diff --git a/book/chapters/00-introduction.md b/book/chapters/00-introduction.md index a5da6958..58e6f312 100644 --- a/book/chapters/00-introduction.md +++ b/book/chapters/00-introduction.md @@ -141,95 +141,119 @@ output = model(input) # YOU know exactly how this works --- -## What You'll Achieve: Complete ML Systems Mastery +## What You'll Achieve: Tier-by-Tier Mastery -### Immediate Achievements (Modules 1-8) -By Module 8, you'll have built a complete neural network framework from scratch: +### šŸ—ļø After Foundation Tier (Modules 01-07) +Build a complete neural network framework from mathematical first principles: ```python # YOUR implementation training real networks on real data model = Sequential([ - Linear(784, 128), # Your linear layer - ReLU(), # Your activation function - Linear(128, 64), # Your architecture design + Linear(784, 128), # Your linear algebra implementation + ReLU(), # Your activation function + Linear(128, 64), # Your gradient-aware layers ReLU(), # Your nonlinearity - Linear(64, 10) # Your final classifier + Linear(64, 10) # Your classification head ]) -# YOUR training loop using YOUR optimizer -optimizer = Adam(model.parameters(), lr=0.001) # Your Adam implementation -for batch in dataloader: # Your data loading - output = model(batch.x) # Your forward pass - loss = CrossEntropyLoss()(output, batch.y) # Your loss function - loss.backward() # Your backpropagation +# YOUR complete training system +optimizer = Adam(model.parameters(), lr=0.001) # Your optimization algorithm +for batch in dataloader: # Your data management + output = model(batch.x) # Your forward computation + loss = CrossEntropyLoss()(output, batch.y) # Your loss calculation + loss.backward() # YOUR backpropagation engine optimizer.step() # Your parameter updates ``` -**Result: 95%+ accuracy on MNIST using 100% your own code.** +**šŸŽÆ Foundation Achievement**: 95%+ accuracy on MNIST using 100% your own mathematical implementations -### Advanced Capabilities (Modules 9-14) -- **Computer Vision**: CNNs achieving 75%+ accuracy on CIFAR-10 -- **Language Models**: TinyGPT built using 95% of your vision components -- **Universal Architecture**: Same mathematical foundations power all modern AI +### šŸ§  After Intelligence Tier (Modules 08-13) +- **Computer Vision Mastery**: CNNs achieving 75%+ accuracy on CIFAR-10 with YOUR convolution implementations +- **Language Understanding**: Transformers generating coherent text using YOUR attention mechanisms +- **Universal Architecture**: Discover why the SAME mathematical principles work for vision AND language +- **AI Breakthrough Recreation**: Implement the architectures that created the modern AI revolution -### Production Systems (Modules 15-20) -- **Performance Engineering**: Profile, measure, and optimize ML systems -- **Memory Optimization**: Understand and implement compression techniques -- **Hardware Acceleration**: Build efficient kernels and vectorized operations -- **TinyMLPerf Competition**: Compete with optimized implementations +### āš” After Optimization Tier (Modules 14-20) +- **Production Performance**: Systems optimized for <100ms inference latency using YOUR profiling tools +- **Memory Efficiency**: Models compressed to 25% original size with YOUR quantization implementations +- **Hardware Acceleration**: Kernels achieving 10x speedups through YOUR vectorization techniques +- **Competition Ready**: TinyMLPerf submissions competitive with industry implementations --- ## The ML Evolution Story You'll Experience -TinyTorch follows the actual historical progression of machine learning breakthroughs: +TinyTorch's three-tier structure follows the actual historical progression of machine learning breakthroughs: -### šŸ§  Era 1: Foundation (1980s) - Modules 1-8 -**The Beginning**: Perceptrons and multi-layer networks -- Build tensor operations and automatic differentiation -- Implement gradient-based optimization (SGD, Adam) -- **Achievement**: Train MLPs to 95%+ accuracy on MNIST +### šŸ—ļø Foundation Era (1980s-1990s) ā†’ Foundation Tier +**The Beginning**: Mathematical foundations that started it all +- **1986 Breakthrough**: Backpropagation enables multi-layer networks +- **Your Implementation**: Build automatic differentiation and gradient-based optimization +- **Historical Milestone**: Train MLPs to 95%+ accuracy on MNIST using YOUR autograd engine -### šŸ‘ļø Era 2: Spatial Intelligence (1989-2012) - Modules 9-10 -**The Revolution**: Convolutional neural networks -- Add spatial processing with Conv2d and pooling operations -- Build efficient data pipelines for real-world datasets -- **Achievement**: Train CNNs to 75%+ accuracy on CIFAR-10 +### šŸ§  Intelligence Era (1990s-2010s) ā†’ Intelligence Tier +**The Revolution**: Specialized architectures for vision and language +- **1998 Breakthrough**: CNNs revolutionize computer vision (LeCun's LeNet) +- **2017 Breakthrough**: Transformers unify vision and language ("Attention is All You Need") +- **Your Implementation**: Build