| import torch | |
| from transformers import BertTokenizer, BertForSequenceClassification | |
| def get_model(): | |
| model = BertForSequenceClassification.from_pretrained('bert-base-uncased', num_labels=2) | |
| return model | |
| # Predicting Function | |
| def predict(model, text, tokenizer): | |
| inputs = tokenizer(text, padding=True, truncation=True, return_tensors="pt") | |
| outputs = model(**inputs) | |
| predictions = torch.argmax(outputs.logits, dim=-1) | |
| return "AI-generated" if predictions.item() == 1 else "Human-written" | |