| import gradio as gr |
| import transformers |
| import numpy as np |
|
|
|
|
| from transformers import DistilBertTokenizer, TFAutoModelForSequenceClassification |
|
|
| tokenizer = DistilBertTokenizer.from_pretrained("Gokce/finetuned_distilBERT_forDetecting_GeneratedReviews_TR") |
| model = TFAutoModelForSequenceClassification.from_pretrained("Gokce/finetuned_distilBERT_forDetecting_GeneratedReviews_TR") |
|
|
|
|
| MAX_LENGTH = 128 |
|
|
| def predict(text): |
| inputs = tokenizer.encode_plus(text, add_special_tokens=True, return_tensors="tf", padding=True, truncation=True, max_length=MAX_LENGTH) |
|
|
| predictions = model.predict([inputs['input_ids'], inputs['attention_mask']]) |
| predicted_class_idx = np.argmax(predictions[0]) |
|
|
| if predicted_class_idx == 0: |
| predicted_class = "Human" |
| else: |
| predicted_class = "AI" |
|
|
| return predicted_class |
|
|
|
|
| iface = gr.Interface( |
| fn=predict, |
| inputs=gr.inputs.Textbox(lines=5, placeholder="Enter text...", label='Review Text'), |
| outputs=gr.outputs.Textbox(label='Detected Label'), |
| title='Generated Text Classifier for Restaurant Reviews' |
| ) |
| iface.launch() |
|
|
|
|
|
|
|
|