CNNs achieving 75%+ on CIFAR-10, then transformers for text generation +- **Historical Milestone**: Recreate both revolutions using YOUR spatial and attention implementations -### šŸ—£ļø Era 3: Universal Architecture (2017-Present) - Modules 11-14 -**The Unification**: Transformers for vision AND language -- Implement attention mechanisms and positional embeddings -- Build TinyGPT using your existing vision infrastructure -- **Achievement**: Language generation with 95% component reuse +### āš” Optimization Era (2010s-Present) ā†’ Optimization Tier +**The Engineering**: Production systems that scale to billions of users +- **2020s Breakthrough**: Efficient inference enables real-time LLMs (GPT, ChatGPT) +- **Your Implementation**: Build KV-caching, quantization, and production optimizations +- **Historical Milestone**: Deploy systems competitive in TinyMLPerf benchmarks -### āš” Era 4: Production Systems (Present) - Modules 15-20 -**The Engineering**: Optimized, deployable ML systems -- Profile performance and identify bottlenecks -- Implement compression, quantization, and acceleration -- **Achievement**: TinyMLPerf competition-ready implementations +**Why This Progression Matters**: You'll understand not just modern AI, but WHY it evolved this way. Each tier builds essential capabilities that inform the next, just like ML history itself. --- -## Systems Engineering Focus: Why It Matters +## Systems Engineering Focus: Why Tiers Matter -Traditional ML courses focus on **algorithms**. TinyTorch focuses on **systems**. +Traditional ML courses teach algorithms in isolation. TinyTorch's tier structure teaches **systems thinking** - how components interact to create production ML systems. -### What Traditional Courses Teach: -- "Use `torch.optim.Adam` for optimization" -- "Transformers use attention mechanisms" -- "Larger models generally perform better" +### Traditional Linear Approach: +``` +Module 1: Tensors ā†’ Module 2: Layers ā†’ Module 3: Training ā†’ ... +``` +**Problem**: Students learn components but miss system interactions -### What TinyTorch Teaches: -- "Why Adam consumes 3Ɨ more memory than SGD and when that matters in production" -- "How attention scales O(NĀ²) with sequence length and limits context windows" -- "How to profile memory usage and identify training bottlenecks" +### TinyTorch Tier Approach: +``` +šŸ—ļø Foundation Tier: Build mathematical infrastructure +šŸ§  Intelligence Tier: Compose intelligent architectures +āš” Optimization Tier: Deploy at production scale +``` +**Advantage**: Each tier builds complete, working systems with clear progression -### Career Impact -After TinyTorch, you become the team member who: -- **Debugs performance issues**: "Your convolution is memory-bound, not compute-bound" -- **Optimizes production systems**: "We can use gradient accumulation to train with less GPU memory" -- **Implements custom operations**: "I'll write a custom kernel for this novel architecture" -- **Designs system architecture**: "Here's why this model won't scale and how to fix it" +### What Traditional Courses Teach vs. TinyTorch Tiers: + +**Traditional**: "Use `torch.optim.Adam` for optimization" +**Foundation Tier**: "Why Adam needs 3Ɨ more memory than SGD and how to implement both from mathematical first principles" + +**Traditional**: "Transformers use attention mechanisms" +**Intelligence Tier**: "How attention creates O(NĀ²) scaling, why this limits context windows, and how to implement efficient attention yourself" + +**Traditional**: "Deploy models with TensorFlow Serving" +**Optimization Tier**: "How to profile bottlenecks, implement KV-caching for 10Ɨ speedup, and compete in production benchmarks" + +### Career Impact by Tier +After each tier, you become the team member who: + +**šŸ—ļø Foundation Tier Graduate**: +- Debugs gradient flow issues: "Your ReLU is causing dead neurons" +- Implements custom optimizers: "I'll build a variant of Adam for this use case" +- Understands memory patterns: "Batch size 64 hits your GPU memory limit here" + +**šŸ§  Intelligence Tier Graduate**: +- Designs novel architectures: "We can adapt transformers for this computer vision task" +- Optimizes attention patterns: "This attention bottleneck is why your model won't scale to longer sequences" +- Bridges vision and language: "The same mathematical principles work for both domains" + +**āš” Optimization Tier Graduate**: +- Deploys production systems: "I can get us from 500ms to 50ms inference latency" +- Leads performance optimization: "Here's our memory bottleneck and my 3-step plan to fix it" +- Competes at industry scale: "Our optimizations achieve TinyMLPerf benchmark performance" --- @@ -254,165 +278,159 @@ After TinyTorch, you become the team member who: --- -## Ready to Begin? +## šŸš€ Start Your Journey -You're about to embark on a journey that will transform how you think about machine learning systems. Instead of using black-box frameworks, you'll understand every component from the ground up. +
+

