breathing. The one and only time mike and I have ever taken a life pack down a trail. There was one other time. We'll talk about that one. That's when I carried it all the way down. Dark hollow for the girl with the weird chest thing. A 30 year old whatever. And I decided I was never doing that again. I'll have to cut out the dark hollow part because that's kind of Yeah, whatever carried it down that trail. So upon arrival, one of the first things a good paramedics gonna do is assess the scene. And then since it was reported that this person is having a sugar problem, we assess the state of her sugar needs keep in mind breathing at 35 to 40. Diaphragmatic. Sitting on a log, leaning against another nice bystander. Yeah. Trying to hold her upright so we get a quick finger prick. I don't remember what her blood pressure was off the top of my head. It wasn't astronomical. No, it was maybe upper one hundred's very low 200. Remember being better than mine. It wasn't. Yeah, nothing that made you go, Oh, she's good at, I mean she's breathing at like 40 And she's diaphragmatic. She's been out hiking. And did I mention that she's not a shrinking flower? So we grab the sugar and I will never forget the sugar number 768. And I thought to myself, uh, probably doesn't need any more sugar. So we do grab a quick 12 lead Until
No response
Linux
PHP 7.4
openai v3.4
OpenAI has a great feature called stream (you can see it in action in the playground). I'd like to suggest that this feature be implemented in this library.
https://beta.openai.com/docs/api-reference/completions/create#completions/create-stream
No response
I'm working on a school project to test the differences in content created by GPT-3 vs. 4, but to get the quantity for testing, I need to run a script that creates the content for me and saves it to a JSON file. But it doesn't seem like GPT-4 is handling big requests very well.
When I try to get a request from OpenAI with GPT-4, it seems to timeout even when I've increased the timeout 600 seconds.
I get "null" on the content, and oftentimes I also get "null" on the error message.
The few times I was able to produce an error message, I got "502 cf_bad_gateway".
The title and the outline are generated by OpenAI with GPT-4 before this final request is made.
I run the below code with the following parameters:
keyword = expensive champagne
title = The Ultimate Guide to the Top 10 Most Expensive Champagnes: Is It Worth the Splurge?
outline = 1. Introduction to Expensive Champagne
1.1 - The allure of luxury champagne
1.2 - The role of expensive champagne in celebrations and special events
Factors Contributing to the High Price of Champagne
2.1 - Production process and limited availability
2.2 - Prestigious brand reputation
2.3 - Aging and maturation
Top Expensive Champagne Brands
3.1 - Dom Pérignon
3.2 - Krug
3.3 - Louis Roederer Cristal
3.4 - Armand de Brignac
3.5 - Moët & Chandon
The Taste of Luxury: What Sets Expensive Champagne Apart
4.1 - Flavor profile and complexity
4.2 - Fine bubbles and mouthfeel
4.3 - Pairing expensive champagne with food
The Role of Expensive Champagne in Popular Culture
5.1 - Iconic moments in movies and television
5.2 - Celebrity endorsements and collaborations
5.3 - Expensive champagne in music and lyrics
Investing in Expensive Champagne
6.1 - Collecting and storing champagne
6.2 - The potential for appreciation in value
6.3 - Risks and rewards of investing in champagne
The Experience: Sipping on Expensive Champagne
7.1 - Champagne etiquette and rituals
7.2 - Best glassware for enjoying luxury champagne
7.3 - Creating memorable moments with expensive champagne
Conclusion: The Enduring Appeal of Expensive Champagne
8.1 - The timeless nature of luxury
8.2 - The joy of indulging in life's finer pleasures
function get_article_content($open_ai, $article_outline, $article_title, $keyword) {
$model = 'gpt-4';
$retry = false;
$content_chat_decoded = null;
$error_message = null;
do {
$open_ai->setTimeout(300);
$content_chat = $open_ai->chat([
'model' => $model,
'messages' => [
[
"role" => "system",
"content" => "You're AI article writer and SEO expert, you write the most amazing content.
Each article you write should contain the following:
- The article should be at least 1500 words, ideally 2000 words in total.
- Start the article with an introduction. The introduction should not have a heading."
