Gemma-2-2b μ΄λ±νμ κΈ λ³νκΈ°
π» Model Description
- Gemma2-2b νκ΅ μ΄λ±νμ κΈ λ³νκΈ°λ μ λ ₯ν κΈμ μ΄λ±νμμ΄ μ΄ κ² κ°μ κΈλ‘ λ³νν΄ μ£Όλ λͺ¨λΈμ λλ€.
- Gemma2-2b-it λͺ¨λΈμ base modelλ‘ μ¬μ©νμκ³ LoRAκΈ°λ²μ μ¬μ©νμ¬ ν¨μ¨μ μΌλ‘ fine-tuning νμμ΅λλ€.
(Gemma2-2b is a model that transforms input text to resemble the writing style of an elementary school student. It is based on the Gemma2-2b-it model and was fine-tuned efficiently using the LoRA technique.)
π¦Usage
import transformers
from huggingface_hub import notebook_login
notebook_login()
BASE_MODEL = "skwh54/Gemma-2-2b-it-elementary-style-document"
model = AutoModelForCausalLM.from_pretrained(BASE_MODEL, device_map={"":0})
tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL)
π Example
Input | Output |
---|---|
μλ νμΈμ. μ λ μΉκ΅¬μ 곡μμμ λ§λμ λμμ΄μ. λ μ¨κ° μ’μμ λκΈ° μ’μμ΄μ. λ§μλ μμλ€λ λ§μ΄ λ¨Ήμμ΄μ. λ΄μΌλ λκ³ μΆλ€! | γ γ λ μΉκ΅¬λ 곡μμμ λ§λμ λμλλ° λ μ¨κ° μ’μμ λ μ’μκ³ λ§μλ μμλ λ§μ΄ λ¨Ήμμ λ΄μΌλ λκ³ μΆλ€ |
μ€λμ νμ¬μμ μ€μν νλ‘μ νΈ νμλ₯Ό νλ€. μ€λΉν λ§νΌ μ’μ νΌλλ°±μ λ°μμ λΏλ―νλ€. ν΄κ·Ό νμλ μ΄λμ νκ³ λμ μ§μμ μ± μ μ½μΌλ©° ν루λ₯Ό λ§λ¬΄λ¦¬νλ€. | μ€λ νμ¬μμ μ€μν νλ‘μ νΈ νμν¨ γ γ μ€λΉν λ§νΌ μ’μ νΌλλ°± λ°μμ λΏλ―ν¨ ν΄κ·Ό νμ μ΄λνκ³ μ§μ μ€λ©΄ μ± μ½μΌλ©° ν루 λ§λ¬΄λ¦¬ν¨ |
π Training data
- korean_smile_style_datasetμ μ¬μ©νμμ΅λλ€.
- λ³Έ λ°μ΄ν°λ Smilegate AIμμ 곡κ°νλ νκ΅μ΄ 문체 μ€νμΌ λ³ν "SmileStyle" λ°μ΄ν°μ μ λλ€.
- μ¬λ¬ μ€νμΌμ λ¬Έμ²΄κ° μ‘΄μ¬νλ©° μ΄ μ€μμ formalκ³Ό choding μ€νμΌμ λ°μ΄ν°λ§μ μΆμΆνμ¬ μ¬μ©νμμ΅λλ€.
(The korean_smile_style_dataset was used for training. This dataset, "SmileStyle," is provided by Smilegate AI and includes various writing styles in Korean. Only the formal and elementary school styles were used in this model.)
πββοΈββ‘οΈ Coworker
- Downloads last month
- 12