TinyTorch/docs/tito/milestones.md

Milestone System

Recreate ML History with YOUR Code

Run the algorithms that changed the world using the TinyTorch you built from scratch

Purpose: The milestone system lets you run famous ML algorithms (1957-2018) using YOUR implementations. Every milestone validates that your code can recreate a historical breakthrough.

See Historical Milestones for the full historical context and significance of each milestone.

What Are Milestones?

Milestones are runnable recreations of historical ML papers that use YOUR TinyTorch implementations:

• 1957 - Rosenblatt's Perceptron: The first trainable neural network
• 1969 - XOR Solution: Solving the problem that stalled AI
• 1986 - Backpropagation: The MLP revival (Rumelhart, Hinton & Williams)
• 1998 - LeNet: Yann LeCun's CNN breakthrough
• 2017 - Transformer: "Attention is All You Need" (Vaswani et al.)
• 2018 - MLPerf: Production ML benchmarks

Each milestone script imports YOUR code from the TinyTorch package you built.

Quick Start

Typical workflow:

# 1. Build the required modules (e.g., Foundation Tier for Milestone 03)
tito module complete 01  # Tensor
tito module complete 02  # Activations
tito module complete 03  # Layers
tito module complete 04  # Losses
tito module complete 05  # Autograd
tito module complete 06  # Optimizers
tito module complete 07  # Training

# 2. See what milestones you can run
tito milestone list

# 3. Get details about a specific milestone
tito milestone info 03

# 4. Run it!
tito milestone run 03

Essential Commands

Discover Milestones

List All Milestones

tito milestone list

Shows all 6 historical milestones with status:

• 🔒 LOCKED - Need to complete required modules first
• 🎯 READY TO RUN - All prerequisites met!
• ✅ COMPLETE - You've already achieved this

Simple View (compact list):

tito milestone list --simple

Learn About Milestones

Get Detailed Information

tito milestone info 03

Shows:

• Historical context (year, researchers, significance)
• Description of what you'll recreate
• Required modules with ✓/✗ status
• Whether you're ready to run it

Run Milestones

Run a Milestone

tito milestone run 03

What happens:

1. Checks prerequisites - Validates required modules are complete
2. Tests imports - Ensures YOUR implementations work
3. Shows context - Historical background and what you'll recreate
4. Runs the script - Executes the milestone using YOUR code
5. Tracks achievement - Records your completion
6. Celebrates! - Shows achievement message 🏆

Skip prerequisite checks (not recommended):

tito milestone run 03 --skip-checks

Track Progress

View Milestone Progress

tito milestone status

Shows:

• How many milestones you've completed
• Your overall progress (%)
• Unlocked capabilities
• Next milestone ready to run

Visual Timeline

tito milestone timeline

See your journey through ML history in a visual tree format.

The 6 Milestones

Milestone 01: Perceptron (1957) 🧠

What: Frank Rosenblatt's first trainable neural network

Requires: Module 01 (Tensor)

What you'll do: Implement and train the perceptron that proved machines could learn

Historical significance: First demonstration of machine learning

Run it:

tito milestone info 01
tito milestone run 01

---

Milestone 02: XOR Crisis (1969) 🔀

What: Solving the problem that stalled AI research

Requires: Modules 01-02 (Tensor, Activations)

What you'll do: Use multi-layer networks to solve XOR - impossible for single-layer perceptrons

Historical significance: Minsky & Papert showed perceptron limitations; this shows how to overcome them

Run it:

tito milestone info 02
tito milestone run 02

---

Milestone 03: MLP Revival (1986) 🎓

What: Backpropagation breakthrough - train deep networks on MNIST

Requires: Modules 01-07 (Complete Foundation Tier)

What you'll do: Train a multi-layer perceptron to recognize handwritten digits (95%+ accuracy)

Historical significance: Rumelhart, Hinton & Williams (Nature, 1986) - the paper that reignited neural network research

Run it:

tito milestone info 03
tito milestone run 03

---

Milestone 04: CNN Revolution (1998) 👁️

What: LeNet - Computer Vision Breakthrough

Requires: Modules 01-09 (Foundation + Spatial/Convolutions)

What you'll do: Build LeNet for digit recognition using convolutional layers

Historical significance: Yann LeCun's breakthrough that enabled modern computer vision

Run it:

tito milestone info 04
tito milestone run 04

---

Milestone 05: Transformer Era (2017) 🤖

What: "Attention is All You Need"

Requires: Modules 01-13 (Foundation + Architecture Tiers)

What you'll do: Implement transformer architecture with self-attention mechanism

Historical significance: Vaswani et al. revolutionized NLP and enabled GPT/BERT/modern LLMs