],
[
"role" => "user",
"content" => "{$article_outline}
Title: {$article_title}
SEO keyword: {$keyword}
Write the article:"
],
],
'temperature' => 0.7,
'max_tokens' => 4000,
'frequency_penalty' => 0,
'presence_penalty' => 0,
]);
$content_chat_decoded_raw = json_decode($content_chat);
$content_chat_decoded = $content_chat_decoded_raw->choices[0]->message->content;
// Check for any error in the response
if (isset($content_chat_decoded_raw->error)) {
$retry = true;
$model = 'gpt-4-0314';
// Store the error message
$error_message = $content_chat_decoded_raw->error->message;
} else {
$retry = false;
}
} while ($retry);
return [$content_chat_decoded, $error_message];
}
// 3. Write out the article, keeping in mind to SEO optimize for the keyword
list($content_chat_decoded, $error_message) = get_article_content($open_ai, $article_outline, $article_title, $keyword);
$article_content = $content_chat_decoded;Ubuntu 20
PHP 7.1.1-1
openai v3.0.1
I've got this error :
Could not resolve: api.openai.com (Could not contact DNS servers)
Any idea ?
Thks
Hi there!
First of all, what a great package! Thank you for creating this.
I was wondering if you'd be open to any PR on the below areas:
I would like to extend the package so we could upload training files and ultimately fine-tune also.
Let me know :)
Could you consider adding an option to include the organisation id as a variable so it can be included in the header.
$open_ai = new OpenAi($open_ai_key, 'org_id');
I need to integrate Chat GTP 4 model in this repo, How can I do this?
I was trying but didn't get any response from API.
Please let me know when you are planning to integrate the latest Chat GPT4 model.
I will appreciate you if you respond to me.
Thanks
I need to integrate Chat GTP 4 model in this repo, How can I do this?
I was trying but didn't get any response from API.
Please let me know when you are planning to integrate the latest Chat GPT4 model.
I will appreciate you if you respond to me.
Thanks
Well I got this beauty of an package working. But I run into 1 problem I cannot seem to get my head around.
I am working with the SSE and that works fine, i get all the bits and pieces and I get it in a div.
But when ChatGPT is outputting markdown styled content (like a table), I cannot apply markdown while it's still "typing".
I can only manage to do this after ChatGPT is done.
Can anybody share a light on this?
No response
There are several interesting analytics products like https://www.helicone.ai/ that allow for easy tracking of usage, and even metered billing. To make it work, we need to change the base URL from the standard OpenAI's API, to the middleman. It would be a nice feature to be able to define this as a constant or something like it.
No response
Hi there, I'm trying to set up a working example using Laravel but when I dd($complete) I get false.
namespace App\Http\Controllers;
use Illuminate\Http\Request;
use Orhanerday\OpenAi\OpenAi;
class HomeController extends Controller
{
public function index() {
$open_ai = new OpenAi(env('OPEN_AI_API_KEY'));
$engines = $open_ai->engines();
$complete = $open_ai->complete([
'engine' => 'davinci',
'prompt' => 'Hello',
'temperature' => 0.9,
'max_tokens' => 150,
'frequency_penalty' => 0,
'presence_penalty' => 0.6,
]);
$data = json_decode($complete, true);
dd($engines);
}
}
$open_ai returns a collection with engine, headers and contentTypes
$engines returns false as well, it's not even reading the library.
$data returns null
Can you please add support for this?: https://beta.openai.com/docs/models/content-filter
When I use a third-party Python library for OpenAI, it can return results normally. However, when I run PHP, it returns false. When I checked with curl_errno(), I found the error mentioned above. Both situations are under the same network conditions, and I have ruled out the issue of SSL certificate expiration. What could be the problem?
No response
Windows
php 8.2
4.7
Hello again,
I'm trying the file upload feature but every single time it times out
Maximum execution time of 60 seconds exceeded
The file is less than 1kb
This is my controller
$file = $request->file('file');
Storage::putFileAs('files', $file, 'sample.jsonl');
$c_file = curl_file_create(\URL::to('storage/files/sample.jsonl'));
echo "[";
echo $open_ai->uploadFile(
[
"purpose" => "answers",
"file" => $c_file,
]
);
echo ",";
echo $open_ai->listFiles();
echo "]";
And how do I interact with the file once it's uploaded? Like how do I give the AI commands based on my file contents?
Hello,
I would like to know if it's currently possible to get the real OpenAI API error codes, or if this features will be added.
