Korean GPT

한국어 GPT 모델입니다.

사용법

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained(
    "oz1115/korean-gpt-quick-test",
    trust_remote_code=True
)

model = AutoModelForCausalLM.from_pretrained(
    "oz1115/korean-gpt-quick-test",
    trust_remote_code=True
)

inputs = tokenizer("안녕하세요", return_tensors="pt")
outputs = model.generate(**inputs, max_length=50)
print(tokenizer.decode(outputs[0]))

모델 정보

Vocabulary: 32,000
Hidden Size: 512
Layers: 8
Attention Heads: 8