Begin Building ML Systems

+

Choose your starting point based on your goals and time commitment

+15-Minute Start ā†’ +Foundation Tier ā†’ +
-**Next Step**: [Module 01: Setup](01-setup.md) - Configure your development environment and build your first TinyTorch function. +**Next Steps**: +- **New to TinyTorch**: Start with [Quick Start Guide](../quickstart-guide.html) for immediate hands-on experience +- **Ready to Commit**: Begin [Module 01: Setup](01-setup.html) to configure your development environment +- **Teaching a Course**: Review [Instructor Guide](../usage-paths/classroom-use.html) for classroom integration -```{admonition} Your Learning Journey Awaits +```{admonition} Your Three-Tier Journey Awaits :class: tip -By the end of this course, you'll have built a complete ML framework that rivals educational implementations like MiniTorch and micrograd, while achieving production-level results: -- **95%+ accuracy on MNIST** (handwritten digit recognition) -- **75%+ accuracy on CIFAR-10** (real-world image classification) -- **TinyGPT language generation** (modern transformer architecture) -- **TinyMLPerf competition entries** (optimized systems performance) +By completing all three tiers, you'll have built a complete ML framework that rivals production implementations: -All using code you wrote yourself, from scratch. +**šŸ—ļø Foundation Tier Achievement**: 95%+ accuracy on MNIST with YOUR mathematical implementations +**šŸ§  Intelligence Tier Achievement**: 75%+ accuracy on CIFAR-10 AND coherent text generation +**āš” Optimization Tier Achievement**: Production systems competitive in TinyMLPerf benchmarks + +All using code you wrote yourself, from mathematical first principles to production optimization. ``` --- -## Complete Learning Timeline & Course Structure +### šŸ—ļø FOUNDATION TIER (Modules 01-07) +**Building Blocks of ML Systems ā€¢ 6-8 weeks ā€¢ All Prerequisites for Neural Networks** -### Capability Progression: Foundation to Production +
-```{mermaid} -:align: center +**What You'll Learn**: Build the mathematical and computational infrastructure that powers all neural networks. Master tensor operations, gradient computation, and optimization algorithms. -timeline - title TinyTorch Capability Development: Building ML Systems +**Prerequisites**: Python programming, basic linear algebra (matrix multiplication) - section Foundation Capabilities - Environment Setup : Checkpoint 00 Complete - : Configure development environment - : Verify dependencies +**Career Connection**: Foundation skills required for ML Infrastructure Engineer, Research Engineer, Framework Developer roles - Tensor Operations : Checkpoint 01 Complete - : N-dimensional arrays - : Mathematical foundations +**Time Investment**: ~20 hours total (3 hours/week for 6-8 weeks) - section Core Learning - Neural Intelligence : Checkpoint 02 Complete - : Nonlinear activations - : ReLU, Sigmoid, Softmax - - Network Building : Checkpoint 03 Complete - : Layer abstractions - : Forward propagation - - section Training Systems - Gradient Computation : Checkpoint 05 Complete - : Automatic differentiation - : Backpropagation mechanics - - Optimization : Checkpoint 06 Complete - : SGD, Adam algorithms - : Learning rate scheduling - - section Advanced Architectures - Computer Vision : Checkpoint 08 Complete - : Convolutional operations - : Spatial feature extraction - - Language Processing : Checkpoint 12 Complete - : Attention mechanisms - : Transformer architectures - - section Production Systems - Performance Analysis : Checkpoint 14 Complete - : Profiling and optimization - : Bottleneck identification - - Complete Mastery : Checkpoint 15 Complete - : End-to-end ML systems - : Production deployment -``` - -### Part I: Core Foundations (Modules 1-8) -**Focus: Neural Network Fundamentals | 8 weeks** - -| Week | Module | Core Capability | Implementation Focus | Checkpoint Unlocked | -|------|--------|-----------------|---------------------|--------------------| -| 1 | Setup | Environment Configuration | Development environment setup | 00: Environment | -| 2 | Tensor | Mathematical Foundations | N-dimensional arrays with gradients | 01: Foundation | -| 3 | Activations | Neural Intelligence | ReLU, Sigmoid, Softmax functions | 02: Intelligence | -| 4 | Layers | Network Components | Linear layers and module system | 03: Components | -| 5 | Losses | Learning Measurement | MSE, CrossEntropy loss functions | 04: Networks | -| 6 | Autograd | Gradient Computation | Automatic differentiation engine | 05: Learning | -| 7 | Optimizers | Parameter Updates | SGD, Adam optimization algorithms | 06: Optimization | -| 8 | Training | Complete Systems | End-to-end training loops | 07: Training | - -**Capability Milestone**: After Module 8, you have complete neural network training capability! - ---- - -### Part II: Computer Vision (Modules 9-10) -**Focus: Spatial Processing | 2 weeks** - -| Week | Module | Core Capability | Implementation Focus | Checkpoint Unlocked | -|------|--------|-----------------|---------------------|--------------------| -| 9 | Spatial | Spatial Processing | Conv2d, MaxPool2d operations | 08: Vision | -| 10 | DataLoader | Data Management | Efficient data loading pipelines | 09: Data | - -**Capability Milestone**: Computer vision systems with spatial feature processing! - ---- - -### Part III: Language Processing (Modules 11-14) -**Focus: Sequence Understanding | 4 weeks** - -| Week | Module | Core Capability | Implementation Focus | Checkpoint Unlocked | -|------|--------|-----------------|---------------------|--------------------| -| 11 | Tokenization | Text Processing | Vocabulary and token systems | 10: Language | -| 12 | Embeddings | Representation Learning | Token and positional encodings | 11: Representation | -| 13 | Attention | Sequence Understanding | Multi-head attention mechanisms | 12: Attention | -| 14 | Transformers | Architecture Mastery | Complete transformer blocks | 13: Architecture | - -**Capability Milestone**: Complete language understanding and generation systems! - ---- - -### Part IV: Production Systems (Modules 15-20) -**Focus: Performance Optimization | 6 weeks** - -| Week | Module | Core Capability | Implementation Focus | Checkpoint Unlocked | -|------|--------|-----------------|---------------------|--------------------| -| 15 | Profiling | Performance Analysis | Memory and compute profiling | 14: Systems | -| 16 | Acceleration | Hardware Optimization | Vectorization and caching | | -| 17 | Quantization | Model Compression | INT8 inference optimization | | -| 18 | Compression | Size Optimization | Pruning and distillation | | -| 19 | Caching | Memory Management | KV-cache for generation | | -| 20 | Capstone | Complete Mastery | End-to-end ML systems | 15: Mastery | - -**Final Capability**: Complete ML systems engineering mastery! - ---- - -## šŸ“ˆ 8-Week Learning Progression Overview - -For a quick visual overview of the main learning phases: - -
- -
-