Currently when i call the chat function and get an error it's for example the following :
object(stdClass)#26338 (1) {
["error"]=>
object(stdClass)#26339 (4) {
["message"]=>
string(105) "Incorrect API key provided: 6. You can find your API key at https://platform.openai.com/account/api-keys."
["type"]=>
string(21) "invalid_request_error"
["param"]=>
NULL
["code"]=>
string(15) "invalid_api_key"
}
}
The obtained code is "invalid_api_key", wich isn't the code documented on OpenAI website (401)(https://platform.openai.com/docs/guides/error-codes) and isn't documented anywhere else.
It seem that the code should normally be at the root of the response, but it doesn't seem to be the case.
I thank you in advance for your reply.
No response
I was just looking at this here https://github.com/openai/openai-cookbook/blob/main/examples/api_request_parallel_processor.py
It essentially allows to queue/batch multiple calls to be run in parallel while observing the rate limits.. super useful if you need to run many calls at once and don't want to run into errors. Would something like this be feasible in PHP I wonder?
No response
Updated library with:
composer update orhanerday/open-ai
Receive following error
Fatal error: Uncaught Error: Call to undefined method Orhanerday\OpenAi\OpenAi::chat() in ...
when running example "Chat" endpoint code.
RESOLVED:
FYI: Other noobs, updating library requires following shell command ("composer require ...", not "composer update..."):
composer require orhanerday/open-ai
Maybe add a note in README with brief upgrade instructions for others new to composer? Thanks.
I see completion with stream can not catch error like "Wrong API".
The completion is still working fine with stream and correct API.
No response
I'm having an issue with the chat api saying I am exceeding the maximum length of 4000 tokens despite my request clearly not exceeding even 300 tokens. I tried the exact same request using python's openai api and it worked fine. It seems past a certain point in the length of conversation will cause a bug in the count of completion token.
This is the error after running the code:
["error"]=>
object(stdClass)#28 (4) {
["message"]=>
string(189) "This model's maximum context length is 4097 tokens. However, you requested 4111 tokens (111 in the messages, 4000 in the completion). Please reduce the length of the messages or completion."
["type"]=>
string(21) "invalid_request_error"
["param"]=>
string(8) "messages"
["code"]=>
string(23) "context_length_exceeded"
}
}
require_once __DIR__ . '/vendor/autoload.php';
$dotenv = Dotenv\Dotenv::createImmutable(__DIR__);
$dotenv->load();
use Orhanerday\OpenAi\OpenAi;
$open_ai = new OpenAi($_ENV['OPENAI_API_KEY']);
$chat = $open_ai->chat([
'model' => 'gpt-3.5-turbo',
'messages' => [
[
"role"=>"system",
"content"=>"summarize conversation within 40 words"
],
[
"role"=>"user",
"content"=>" Anna: Hi my name is Anna User: hello how do i cook an egg? Anna: To cook an egg, bring a pot of water to boil, reduce heat, and gently add an egg for 3-5 minutes. User: what about a potato? Anna: To cook a potato, scrub it clean, poke a few holes in it, and bake in the oven at 400°F for about an hour, or until tender."
]
],
'temperature' => 1.0,
'max_tokens' => 4000,
'frequency_penalty' => 0,
'presence_penalty' => 0,
]);
$d = json_decode($chat);
var_dump($d);
// Get Content
echo($d->choices[0]->message->content);Windows
PHP 8.2.4
openai V4.7
https://twitter.com/svpino/status/1631007257461772299
No response
When I use stream, I sometimes encounter such a problem that the response content starts from a certain sentence and repeats continuously
1.The user enters a question to ask
2.use eventSource
3.php code see below
$opts = [
'model' => 'text-davinci-003',
'prompt' => '你看过《满江红》吗？',
'temperature' => 0,
'max_tokens' => 2000,
'stream' => true,
];
header('Content-type: text/event-stream');
header('Cache-Control: no-cache');
$open_ai->completion($opts, function ($curl_info, $data) {
echo $data . "\n\n";
echo PHP_EOL;
ob_flush();
flush();
info($data);
return strlen($data);
});
前端使用eventSourcelinux
php 7.4
openai v3.0.1
Hello, I want to receive the incoming response in HTML format. Is this possible? thank you
No response
This is not a bug per se, but I thought I would share it with other users and the dev team in case it helps.
