DaramGPT inference example

infer.py

from transformers import AutoTokenizer, AutoModelForCausalLM, GPTJConfig, GPTJForCausalLM
import torch

tokenizer = AutoTokenizer.from_pretrained("d:/MODEL/DaramGPT")
model = GPTJForCausalLM.from_pretrained("d:/MODEL/DaramGPT")

tokens = tokenizer("서울의 날씨는 ")

print(tokenizer.decode(model.generate(**{
    "input_ids": torch.tensor([tokens["input_ids"]]),
    "attention_mask": torch.tensor([tokens["attention_mask"]]),
    "max_length": 256,
    "repetition_penalty": 1.1
})[0], skip_special_tokens=True))