Views
No views yet
pip install transformers torch1
2from transformers import BertForSequenceClassification, BertTokenizer
3import torch
4
5# Load quantized model
6quantized_model_path = "AventIQ-AI/sentiment-analysis-for-fake-news-detection"
7quantized_model = BertForSequenceClassification.from_pretrained(quantized_model_path)
8quantized_model.eval() # Set to evaluation mode
9quantized_model.half() # Convert model to FP16
10
11# Load tokenizer
12tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
13
14# Define a test sentence
15test_sentence = "The government has completely failed its people again! Hospitals are overflowing, food prices are out of control, and yet officials are busy attending luxury summits abroad. It's clear they don't care about ordinary citizens anymore—only their power and perks. This country is on the brink, and no one in charge is doing anything about it."
16
17# Tokenize input
18inputs = tokenizer(test_sentence, return_tensors="pt", padding=True, truncation=True, max_length=128)
19
20# Ensure input tensors are in correct dtype
21inputs["input_ids"] = inputs["input_ids"].long() # Convert to long type
22inputs["attention_mask"] = inputs["attention_mask"].long() # Convert to long type
23
24# Make prediction
25with torch.no_grad():
26 outputs = quantized_model(**inputs)
27
28# Get predicted class
29predicted_class = torch.argmax(outputs.logits, dim=1).item()
30print(f"Predicted Class: {predicted_class}")
31
32
33label_mapping = {0: "very_negative", 1: "negative", 2: "neutral", 3: "positive", 4: "very_positive"} # Example
34
35predicted_label = label_mapping[predicted_class]
36print(f"Predicted Label: {predicted_label}")
37.
├── model/ # Contains the quantized model files
├── tokenizer_config/ # Tokenizer configuration and vocabulary files
├── model.safensors/ # Fine Tuned Model
├── README.md # Model documentation