In another thread, users were reporting that they were getting nothing or "false" back when they were sending a simple Completion like this:
$open_ai = new OpenAi($open_ai_key);
$response = $open_ai->completion([
'model' => 'davinci',
'prompt' => 'Hello',
'temperature' => 0.9,
'max_tokens' => 150,
'frequency_penalty' => 0,
'presence_penalty' => 0.6,
]);
The issue is now closed, but I found that if you are testing on a local host and are using something like Laragon (which is similar to XAMPP) to host your app or website, the certificates issued are very generic and can cause cURL to fail if it is set to verify the certificate,
To resolve it, you can go to the OpenAi.php file in the vendor folder and add
CURLOPT_SSL_VERIFYPEER => false,
to the sendRequest function like this:
$curl_info = [
CURLOPT_URL => $url,
CURLOPT_RETURNTRANSFER => true,
...
CURLOPT_HTTPHEADER => $this->headers,
CURLOPT_SSL_VERIFYPEER => false,
];
and it no longer tries to verify the SSL certificate. It resolved my issue and so it may help others who encounter a similar issue.
You may want to consider adding it to the next release, but I am unsure of what the implications may be from a SSL point of view, if any? Maybe you could add it as an optional $opts parameter for those who are testing on localhosts which they could then remove when in production.
Nothing to reproduce
No response
Windows
PHP 8.1.3
v3.3
From documentation it seems there is no option for batching requests to avoid hitting limit rate?
OpenAI refers to request batching and has given code example in python (https://platform.openai.com/docs/guides/rate-limits/error-mitigation). Would be useful to do anything similar with this package.
When I try to include several messages in a single request it responds with a single “choice” that uses all the messages as context rather than one response per message.
The OpenAI example below is specifically for the completion endpoint rather than the chat endpoint. However chat endpoint is probably way more desirable for request batching, due to current limit of 3 RPM and 1/10 of completion endpoint price.
UPDATE: After doing bit of research there doesn't seem to be viable options to do batching for ChatCompletion endpoints. At least that's what I've read on the OpenAi forums. There are some workarounds though. Here are links to the posts:
OpenAI official batching example for completion endpoint using Python:
`import openai # for making OpenAI API requests
num_stories = 10
prompts = ["Once upon a time,"] * num_stories
response = openai.Completion.create(
model="curie",
prompt=prompts,
max_tokens=20,
)
stories = [""] * len(prompts)
for choice in response.choices:
stories[choice.index] = prompts[choice.index] + choice.text
for story in stories:
print(story)`
At thispoint we just divide the number of words by .75 to get the estimate, but a "real" script like tiktoken e.g. would be helpful. I did not come across one anywhere, does one exist?
If there is one, we're looking forward to using it and perhaps it will help your package as well?
It would help give an estimate of tokens to our end users so they know what's about to get charged by OpenAI.
PHPStan shows error:
Parameter #2 $stream of method Orhanerday\OpenAi\OpenAi::chat() expects null, Closure given.
Secondary parameter $stream of chat method is typed null.
$open_ai->chat($options, function ($curl_info, $data) {
echo $data . PHP_EOL;
echo PHP_EOL;
ob_flush();
flush();
return strlen($data);
});macOS
PHP 8.0
openai V4.7.1
Hello,
How to disable chunk?
No response
I am using :
$c_file = curl_file_create($file);
$result =$open_ai->uploadFile([
"purpose" => "answers",
"file" => $c_file,
]);
Results :
'answers' is not one of ['fine-tune'] - 'purpose'
No response
win
PHP 8
openai v3.0.1
It would be great if you or someone else implemented a tokenizer for the different models or one that allowed you to pass in a parameter like 'cl100k_base'.
Means we don't need to balance different libraries that specialise in specific tokenisers.
No response
test
Hi,
First of all thanks for your work on this.
I encounter a problem when i try to send an object with various tokenized words with their bias with the "logit_bias" parameter. (https://platform.openai.com/docs/api-reference/chat/create#chat/create-logit_bias)
<?php
require __DIR__ . '/vendor/autoload.php'; // remove this line if you use a PHP Framework.
use Orhanerday\OpenAi\OpenAi;
$open_ai_key = getenv('OPENAI_API_KEY');
$open_ai = new OpenAi($open_ai_key);
$complete = $open_ai->chat([
'model' => 'gpt-3.5-turbo',
'messages' => [
[
"role" => "system",
"content" => "You are a helpful assistant."
