| # -*- coding: utf-8 -*- | |
| """ | |
| Spyder Editor | |
| This is a temporary script file. | |
| """ | |
| from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline | |
| import torch | |
| tokenizer = AutoTokenizer.from_pretrained("nomsgadded/opt_RestaurantReview") | |
| model = AutoModelForSequenceClassification.from_pretrained("nomsgadded/opt_RestaurantReview", num_labels=9) | |
| prefix = "##Rating: " | |
| text1 = "Bad" | |
| text2 = "It was really nice to dine there, however the waiter is very mean." | |
| text3 = "Nice" | |
| def inference(text): | |
| text = prefix + text | |
| inputs = tokenizer(text, return_tensors="pt") | |
| classifier = pipeline("text-classification", model=model, tokenizer=tokenizer) | |
| with torch.no_grad(): | |
| logits = model(**inputs).logits | |
| predicted_class_id = logits.argmax().item() | |
| print((predicted_class_id+2)/2) | |
| inference(text1) | |
| inference(text2) | |
| inference(text3) |