diff --git a/book/intro.md b/book/intro.md index 14ac6101..92d9e66b 100644 --- a/book/intro.md +++ b/book/intro.md @@ -198,57 +198,6 @@ Want to see what TinyTorch feels like? **[Launch the Setup chapter](chapters/01- --- -## ❓ **Common Questions** - -
-🧮 "Do I need to know advanced math to succeed?" - -**You need high school math + willingness to learn.** We explain the math as we go: - -- **Linear algebra**: We explain matrix multiplication when we build Dense layers -- **Calculus**: We cover derivatives when implementing backpropagation -- **Statistics**: We introduce concepts like gradients in context of optimization - -**Key insight:** You learn the math by implementing it, not the other way around. Most students find this more intuitive than traditional math courses. -
- -
-✅ "How do I know if I'm implementing things correctly?" - -**Built-in validation at every step:** -- **200+ automated tests** that check your implementations -- **Progressive testing**: Start simple, then add complexity -- **Clear error messages**: "Your ReLU should return 0 for negative inputs" -- **Expected outputs**: Know exactly what your code should produce - -**Example:** Your tensor multiplication either passes the test or gets specific feedback about what went wrong. -
- -
-🔗 "Can I skip around or must I do modules in order?" - -**Designed for flexibility with recommended paths:** -- **Core foundation**: Modules 1-4 build on each other (do in order) -- **Parallel tracks**: Modules 5-8 can be done in different sequences -- **Advanced modules**: Pick what interests you most - -**Smart approach:** Do Setup → Tensor → Activations in order, then choose your adventure. -
- -
-🚀 "Will this actually work with real data and real problems?" - -**Yes - we use production datasets from day one:** -- **CIFAR-10**: Train on 60,000 real images, not toy data -- **Standard benchmarks**: Compare your results to published papers -- **Real performance**: Your CNN will achieve 85%+ accuracy on image classification -- **Production patterns**: CLI tools, testing, packaging like professional frameworks - -**This isn't a toy - it's a real framework that handles real problems.** -
- ---- - ## 🙏 **Acknowledgments** TinyTorch originated from CS249r: Tiny Machine Learning Systems at Harvard University. We're inspired by projects like [tinygrad](https://github.com/geohot/tinygrad), [micrograd](https://github.com/karpathy/micrograd), and [MiniTorch](https://minitorch.github.io/) that demonstrate the power of minimal implementations.