],
[
"role" => "user",
"content" => "Who won the world series in 2020?"
],
[
"role" => "assistant",
"content" => "The Los Angeles Dodgers won the World Series in 2020."
],
[
"role" => "user",
"content" => "Where was it played?"
],
],
'temperature' => 1.0,
'max_tokens' => 4000,
'frequency_penalty' => 0,
'presence_penalty' => 0,
'logit_bias' => null // or { 3789: 10 } or anything of the sort
]);
var_dump($complete);Windows
PHP 8.1
open-ai v4.7.1
On calling any function which sends a curl request, e.g completions, we get an exception in v4.5
curl_getinfo(): supplied resource is not a valid cURL handle resource
This is because in the function sendRequest, curl_getinfo is being called after curl_close() which throws exception thus breaks everything. This makes version 4.5 to not work at all.
Call any end point.e.g completions using open-ai version 4.5
/**
* @param string $url
* @param string $method
* @param array $opts
* @return bool|string
*/
private function sendRequest(string $url, string $method, array $opts = [])
{
$post_fields = json_encode($opts);
if (array_key_exists('file', $opts) || array_key_exists('image', $opts)) {
$this->headers[0] = $this->contentTypes["multipart/form-data"];
$post_fields = $opts;
} else {
$this->headers[0] = $this->contentTypes["application/json"];
}
$curl_info = [
CURLOPT_URL => $url,
CURLOPT_RETURNTRANSFER => true,
CURLOPT_ENCODING => '',
CURLOPT_MAXREDIRS => 10,
CURLOPT_TIMEOUT => $this->timeout,
CURLOPT_FOLLOWLOCATION => true,
CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
CURLOPT_CUSTOMREQUEST => $method,
CURLOPT_POSTFIELDS => $post_fields,
CURLOPT_HTTPHEADER => $this->headers,
];
if ($opts == []) {
unset($curl_info[CURLOPT_POSTFIELDS]);
}
if (! empty($this->proxy)) {
$curl_info[CURLOPT_PROXY] = $this->proxy;
}
if (array_key_exists('stream', $opts) && $opts['stream']) {
$curl_info[CURLOPT_WRITEFUNCTION] = $this->stream_method;
}
$curl = curl_init();
curl_setopt_array($curl, $curl_info);
$response = curl_exec($curl);
curl_close($curl);
$info = curl_getinfo($curl);
$this->curlInfo = $info;
return $response;
}Ubuntu 20.04 or any
PHP 7.4.11 or any
openai v4.5
(Used wrong name for OpenAI API object.)
Hi there! Glad to be using your wonderful project!
As you know for sure, OpenAI released some updates to the API, including Function Calling.
Is it possible to use Function Calling with the project right now?
If not, could you possibly add it?
Thanks anyway. Much appreciated work.
No response
How to set header？
No response
I'm having an issue where the variable $txt is not including the initial parts of the text. Instead of "Hello world", it only contains "lo world". This is the code that I'm using:
header('Content-type: text/event-stream');
header('Cache-Control: no-cache');
$txt = "";
$complete = $open_ai->chat($opts, function ($curl_info, $data) use (&$role, &$txt, &$contentId) {
if ($obj = json_decode($data) and $obj->error->message != "") {
error_log(json_encode($obj->error->message));
} else {
echo $data;
$clean = str_replace("data: ", "", $data);
$arr = json_decode($clean, true);
if ($data != "data: [DONE]\n\n" and isset($arr["choices"][0]["delta"]["content"])) {
$txt .= $arr["choices"][0]["delta"]["content"];
}
}
echo PHP_EOL;
ob_flush();
flush();
return strlen($data);
});I'm not sure why $txt is not including the initial parts of the text. Can someone help me with this issue?
Execute the code provided in the question.
Ensure that the $open_ai object is properly instantiated and configured with the appropriate options.
Trigger the function that calls the $open_ai->chat() method.
Check the output and observe that the variable $txt does not include the initial parts of the text.