Weeks 1-2: Mathematical Foundations

-

Implement tensor operations, understand memory layout, build arithmetic foundations. Core mathematical building blocks.

-
-

Weeks 3-4: Neural Network Components

-

Linear transformations, activation functions, loss functions. Build the mathematical components of neural computation.

+| Module | Component | Core Capability | Real-World Connection | +|--------|-----------|-----------------|----------------------| +| **01** | **Tensor** | Data structures and operations | NumPy, PyTorch tensors | +| **02** | **Activations** | Nonlinear functions | ReLU, attention activations | +| **03** | **Layers** | Linear transformations | `nn.Linear`, dense layers | +| **04** | **Losses** | Optimization objectives | CrossEntropy, MSE loss | +| **05** | **Autograd** | Automatic differentiation | PyTorch autograd engine | +| **06** | **Optimizers** | Parameter updates | Adam, SGD optimizers | +| **07** | **Training** | Complete training loops | Model.fit(), training scripts | + +**šŸŽÆ Tier Milestone**: Train neural networks achieving **95%+ accuracy on MNIST** using 100% your own implementations! + +**Skills Gained**: +- Understand memory layout and computational graphs +- Debug gradient flow and numerical stability issues +- Implement any optimization algorithm from research papers +- Build custom neural network architectures from scratch + +--- + +### šŸ§  INTELLIGENCE TIER (Modules 08-13) +**Modern AI Algorithms ā€¢ 4-6 weeks ā€¢ Vision + Language Architectures** + +
+ +**What You'll Learn**: Implement the architectures powering modern AI: convolutional networks for vision and transformers for language. Discover why the same mathematical principles work across domains. + +**Prerequisites**: Foundation Tier complete (Modules 01-07) + +**Career Connection**: Computer Vision Engineer, NLP Engineer, AI Research Scientist, ML Product Manager roles + +**Time Investment**: ~25 hours total (4-6 hours/week for 4-6 weeks) +
-
-

Weeks 5-6: Learning Algorithms

-

Automatic differentiation, optimization algorithms, training procedures. Understand how neural networks learn.

+| Module | Component | Core Capability | Real-World Connection | +|--------|-----------|-----------------|----------------------| +| **08** | **Spatial** | Convolutions and regularization | CNNs, ResNet, computer vision | +| **09** | **DataLoader** | Batch processing | PyTorch DataLoader, tf.data | +| **10** | **Tokenization** | Text preprocessing | BERT tokenizer, GPT tokenizer | +| **11** | **Embeddings** | Representation learning | Word2Vec, positional encodings | +| **12** | **Attention** | Information routing | Multi-head attention, self-attention | +| **13** | **Transformers** | Modern architectures | GPT, BERT, Vision Transformer | + +**šŸŽÆ Tier Milestone**: Achieve **75%+ accuracy on CIFAR-10** with CNNs AND generate coherent text with transformers! + +**Skills Gained**: +- Understand why convolution works for spatial data +- Implement attention mechanisms from scratch +- Build transformer architectures for any domain +- Debug sequence modeling and attention patterns + +--- + +### āš” OPTIMIZATION TIER (Modules 14-20) +**Production & Performance ā€¢ 4-6 weeks ā€¢ Deploy and Scale ML Systems** + +
+ +**What You'll Learn**: Transform research models into production systems. Master profiling, optimization, and deployment techniques used by companies like OpenAI, Google, and Meta. + +**Prerequisites**: Intelligence Tier complete (Modules 08-13) + +**Career Connection**: ML Systems Engineer, Performance Engineer, MLOps Engineer, Senior ML Engineer roles + +**Time Investment**: ~30 hours total (5-7 hours/week for 4-6 weeks) +
-
-

Weeks 7-8: Systems Engineering

-

Performance analysis, computational kernels, benchmarking. Study the engineering principles behind ML systems.

