This project generate everytime community style korean text using GPT-2 model.
Original github: jason9693/UCK-GPT2
Original notebook: Colab
* First, Fill what the character will say in text. This will be base of script.
* Second, Choose text style category.
* And then, Fill number in length. Text is created as long as "length". I recommend between 100 and 300.
* If length is so big, generate time will be long.
* Finally, select the desired type and send the request. ['natural', 'fix-length']
* 'natural' stops generating when a newline occurs, generating a natural sentence. With 'fix-length', sentences as long as the length are generated.
text: The base of generated text.
category: The text style category. [숭실대 에타, 대학생 잡담방, 모두의 연애]
length: The size of generated text.
{"0": generated text(utf-8, korean)}
Request url: https://master-gpt2-everytime-fpem123.endpoint.ainize.ai/everytime/natural
curl -X POST "https://master-gpt2-everytime-fpem123.endpoint.ainize.ai/everytime/natural" -H "accept: application/json" -H "Content-Type: multipart/form-data" -F "text=님들 오늘 수업" -F "category=숭실대 에타" -F "length=100"
{
"0": "님들 오늘 수업시간끝나고 프린트할수있나요??"
}
Request url: https://master-gpt2-everytime-fpem123.endpoint.ainize.ai/everytime/fix-length
curl -X POST "https://master-gpt2-everytime-fpem123.endpoint.ainize.ai/everytime/fix-length" -H "accept: application/json" -H "Content-Type: multipart/form-data" -F "text=님들 오늘 수업" -F "category=대학생 잡담방" -F "length=100"
{
"0": "님들 오늘 수업끝나고 수업시간에 뭐하나요? ᄉᄇ\n어제 전공공부 다 ⁇ 끝나서 늦게까지 남아서 하고 나왔는데\n>아 ⁇
끝나갈시간이네 다들 과제끝났다 ⁇ 얘두라 저번에 술취한 애한테 카톡보냈는데 답장이 없네? 나랑 대화 가능? 이거 뭐냐 이 미친새끼... ⁇
와 진짜 자살생각하구 이 시간에 전화해서 내가 전화"
}
API page: Ainize
Demo page: End-point