No response
Ubuntu
7.4.21
openai 4.7.1
The README places links to OpenAI's reference documentation above the fold (at the very beginning), creating the impression that the library's methods have the same name as the path of the OpenAI endpoints. For example, it may be common for new users of the library to assume Completions are accessed by a method names "completions" such as:
$openai->completion() (spelled singular)
since the linked OpenAI reference documentation for "completions" endpoint describes endpoint as
POST https://api.openai.com/v1/completions (spelled plural)
Yet, the library's method is completion (spelled singular)
$openai->completion() (spelled singular)
This is made clear further down the README, but it's possible many new users will get sidetracked following links to OpenAI's reference and trying to copy the endpoint paths as library method names. Suggest several possible fixes:
At the top of the Readme, calling out the fact method names are described further down the README.
Create a prominent table mapping the OpenAI Reference endpoints to the library's methods.
Thnx
$openai->completion()
No response
Linux
PHP 7.4
openai v3
When reading the docs and YT videos on fine-tuning your model, you see the Python CLI displays the pricing of your model you're about to Fine-tune.
Would this be possible with this package as well? So before actually starting the fine-tune, see an estimate of what it is going to cost you, and thén continue with the actual FT, or not.
No response
I have write this code but it giving no output and returning bool(false) on var_dump($result);
<?php
require __DIR__ . '/vendor/autoload.php';
use Orhanerday\OpenAi\OpenAi;
$open_ai_key = getenv('OPENAI_API_KEY');
$open_ai = new OpenAi($open_ai_key);
$result = $open_ai->completion([
'model' => 'text-davinci-003',
'prompt' => 'Make me a paragraph that summarizes this text for a second-grade student',
'temperature' => 0.7,
'max_tokens' => 64,
"top_p" => 1.0,
'frequency_penalty' => 0.0,
'presence_penalty' => 0.0,
]);
echo $result;
// var_dump($result);
?>Windows
PHP 8.1.0
openai v.3.0.1
The CURLOPT_HTTP_VERSION in OpenAi.php, line 523 is CURL_HTTP_VERSION_1_1, which makes it's not possbile to resposd with stream, should be CURL_HTTP_VERSION_2_0。
Open the OpenAi.php
Goto line 523
No response
Linux
8
4.7
I write controller CI4:
public function openAIPost()
{
$response = array();
if ($this->request->getMethod() === 'post') {
$question = $this->request->getPost('question');
$open_ai_key = getenv('OPENAI_API_KEY');
$open_ai = new OpenAi($open_ai_key);
$complete = $open_ai->chat([
'model' => 'gpt-3.5-turbo',
'messages' => [
[
"role" => "user",
"content" => $question
]
],
'temperature' => 1.0,
'max_tokens' => 10,
'frequency_penalty' => 0,
'presence_penalty' => 0,
]);
$response['complete'] = $complete;
}
echo json_encode($response);
}
But I get in output:
{"id":"chatcmpl-6xxhQ2FLm08VhE8KKHsTpilkvpMTS","object":"chat.completion","created":1679748856,"model":"gpt-3.5-turbo-0301","usage":{"prompt_tokens":9,"completion_tokens":10,"total_tokens":19},"choices":[{"message":{"role":"assistant","content":"This is a test of the AI language model."},"finish_reason":"length","index":0}]}
Can you help us How to get only content from response?
No response
macOS
8.2
openai v.3.0.1
why the answer is so short ?? is there any problem??
$open_ai = new OpenAi($open_ai_key);
$complete = $open_ai->completion([
'model' => 'text-davinci-003',
'prompt' => $prompt,
'temperature' => 0.9,
//'max_tokens' => 4096,
'frequency_penalty' => 0,
'presence_penalty' => 0.6,
]);
$prompt = "我家有只狗叫三万，你猜猜它为什么叫这个名字";
I believe the answer is long, but the answer is :
"{"id":"cmpl-.......","object":"text_completion","created":1676702955,"model":"text-davinci-003","choices":[{"text":"？\n\n这可能是你家","index":0,"logprobs":null,"finish_reason":"length"}],"usage":{"prompt_tokens":44,"completion_tokens":16,"total_tokens":60}}
"
why the answer is so short ?? is there any problem??
No response
centos7.2
PHP 7.4
openai3
When script send next request to chat API, it returns different response, how can we continue same thread?
