README.md

metadata

license: gemma
library_name: transformers
pipeline_tag: text-generation
extra_gated_heading: Access Gemma on Hugging Face
extra_gated_prompt: >-
  To access Gemma on Hugging Face, you’re required to review and agree to
  Google’s usage license. To do this, please ensure you’re logged in to Hugging
  Face and click below. Requests are processed immediately.
extra_gated_button_content: Acknowledge license
tags:
  - conversational
base_model:
  - google/gemma-3-270m-it

gemma3-270m-it-sms-verification_code_extraction

SMS에서 인증번호를 추출 하는 gemma3 SFT 모델

• 모바일 앱에서 사용하기 위한 Quantization 모델은 아래 참조.

  • https://huggingface.co/sg2023/gemma3-270m-it-sms-verification_code_extraction-int8-tflite
  • https://huggingface.co/sg2023/gemma3-270m-it-sms-verification_code_extraction-fp16-tflite

• Input / Output

  • "본인인증번호는 315611 입니다. 정확히 입력해주세요." -> 315611
  • "안녕하세요" -> 0

from transformers import Gemma3ForCausalLM, AutoTokenizer

repo_id = "sg2023/gemma3-270m-it-sms-verification_code_extraction"
tokenizer = AutoTokenizer.from_pretrained(repo_id)
model = Gemma3ForCausalLM.from_pretrained(repo_id)
model.eval()

prompt = "본인인증번호는 315611 입니다. 정확히 입력해주세요."
request_template = [{"role": "user", "content": prompt}]
response_template = tokenizer.apply_chat_template(request_template, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(response_template, return_tensors="pt").to(model.device)
input_ids = inputs["input_ids"][0]  # Tensor shape: (seq_len,)
input_len = input_ids.shape[0]

outputs = model.generate(**inputs, max_new_tokens=64)[0]
outputs = outputs[input_len:]
outputs = tokenizer.decode(outputs, skip_special_tokens=True)
outputs = outputs.strip()
print(outputs)  # 315611