+| Module | Component | Core Capability | Real-World Connection | +|--------|-----------|-----------------|----------------------| +| **14** | **Profiling** | Performance analysis | PyTorch Profiler, TensorBoard | +| **15** | **Acceleration** | Speed improvements | CUDA kernels, vectorization | +| **16** | **Quantization** | Memory efficiency | INT8 inference, model compression | +| **17** | **Compression** | Model optimization | Pruning, distillation, ONNX | +| **18** | **Caching** | Memory management | KV-cache for generation | +| **19** | **Benchmarking** | Measurement systems | MLPerf, production monitoring | +| **20** | **Capstone** | Full system integration | End-to-end ML pipeline | + +**šŸŽÆ Tier Milestone**: Build **production-ready systems** competitive in TinyMLPerf benchmarks! + +**Skills Gained**: +- Profile memory usage and identify bottlenecks +- Implement efficient inference optimizations +- Deploy models with <100ms latency requirements +- Design scalable ML system architectures + +--- + +## šŸŽÆ Learning Path Recommendations + +### Choose Your Learning Style + +
+ +
+

šŸš€ Complete Builder

+

Implement every component from scratch

+

Time: 14-18 weeks
Ideal for: CS students, aspiring ML engineers

+
+ +
+

āš” Focused Explorer

+

Pick one tier based on your goals

+

Time: 4-8 weeks
Ideal for: Working professionals, specific skill gaps

+
+ +
+

šŸ“š Guided Learner

+

Study implementations with hands-on exercises

+

Time: 8-12 weeks
Ideal for: Self-directed learners, bootcamp graduates

diff --git a/book/intro.md b/book/intro.md index ff6b312b..cf3b2d99 100644 --- a/book/intro.md +++ b/book/intro.md @@ -31,11 +31,17 @@ TinyTorch is an educational ML systems course where you **build complete neural **Core Learning Approach**: Build ā†’ Profile ā†’ Optimize. You'll implement each system component, measure its performance characteristics, and understand the engineering trade-offs that shape production ML systems. -## The ML Evolution Story You'll Experience +## Three-Tier Learning Pathway -Journey through 40+ years of ML breakthroughs by building each era yourself: **1980s neural foundations** ā†’ **1990s backpropagation** ā†’ **2012 CNN revolution** ā†’ **2017 transformer unification** ā†’ **2024 production optimization**. Each module teaches both the breakthrough AND the systems engineering that made it possible. +TinyTorch organizes learning through **three pedagogically-motivated tiers** that follow ML history: -**šŸ“– See [Complete ML Evolution Timeline](chapters/00-introduction.html#the-ml-evolution-story-youll-experience)** for the full historical context and technical progression. +**šŸ—ļø Foundation Tier (Modules 01-07)**: Build mathematical infrastructure - tensors, autograd, optimizers +**šŸ§  Intelligence Tier (Modules 08-13)**: Implement modern AI - CNNs for vision, transformers for language +**āš” Optimization Tier (Modules 14-20)**: Deploy production systems - profiling, quantization, acceleration + +Each tier builds complete, working systems with clear career connections and practical skills. + +**šŸ“– See [Complete Three-Tier Structure](chapters/00-introduction.html#three-tier-learning-pathway-build-complete-ml-systems)** for detailed tier breakdown, time estimates, and learning outcomes. ## šŸ† Prove Your Mastery Through History @@ -167,7 +173,7 @@ You master modern LLM optimizations ## How to Choose Your Learning Path -**Two Learning Approaches**: You can either **build it yourself** (work through student notebooks and implement from scratch) or **learn by reading** (study the solution notebooks to understand how ML systems work). Both approaches use the same **Build ā†’ Profile ā†’ Optimize** methodology at different scales. +**Three Learning Approaches**: You can **build complete tiers** (implement all 20 modules), **focus on specific tiers** (target your skill gaps), or **explore selectively** (study key concepts). Each tier builds complete, working systems.
