gpt3, gpt4 supported.
Built in Chat API, web socket, context management
Clone the project to your local machine
git clone project_url
Open application.properties
file, set up your own configuration
Here is the sample:
# proxy = http://127.0.0.1:7890
open_ai_key = sk-xxx
max_token = 3000
timeout = 10000
open_ai_host = https://api.openai.com/
open_ai_model = gpt-3.5-turbo
redis_host = localhost
redis_port = 6379
#redis_password = 1234
Start your project
mvn spring-boot:run
To call the API
curl -X POST http://localhost:8080/app/chat?text=?text=your text&mid=yourmid&cid=yourcid
There are three query parameters you can use.
text
(required): Your prompts.
mid
: the message id you want to respond to.
cid
: conversation id (session id).
Response sample
{
"text": "Sure, what do you wonder about? I'm here to help you find answers to your questions or provide assistance in any way I can.",
"messageId": "b095773b-e207-4cca-a6ee-14cb417c3c02",
"conversationId": "e0c9fe58-bd2d-4f95-a431-e95955b2661c"
}
ws://localhost:8080/chat
To send stream to server, the request data format is in json string
{
"text": "hello world",
"mid": "adoasjdowadasdasdasdas",
"cid": "sdjaodjoasjokdaskdkaskdasp"
}
Response format is the same as REST API.