Examples

This section provides example scripts to illustrate how to use various functionalities of the textpredict package.

Simple Text Prediction

The following example demonstrates how to use the textpredict package to perform simple sentiment analysis and emotion detection.

from datasets import load_dataset
import textpredict as tp

# Function to test simple prediction using default model
def text_simple_prediction():
    # Sentiment analysis
    texts = ["I love this product!", "I love this product!"]
    model = tp.initialize(task="sentiment")
    result = model.analyze(texts, return_probs=False)
    print(f"Simple Prediction Result: {result}")

    # Emotion detection
    texts = ["I am happy today", "I am happy today"]
    model = tp.initialize(task="emotion")
    result = model.analyze(texts, return_probs=False)
    print(f"Emotion Prediction Result: {result}")

if __name__ == "__main__":
    text_simple_prediction()

Sequence Classification Training

The following example demonstrates how to use the SequenceClassificationTrainer class to train a sequence classification model.

from datasets import load_dataset
import textpredict as tp

# Function to test sequence classification trainer
def sequence_classification_training():
    dataset = load_dataset('imdb')
    trainer = tp.SequenceClassificationTrainer(model_name='bert-base-uncased', output_dir='./model', config=None, device='cpu')
    trainer.train_dataset = dataset['train']
    trainer.val_dataset = dataset['test']
    trainer.train()
    trainer.save()
    results = trainer.evaluate(dataset['test'])
    print(f"Training Results: {results}")

if __name__ == "__main__":
    sequence_classification_training()

Explainability

The following example demonstrates how to use the Explainability class to get feature importance for a given text.

from textpredict import initialize, Explainability

# Initialize the predictor
predictor = initialize(task='sentiment', device='cpu', model_name='bert-base-uncased')

# Explainability instance
explain = Explainability(model_name='bert-base-uncased', task='sentiment')

# Get feature importance
importance = explain.feature_importance("This is a great product!")
print(importance)

Benchmarking

The following example demonstrates how to use the Benchmarking class to benchmark a model.

from textpredict import Benchmarking
from datasets import load_dataset

def benchmark_model():
    dataset = load_dataset('imdb', split='test')
    benchmark = Benchmarking(model_name='bert-base-uncased')
    metrics = benchmark.benchmark(dataset)
    inference_time = benchmark.measure_inference_time(dataset)
    memory_usage = benchmark.measure_memory_usage(dataset)
    print(metrics, inference_time, memory_usage)

if __name__ == "__main__":
    benchmark_model()