AI PLAYGROUND
Handwritten digit
recognition
Draw a digit or upload an image. Your trained model will do the rest.
MNIST classifier
INPUT
Create your sample
Draw a digit here
20 px
Mouse, touch or stylus
PREDICTION
Recognition result
Your result will appear here
Draw or upload a digit, then predict.
MODEL RECOGNIZED
Digit -
Predicted from your input sample
Model confidence
--%
Confidence is the model's probability for its top prediction.
PROBABILITY DISTRIBUTION
All digit classes
Chart.js could not load. Prediction values are still available after a successful request.
SESSION ACTIVITY
Recent predictions
Your predictions will show up here.
EXPERIMENT REPORT
Model evaluation
Held-out test results from the training notebook, separate from live predictions.
Loading experiment results...
DATA AND PREPROCESSING
MNIST dataset
EXPERIMENT SETUP
Hyperparameters and rationale
NOTEBOOK CONFIGURATIONS
Architectures and design rationale
Chart.js could not load; the recorded values and saved plots remain available.
VALIDATION + HELD-OUT TEST
ANN and CNN comparison
| Model | Validation accuracy | Test accuracy | Macro precision | Macro recall | Macro F1 |
|---|
TRAINING HISTORY
Accuracy by epoch
TRAINING HISTORY
Loss by epoch
SELECTED MODEL
Test-set metrics
SELECTED MODEL
Confusion matrix
Counts are from the original 10,000-image MNIST test partition.
SAVED NOTEBOOK PLOT
Training and validation curves
SAVED NOTEBOOK PLOT