---
license: gemma
datasets:
- jojo0217/korean_safe_conversation
language:
- ko
- en
base_model:
- google/gemma-2-2b-it
---
# Toadx2 LLM Model | Gemma2 Fine-Tune
- Huggingface Korean Dataset & Custom Preprocessed KB Real Estate Data
- Google Machine Learning Bootcamp 2024, 5th

<br />

by. Kim Basilri

---

<br/>

### What is this project?
- This project aims to create a web application that predicts future real estate prices by utilizing machine learning techniques through Gemma2 model based on Korean real estate price data. This allows users to make clear and smart judgments through data-based future prediction results without being swept away by the uncertain trend of the Korean real estate market.

<br />

### Based Model
- [Google's Gemma2-2-2b-it](https://huggingface.co/google/gemma-2-2b-it)

<br />

### Fine Tuned Model
- [basilry/gemma2-2-2b-it-fine-tuned-korean-real-estate-model](https://huggingface.co/basilry/gemma2-2-2b-it-fine-tuned-korean-real-estate-model)

<br />

### Used Dataset
1. [Korean Safe Conversation Dataset](https://huggingface.co/datasets/jojo0217/korean_safe_conversation)

2. [KB Real Estate Data Hub's Apartment Dataset](https://data.kbland.kr/)

3. [S.Korea Apartment Market Prediction Dataset](https://github.com/basilry/toadx2_api)

<br />

### Web Application Github Address
1. [ToadX2 Frontend Github](https://github.com/basilry/toadx2_fe)

2. [ToadX2 Backend Github](https://github.com/basilry/toadx2_api.git)

<br />

### Finetuning Environment
1. Google Colab

2. Huggingface Transformers