gamzaS2's picture
Upload README.md with huggingface_hub
0acede6 verified
---
language: ko
license: mit
library_name: transformers
tags:
- text-classification
- korean
- mental-health
- depression-detection
- bert
pipeline_tag: text-classification
---
# Korean Depression/Anxiety Detection Model
ํ•œ๊ตญ์–ด ํ…์ŠคํŠธ ๊ธฐ๋ฐ˜ ์šฐ์šธ/๋ถˆ์•ˆ ๊ฐ์ง€ ๋ชจ๋ธ์ž…๋‹ˆ๋‹ค.
## Model Description
- **Model Type:** BERT for Sequence Classification
- **Language:** Korean (ko)
- **Task:** Binary Classification (์ •์ƒ vs ์šฐ์šธ/๋ถˆ์•ˆ)
- **Base Model:** BERT (Korean)
## Labels
| Label | Description |
|-------|-------------|
| 0 | ์ •์ƒ (Normal) |
| 1 | ์šฐ์šธ/๋ถˆ์•ˆ (Depression/Anxiety) |
## Usage
```python
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
# ๋ชจ๋ธ ๋กœ๋“œ
tokenizer = AutoTokenizer.from_pretrained("YOUR_USERNAME/final_depression_model")
model = AutoModelForSequenceClassification.from_pretrained("YOUR_USERNAME/final_depression_model")
model.eval()
# ์˜ˆ์ธก
def predict(text):
inputs = tokenizer(text, return_tensors="pt", padding=True, truncation=True)
with torch.no_grad():
outputs = model(**inputs)
probs = torch.softmax(outputs.logits, dim=-1)
prediction = torch.argmax(probs, dim=-1).item()
return {
"label": prediction, # 0=์ •์ƒ, 1=์šฐ์šธ/๋ถˆ์•ˆ
"confidence": probs[0][prediction].item()
}
# ์‚ฌ์šฉ ์˜ˆ์‹œ
result = predict("์š”์ฆ˜ ๋„ˆ๋ฌด ํž˜๋“ค๊ณ  ์•„๋ฌด๊ฒƒ๋„ ํ•˜๊ธฐ ์‹ซ์–ด์š”")
print(result)
```
## Model Details
- **Architecture:** BertForSequenceClassification
- **Hidden Size:** 768
- **Attention Heads:** 12
- **Hidden Layers:** 12
- **Vocab Size:** 30,000
- **Max Position Embeddings:** 300
## Intended Use
์ด ๋ชจ๋ธ์€ ์ •์‹ ๊ฑด๊ฐ• ๊ด€๋ จ ์—ฐ๊ตฌ ๋ฐ ์ฑ—๋ด‡ ์„œ๋น„์Šค์—์„œ ์‚ฌ์šฉ์ž์˜ ๊ฐ์ • ์ƒํƒœ๋ฅผ ํŒŒ์•…ํ•˜๊ธฐ ์œ„ํ•œ ๋ชฉ์ ์œผ๋กœ ๊ฐœ๋ฐœ๋˜์—ˆ์Šต๋‹ˆ๋‹ค.
## Limitations
- ์ด ๋ชจ๋ธ์€ ์ „๋ฌธ์ ์ธ ์˜๋ฃŒ ์ง„๋‹จ ๋„๊ตฌ๊ฐ€ ์•„๋‹™๋‹ˆ๋‹ค.
- ์‹ค์ œ ์šฐ์šธ์ฆ/๋ถˆ์•ˆ์žฅ์•  ์ง„๋‹จ์€ ๋ฐ˜๋“œ์‹œ ์ „๋ฌธ ์˜๋ฃŒ์ง„๊ณผ ์ƒ๋‹ดํ•˜์„ธ์š”.
- ๋ชจ๋ธ์˜ ์˜ˆ์ธก ๊ฒฐ๊ณผ๋Š” ์ฐธ๊ณ ์šฉ์œผ๋กœ๋งŒ ์‚ฌ์šฉํ•ด์•ผ ํ•ฉ๋‹ˆ๋‹ค.
## License
MIT License