Run it:

tito milestone info 05
tito milestone run 05

---

Milestone 06: MLPerf Benchmarks (2018) 🏆

What: Production ML Systems

Requires: Modules 01-19 (Foundation + Architecture + Optimization Tiers)

What you'll do: Optimize for production deployment with quantization, compression, and benchmarking

Historical significance: MLPerf standardized ML system benchmarks for real-world deployment

Run it:

tito milestone info 06
tito milestone run 06

---

Prerequisites and Validation

How Prerequisites Work

Each milestone requires specific modules to be complete. The run command automatically validates:

Module Completion Check:

tito milestone run 03

🔍 Checking prerequisites for Milestone 03...
  ✓ Module 01 - complete
  ✓ Module 02 - complete
  ✓ Module 03 - complete
  ✓ Module 04 - complete
  ✓ Module 05 - complete
  ✓ Module 06 - complete
  ✓ Module 07 - complete

✅ All prerequisites met!

Import Validation:

🧪 Testing YOUR implementations...
  ✓ Tensor import successful
  ✓ Activations import successful
  ✓ Layers import successful

✅ YOUR TinyTorch is ready!

If Prerequisites Are Missing

You'll see a helpful error:

❌ Missing Required Modules

Milestone 03 requires modules: 01, 02, 03, 04, 05, 06, 07
Missing: 05, 06, 07

Complete the missing modules first:
  tito module start 05
  tito module start 06
  tito module start 07

Achievement Celebration

When you successfully complete a milestone, you'll see:

╔════════════════════════════════════════════════╗
║  🎓 Milestone 03: MLP Revival (1986)           ║
║  Backpropagation Breakthrough                  ║
╚════════════════════════════════════════════════╝

🏆 MILESTONE ACHIEVED!

You completed Milestone 03: MLP Revival (1986)
Backpropagation Breakthrough

What makes this special:
• Every line of code: YOUR implementations
• Every tensor operation: YOUR Tensor class
• Every gradient: YOUR autograd

Achievement saved to your progress!

🎯 What's Next:
Milestone 04: CNN Revolution (1998)
Unlock by completing modules: 08, 09

Understanding Your Progress

Three Tracking Systems

TinyTorch tracks progress in three ways (all are related but distinct):

1. Module Completion (tito module status)

• Which modules (01-20) you've implemented
• Tracked in .tito/progress.json
• Required for running milestones

2. Milestone Achievements (tito milestone status)

• Which historical papers you've recreated
• Tracked in .tito/milestones.json
• Unlocked by completing modules + running milestones

3. Capability Checkpoints (tito checkpoint status) - OPTIONAL

• Gamified capability tracking
• Tracked in .tito/checkpoints.json
• Purely motivational; can be disabled

Relationship Between Systems

Complete Modules (01-07)
    ↓
Unlock Milestone 03
    ↓
Run: tito milestone run 03
    ↓
Achievement Recorded
    ↓
Capability Unlocked (optional checkpoint system)

Tips for Success

1. Complete Modules in Order

While you can technically skip around, the tier structure is designed for progressive learning:

• Foundation Tier (01-07): Required for first milestone
• Architecture Tier (08-13): Build on Foundation
• Optimization Tier (14-19): Build on Architecture

2. Test as You Go

Before running a milestone, make sure your modules work:

# After completing a module
tito module complete 05

# Test it works
python -c "from tinytorch import Tensor; print(Tensor([[1,2]]))"

3. Use Info Before Run

Learn what you're about to do:

tito milestone info 03  # Read the context first
tito milestone run 03   # Then run it

4. Celebrate Achievements

Share your milestones! Each one represents recreating a breakthrough that shaped modern AI.

Troubleshooting

"Import Error" when running milestone

Problem: Module not exported or import failing

Solution:

# Re-export the module
tito module complete XX

# Test import manually
python -c "from tinytorch import Tensor"

"Prerequisites Not Met" but I completed modules

Problem: Progress not tracked correctly

Solution:

# Check module status
tito module status

# If modules show incomplete, re-run complete
tito module complete XX

Milestone script fails during execution

Problem: Bug in your implementation

Solution:

1. Check error message for which module failed
2. Edit modules/source/XX_name/ (NOT tinytorch/)
3. Re-export: tito module complete XX
4. Run milestone again

Next Steps

Ready to Recreate ML History?

Start with the Foundation Tier and work toward your first milestone

Foundation Tier → Historical Context →

---

Every milestone uses YOUR code. Every achievement is proof you understand ML systems deeply. Build from scratch, recreate history, master the fundamentals.