AI Model Playground

MNIST Digit Classifier

Draw a digit and see the neural network's prediction

Prediction Results

Draw a digit to see predictions

Model Information

Model Type: Neural Network

Architecture: Flatten → Dense(128, ReLU) → Dense(10, Softmax)

Test Accuracy: ~98%

Sentiment Analyzer

Enter text to analyze its sentiment (positive/negative)

Enter your text:

Enter some text to analyze sentiment

Model Information

Model Type: Naive Bayes Classifier

Features: Bag-of-words with CountVectorizer

Training Data: Custom sentiment examples

How It Works

MNIST Classifier

The MNIST classifier uses a simple neural network with one hidden layer (128 neurons) trained on 60,000 handwritten digits. Draw a digit and see the model's prediction confidence for each digit (0-9).

Sentiment Analysis

The sentiment analyzer uses a Naive Bayes classifier trained on example sentences. It analyzes the words in your text to predict whether the sentiment is positive (1) or negative (0).

Technical Details

This demo simulates the behavior of TensorFlow/Keras models for MNIST and scikit-learn for NLP. In a real implementation, you would connect to a backend running the actual models.