|
|
--- |
|
|
license: mit |
|
|
tags: |
|
|
- indobert |
|
|
- sentiment-analysis |
|
|
- tokopedia |
|
|
- indo |
|
|
language: |
|
|
- id |
|
|
--- |
|
|
|
|
|
# IndoBERT Tokopedia Sentiment Classifier |
|
|
|
|
|
Model ini dilatih menggunakan IndoBERT (`indobenchmark/indobert-base-p1`) untuk klasifikasi sentimen komentar pelanggan Tokopedia (positif / negatif). |
|
|
|
|
|
## Dataset |
|
|
Dataset berupa komentar dari Tokopedia dengan label berdasarkan rating (≥4 = positif, ≤3 = negatif). |
|
|
|
|
|
## Penggunaan |
|
|
|
|
|
```python |
|
|
from transformers import BertTokenizer, BertForSequenceClassification |
|
|
import torch |
|
|
|
|
|
tokenizer = BertTokenizer.from_pretrained("username/indo-sentimen-tokopedia") |
|
|
model = BertForSequenceClassification.from_pretrained("username/indo-sentimen-tokopedia") |
|
|
|
|
|
text = "Barang sangat buruk dan tidak sesuai" |
|
|
inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True) |
|
|
|
|
|
with torch.no_grad(): |
|
|
logits = model(**inputs).logits |
|
|
probs = torch.nn.functional.softmax(logits, dim=1) |
|
|
pred = torch.argmax(probs) |
|
|
|
|
|
print("Sentimen:", "Positif" if pred.item() == 1 else "Negatif") |
|
|
|
