HttpGPT is an Unreal Engine plugin that facilitates integration with OpenAI's GPT based services (ChatGPT and DALL-E) through asynchronous REST requests, making it easy for developers to communicate with these services.
License: MIT License
The access problem of a single window is suspected to be restricted. After asking a few questions, there is no response. It can be used normally after restarting, but it is not related to the problem of the previous window.
Greetings,
HttpGPTHelper.cpp (version 1.5.4, HttpGPT_UE51.zip file) does not include EHttpGPTChatModel::gpt35turbo16k support for these two functions:
UHttpGPTHelper::GetEndpointForModel
UHttpGPTHelper::ModelSupportsChat
Example excerpt from the log that shows the problem when trying to use the gpt-35-turbo-16k model:
LogHttpGPT: Error: OnProgressCompleted (47373): Request failed
LogHttpGPT: Display: SetReadyToDestroy (47373): Setting task as Ready to Destroy
LogBlueprintUserMessages: [BP_HttpGPT_Test_C_2] Invalid URL (POST /None)
Adding the missing ENUM case fixes the problem.
Thank you for writing HttpGPT!
Jen
Hello! there is something weird that is happening lately with this plugin in my project. Before was running perfect with any crashes, but lately started to crash (overall on packaged builds) and I don't understand why. Any idea what could be? Here is a link with the crash screenshot: https://pbs.twimg.com/media/F0qEnmnXoAAtJqW?format=jpg&name=large
Image generation works fine. But when i try to open a chat window i get a completely empty window (UE 5.2)...
Heya! :)
Am I doing something wrong? This was working before as I have created a system based off these results - latest version all returning zero for some reason.
The code is currently disorganized.
As it's an editor module with small code pieces, there's no need to focus on POO concepts. I'll organize the code to try improve the readability, modularity and reduce compilation impacts, also trying to find a way to turn better/easier to customize.
Instead setting the custom context in project settings, add the option to the chat.
This feature is aiming to support multiple chat sections. Each section can contains it's context: #62
Main hotkey: Press ENTER to send message
I would like to be able to configure azure openAI endpoints
Hello,
When I try to package for HL2 I get the following:
ERROR: Missing precompiled manifest for 'HttpGPTChatModule', 'C:\Program Files\Epic Games\UE_5.0\Engine\Plugins\Marketplace\HttpGPT\Intermediate\Build\HoloLens\arm64\UnrealGame\Development\HttpGPTChatModule\HttpGPTChatModule.precompiled'. This module was most likely not flagged for being included in a precompiled build - set 'PrecompileForTargets = PrecompileTargetsType.Any;' in HttpGPTChatModule.build.cs to override. If part of a plugin, also check if its 'Type' is correct.
I have tried editing the instructed files and it only resulted in other errors. I can package the same exact project to Windows/ other platforms without issues. My teammates informed me this issue is probably on the plugin side and it seems like a new build target needs to be added. Is there something you can do to fix this?
The access problem of a single window is suspected to be restricted. After asking a few questions, there is no response. It can be used normally after restarting, but it is not related to the problem of the previous window.
Hi @lucoiso , I noticed that GPT api actually replies word by word. And to generate the msg string, it would take a while for GPT to reply all the words. For example, a 200 words paragaph, it might take GPT 10 seconds to reply the whole paragraph.
So is it possible to get the response sentence by sentence, for example I could set a range, generate a message string when the accumulated response has been larger than 10 words, and when they system comes across symbols such as [, . ! ?] ?
The reason to do that is to reduce the wait time. As long as I got a senetence, I'll feed the message to Azure to get audio and blendshapes, instead of waiting for the whole thing to be generated.
Open ai it has update the models, maybe you can add this to the setting, i am really bad with c++ code yet to modify the plugin, and i guess you can add it, or maybe let an open variable to put an string in case Openai will release more updates in the future.
Hello, thank you for sharing this great work! I tested using the plugin on OSX and on Windows 10, and I noticed that on OSX the HttpGPT Image Generator option doesn't pop up in the Unreal Editor. I'm running Unreal 5 on both machines. Cheers
Log:
[2023.04.18-02.18.26:073][ 91]LogHttpGPT_Internal: Display: OnProgressCompleted (37236): Process Completed [2023.04.18-02.18.26:073][ 91]LogHttpGPT_Internal: Display: OnProgressCompleted (37236): Content: { "error": { "message": "'User' is not one of ['system', 'assistant', 'user'] - 'messages.0.role'", "type": "invalid_request_error", "param": null, "code": null } }
[2023.04.18-02.18.26:073][ 91]LogHttpGPT: Error: OnProgressCompleted (37236): Request failed
Thanks Snappy07 for letting me know about this! :)
Unhandled Exception: EXCEPTION_ACCESS_VIOLATION reading address 0x0000006400000063
VCRUNTIME140
UnrealEditor_Core
UnrealEditor_HttpGPTEditor!SHttpGPTChatItem::GetMessageText() [D:\build\U5M-Marketplace\Sync\LocalBuilds\PluginTemp\HostProject\Plugins\HttpGPT\Source\HttpGPTEditor\Private\SHttpGPTChatView.cpp:108]
UnrealEditor_HttpGPTEditor!TBaseSPMethodDelegateInstance<1,SHttpGPTChatItem const ,1,FText __cdecl(void),FDefaultDelegateUserPolicy>::Execute() [D:\RocketSync\5.1.0-23058290+++UE5+Release-5.1\Working\Engine\Source\Runtime\Core\Public\Delegates\DelegateInstancesImpl.h:295]
UnrealEditor_Slate
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_SlateCore
UnrealEditor_Slate
UnrealEditor_Slate
UnrealEditor_Slate
UnrealEditor_Slate
UnrealEditor_Slate
UnrealEditor_Slate
UnrealEditor
UnrealEditor
UnrealEditor
UnrealEditor
UnrealEditor
UnrealEditor
kernel32
ntdll
Feature Request.
