/**
* Interactive ML History Timeline
* Handles popup functionality for milestone cards
*/
document.addEventListener('DOMContentLoaded', function() {
const timelineData = {
perceptron: {
year: "1958",
title: "The Perceptron",
researcher: "Frank Rosenblatt",
subtitle: "The first trainable neural network proves machines can learn from data",
achievement: "Binary classification with gradient descent",
architecture: "Input → Linear → Sigmoid → Output",
whatYouBuild: [
"Binary classification with gradient descent",
"Simple but revolutionary architecture",
"YOUR Linear layer recreates history"
],
systemsInsights: [
"Memory: O(n) parameters",
"Compute: O(n) operations",
"Limitation: Only linearly separable problems"
],
modules: "After Modules 02-04",
expectedResults: "~50% (untrained) → 95%+ (trained) accuracy",
commands: [
"tito milestone run perceptron"
]
},
xor: {
year: "1969",
title: "The XOR Crisis",
researcher: "Minsky & Papert",
subtitle: "Hidden layers solve non-linear problems that nearly ended AI research",
achievement: "Non-linear learning through hidden representations",
architecture: "Input → Linear → ReLU → Linear → Output",
whatYouBuild: [
"Hidden layers enable non-linear solutions",
"Multi-layer networks break through limitations",
"YOUR autograd makes it possible"
],
systemsInsights: [
"Memory: O(n²) with hidden layers",
"Compute: O(n²) operations",
"Breakthrough: Hidden representations"
],
modules: "After Modules 02-06",
expectedResults: "50% (single layer) → 100% (multi-layer) on XOR",
commands: [
"tito milestone run xor"
]
},
mlp: {
year: "1986",
title: "MLP Revival",
researcher: "Backpropagation Era",
subtitle: "Backpropagation enables training deep networks on real datasets",
achievement: "Multi-class digit recognition",
architecture: "Images → Flatten → Linear → ReLU → Linear → ReLU → Linear → Classes",
whatYouBuild: [
"Multi-class digit recognition",
"Complete training pipelines",
"YOUR optimizers achieve 95%+ accuracy"
],
systemsInsights: [
"Memory: ~100K parameters for MNIST",
"Compute: Dense matrix operations",
"Architecture: Multi-layer feature learning"
],
modules: "After Modules 02-08",
expectedResults: "95%+ accuracy on MNIST",
commands: [
"tito milestone run mlp"
]
},
cnn: {
year: "1998",
title: "CNN Revolution",
researcher: "Yann LeCun",
subtitle: "CNNs exploit spatial structure for computer vision—enabling modern AI",
achievement: "Spatial intelligence for computer vision",
architecture: "Images → Conv → ReLU → Pool → Conv → ReLU → Pool → Flatten → Linear → Classes",
whatYouBuild: [
"Convolutional feature extraction",
"Natural image classification (CIFAR-10)",
"YOUR Conv2d + MaxPool2d unlock spatial intelligence"
],
systemsInsights: [
"Memory: ~1M parameters (weight sharing reduces vs dense)",
"Compute: Convolution is intensive but parallelizable",
"Architecture: Local connectivity + translation invariance"
],
modules: "After Modules 02-09",
expectedResults: "75%+ accuracy on CIFAR-10 ✨",
commands: [
"tito milestone run cnn"
],
northStar: true
},
transformer: {
year: "2017",
title: "Transformer Era",
researcher: "Vaswani et al.",
subtitle: "Attention mechanism launches the LLM revolution (GPT, BERT, ChatGPT)",
achievement: "Self-attention for language understanding",
architecture: "Tokens → Embeddings → Attention → FFN → ... → Attention → Output",
whatYouBuild: [
"Self-attention mechanisms",
"Autoregressive text generation",
"YOUR attention implementation generates language"
],
systemsInsights: [
"Memory: O(n²) attention requires careful management",
"Compute: Highly parallelizable",
"Architecture: Long-range dependencies"
],
modules: "After Modules 02-13",
expectedResults: "Loss < 1.5, coherent responses to questions",
commands: [
"tito milestone run transformer"
]
},
olympics: {
year: "2018",
title: "MLPerf Torch Olympics",
researcher: "MLCommons (founded 2018)",
subtitle: "Systematic optimization becomes essential as models grow larger",
achievement: "Production-ready optimization",
architecture: "Profile → Compress → Accelerate",
whatYouBuild: [
"Performance profiling and bottleneck analysis",
"Model compression (quantization + pruning)",
"Inference acceleration (KV-cache + batching)"
],
systemsInsights: [
"Memory: 4-16× compression through quantization/pruning",
"Speed: 12-40× faster generation with KV-cache + batching",
"Workflow: Systematic 'measure → optimize → validate' methodology"
],
modules: "After Modules 14-18",
expectedResults: "8-16× smaller models, 12-40× faster inference",
commands: [
"tito milestone run mlperf"
]
}
};
// Create popup HTML if not exists
let popup = document.getElementById('ml-timeline-popup');
if (!popup) {
popup = document.createElement('div');
popup.id = 'ml-timeline-popup';
popup.className = 'ml-timeline-popup';
popup.innerHTML = '';
document.body.appendChild(popup);
}
// Handle clicks on timeline items
document.querySelectorAll('.ml-timeline-content').forEach(card => {
card.addEventListener('click', function(e) {
const item = this.closest('.ml-timeline-item');
const milestoneType = item.classList[1]; // Get the milestone class (perceptron, xor, etc.)
const data = timelineData[milestoneType];
if (!data) return;
const popupContent = popup.querySelector('.ml-timeline-popup-content');
popupContent.innerHTML = `
${data.title}
${data.northStar ? '🎯 North Star Achievement — This is a major milestone in your TinyTorch journey!
' : ''}
`;
popup.classList.add('active');
// Close button handler
const closeBtn = popupContent.querySelector('.ml-timeline-popup-close');
closeBtn.addEventListener('click', function() {
popup.classList.remove('active');
});
});
});
// Close popup on background click
popup.addEventListener('click', function(e) {
if (e.target === popup) {
popup.classList.remove('active');
}
});
// Close popup on ESC key
document.addEventListener('keydown', function(e) {
if (e.key === 'Escape' && popup.classList.contains('active')) {
popup.classList.remove('active');
}
});
});