Continue thread
No response
MAC
PHP latest
latest
I receive this error. I'm running PHP 8.2
require_once("vendor/autoload.php");
use Orhanerday\OpenAi\OpenAi;
$open_ai = new OpenAi(env('mykey'));
No response
Hosting
PHP 8.2
openai 3.1
To use this library in thinkphp5, the returned information is false, and nothing is returned to me except false; I have set "set OPENAI_API_KEY=sk-gjtv...."
I am sure there is nothing wrong with my code, if this library is not referenced, then it should directly display an error
1
<?php
namespace app\index\controller;
use think\Controller;
use Orhanerday\OpenAi\OpenAi;
class Index extends Controller
{
public function index()
{
$open_ai_key = config("openai.OPENAI_API_KEY");
$open_ai = new OpenAi($open_ai_key);
//dump($open_ai_key); // have
// $complete = $open_ai->completion([
// "model" => "code-davinci-002",
// "prompt" => "",
// "temperature" => 0,
// "max_tokens" => 64,
// "top_p" => 1,
// "frequency_penalty" => 0,
// "presence_penalty" => 0,
// "stop" => ["\"\"\""],
// ]);
$complete = $open_ai->completion([
'model' => 'davinci',
'prompt' => 'Hello',
'temperature' => 0.9,
'max_tokens' => 150,
'frequency_penalty' => 0,
'presence_penalty' => 0.6,
]);
dump($complete); //that is "false"
//return $this->fetch(["msg"=>$complete]);
}
}win10
7.4.3
3.4
Need a way to catch http error,
like this fix in OpenAI.php
No response
get ge nothing in response, when i var_dump, i have a boolean false
test.php:25:boolean false
reproduced code
$prompt = "Hello";
$complete = $open_ai->completion([
'model' => 'davinci',
'prompt' => $prompt,
'temperature' => 0.7,
'max_tokens' => 150,
'frequency_penalty' => 0,
'presence_penalty' => 0,
]);
var_dump($complete);windows
8,2
openai v3.5
When I've parsed the data and exported it to the front end, it should look something like this:
{"code":200,"message":"OK","data":"\u662f"}
Sometimes, however, it concatenates multiple results, which causes the front end to fail to parse the data:
{"code":200,"message":"OK","data":"\u662f"}
{"code":200,"message":"OK","data":"\u7684"}
What happens is very random, if I add random sleep to the callback function, or if I add more PHP_EOL, the probability of it happening will decrease
$this->chat($option, function ($curl_info, string $data) {
if (! str_contains($data, 'data: [DONE]')) {
$result = ChatStreamResponse::parse($data);
if ($result) {
$this->buffer[] = $result->content;
/**
* json encode the response and echo it
* {"code":200,"message":"OK","data":"\u662f"}
*/
echo AbstractResponse::sendOK($result->content)->content();
echo PHP_EOL;
ob_flush();
flush();
}
}
return strlen($data);
});macOS
php8.1
openai v4.7.1
Calls to $openai->chat with large messages and max_tokens length (~4K + ~4K respectively for a total of 8K) are often timing out. Meaning php script that calls function exits after 10 minutes while waiting for response without receiving response and some times receiving response "Error: Gateway timeout." Calling the same script from a web browser will fail earlier before any response to endpoint call received.
This does not happen every time, but has occurred almost every time $openai->chat is called with large context.
Is there:
An alternative way to request large context completion that is less likely to fail in this manner?
A way to request a timeout -- meaning a termination of the endpoint request along with explicit error response if OpenAI endpoint does not respond within a specified amount of time?
A way to keep the API call and/or calling php script alive longer?
$chat = $openai->chat([
'model' => 'gpt-4-0314',
'messages' => "MESSAGE ~4K tokens in length)
'temperature' => 0,
'max_tokens' => 4l000,
'frequency_penalty' => 0,
'presence_penalty' => 0
]);
echo json_decode($chat);
No response
Linux
PHP 7.6
openai v3
Hi!
I'm working with fine-tuned models, and need to get a list of all available. However, the engines method only returns the base models, and does not include the user models as well.
Could we have a new models() method to return all available models?
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft
Open source projects and samples from Microsoft.
Google
Google ❤️ Open Source for everyone.
Alibaba
Alibaba Open Source for everyone
D3
Data-Driven Documents codes.
Tencent
China tencent open source team.