@@ -201,7 +207,7 @@ You master modern LLM optimizations ## Getting Started -Whether you're just exploring or ready to dive in, here are helpful resources: **šŸ“– See [Essential Commands](tito-essentials.html)** for complete setup and command reference, or **šŸ“– See [Complete Course Structure](chapters/00-introduction.html)** for detailed module descriptions. +Whether you're just exploring or ready to dive in, here are helpful resources: **šŸ“– See [Essential Commands](tito-essentials.html)** for complete setup and command reference, or **šŸ“– See [Three-Tier Learning Structure](chapters/00-introduction.html#three-tier-learning-pathway-build-complete-ml-systems)** for detailed tier breakdown and learning outcomes. **Additional Resources**: - **[Progress Tracking](learning-progress.html)** - Monitor your learning journey with 21 capability checkpoints diff --git a/book/learning-progress.md b/book/learning-progress.md index 556e47c5..9b0dfdf0 100644 --- a/book/learning-progress.md +++ b/book/learning-progress.md @@ -22,23 +22,21 @@ Use TinyTorch's 21-checkpoint system to monitor your capability development. Tra ## Your Learning Path Overview -TinyTorch organizes learning through four major phases, each building essential ML systems capabilities: +TinyTorch organizes learning through **three pedagogically-motivated tiers**, each building essential ML systems capabilities: -**šŸ“– See [Complete Course Structure](chapters/00-introduction.html)** for the full learning timeline and detailed module descriptions. +**šŸ“– See [Three-Tier Learning Structure](chapters/00-introduction.html#three-tier-learning-pathway-build-complete-ml-systems)** for detailed tier breakdown, time estimates, and learning outcomes. ## Student Learning Journey -### Typical Student Progression -- **Week 1-2**: Foundation capabilities (Environment, Tensors, Activations) -- **Week 3-4**: Core learning systems (Layers, Losses, Autograd) -- **Week 5-6**: Training and optimization (Optimizers, Training loops) -- **Week 7-8**: Advanced architectures (Spatial processing, Attention) -- **Week 9-12**: Production systems (Profiling, Optimization, Deployment) +### Typical Student Progression by Tier +- **šŸ—ļø Foundation Tier (6-8 weeks)**: Build mathematical infrastructure - tensors, autograd, optimizers, training loops +- **šŸ§  Intelligence Tier (4-6 weeks)**: Implement modern AI architectures - CNNs for vision, transformers for language +- **āš” Optimization Tier (4-6 weeks)**: Deploy production systems - profiling, quantization, acceleration ### Study Approaches -- **Full Implementation** (8-12 weeks): Build every component from scratch -- **Guided Study** (4-6 weeks): Study solution notebooks with implementation exercises -- **Quick Exploration** (2 weeks): Focus on key concepts with provided implementations +- **Complete Builder** (14-18 weeks): Implement all three tiers from scratch +- **Focused Explorer** (4-8 weeks): Pick specific tiers based on your goals +- **Guided Learner** (8-12 weeks): Study implementations with hands-on exercises **šŸ“– See [Quick Start Guide](quickstart-guide.html)** for immediate hands-on experience with your first module. diff --git a/book/usage-paths/classroom-use.md b/book/usage-paths/classroom-use.md index 0105324b..b16c7fe7 100644 --- a/book/usage-paths/classroom-use.md +++ b/book/usage-paths/classroom-use.md @@ -21,7 +21,7 @@
    -
  • 20 progressive modules with NBGrader integration
    • +
  • Three-tier progression (20 modules) with NBGrader integration
  • 200+ automated tests for immediate feedback
  • Professional CLI tools for development workflow
  • Real datasets (CIFAR-10, text generation)