Is it posible to dd Stable Diffusion through using API key?
It will be great if we can use it with Automatic1111 API
Greetings,
HttpGPT.uplugin (1.5.4, HttpGPT_UE51.zip file) uses the 4.27 term "WhitelistPlatforms" instead of the 5.x term "PlatformAllowList."
This shouldn't cause a problem for most users, since Unreal 5.x will ignore "WhitelistPlatforms" in the UPLUGIN file. However, I thought I should report it.
Thank you again for writing HttpGPT,
Jen
It seems that if I talk too many rounds, there will always be a error like this, and after this error, no more conversation could be made
Create an editor tool to integrate ChatGPT directly in the engine.
All text is replaced when updated and the text block is losing the selected reference causing a strange behavior on selection.
Hi, I can't use this plugin due insufficient_quota issue, is this plugin related issue? I've changed only API and user in plugin options.
Error Details:
Error Code: None
Error Type: insufficient_quota
Error Message: You exceeded your current quota, please check your plan and billing details.
Here is more detailed log:
ogHttpGPT: Display: Activate (115036): Activating task
LogHttpGPT_Internal: Display: InitializeRequest (115036): Initializing request object
LogHttpGPT_Internal: Display: SetRequestContent (115036): Mounting content
LogHttpGPT_Internal: Display: SetRequestContent (115036): Selected model supports Chat API. Mounting section history.
LogHttpGPT_Internal: Display: BindRequestCallbacks (115036): Binding callbacks
LogHttpGPT: Display: SendRequest (115036): Sending request
LogHttpGPT: Display: SendRequest (115036): Request sent
LogHttpGPT_Internal: Display: OnProgressUpdated (115036): Progress Updated
LogHttpGPT_Internal: Display: OnProgressUpdated (115036): Content: {
"error": {
"message": "You exceeded your current quota, please check your plan and billing details.",
"type": "insufficient_quota",
"param": null,
"code": null
}
}
; Bytes Sent: 1563; Bytes Received: 206
LogHttpGPT_Internal: Display: OnProgressCompleted (115036): Process Completed
LogHttpGPT_Internal: Display: OnProgressCompleted (115036): Content: {
"error": {
"message": "You exceeded your current quota, please check your plan and billing details.",
"type": "insufficient_quota",
"param": null,
"code": null
}
}
LogHttpGPT: Error: OnProgressCompleted (115036): Request failed
LogHttpGPT: Display: SetReadyToDestroy (115036): Setting task as Ready to Destroy
Hi Everyone,
I'm having an issue with the ChatGPT Plugin in UNreal Engine 5.1.1.
I enabled the plugin and turned on the Internal Logs as the assistant said but it still not working and keep saying to me to enable the internal logs.
Any suggestions?
When the request fails to be sent, the assistant message doesn't update and still displays the text "Waiting response...". To improve user experience, update the text to display an error message if one is returned and inform the user that more information is available in the internal logs.
I sometimes experience a crash when I switch map in the Editor:
If any data was lost, recreating will restore them in a next response.
Issue created based on these comments:
Just like ChatGPT, add multiple chats instead just one.
I'm not sure what's wrong. I can access it in tools, but it does not work. And when I go to editor preferences I cant find the httpchatgpt plug in
The tasks initialized by the editor tool are being destroyed in these situations:
Is there a place that we could set the system role? Just like below or the playgournd in chat mode, that you could set various roles and cusotmize the response.
openai.ChatCompletion.create(
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?"}
]
)
the plugin doesn't seem to work
and I don't know where to find the logs
https://platform.openai.com/docs/api-reference/chat/create
Include the new macro in source files.
// Used to inline generated cpp files from UObject headers
#define UE_INLINE_STRINGIFY(name) #name
#if UE_DISABLE_INLINE_GEN_CPP
#define UE_INLINE_GENERATED_CPP_BY_NAME(name) UE_INLINE_STRINGIFY(CoreTypes.h)
#else
#define UE_INLINE_GENERATED_CPP_BY_NAME(name) UE_INLINE_STRINGIFY(name.gen.cpp)
#endif
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.