mirror of
https://github.com/MLSysBook/TinyTorch.git
synced 2026-06-04 21:42:14 -05:00
48e62a954a
- Changed all module titles to be short and clean (e.g., 'Autograd' not 'Autograd - Automatic Differentiation') - Updated metadata generation template to use concise titles by default - Fixed CLI reference in metadata generator to use new hierarchical structure - Titles are now consistent: just the module name capitalized - Detailed descriptions remain in the description field where they belong
101 lines
2.8 KiB
YAML
101 lines
2.8 KiB
YAML
# TinyTorch Module Metadata
|
|
# This file contains structured information about the module for CLI tools and documentation
|
|
|
|
# Basic Information
|
|
name: "activations"
|
|
title: "Activations"
|
|
description: "Implement the mathematical functions that give neural networks their power to learn complex patterns"
|
|
version: "1.0.0"
|
|
author: "TinyTorch Team"
|
|
last_updated: "2024-12-19"
|
|
|
|
# Module Status
|
|
status: "complete" # complete, in_progress, not_started, deprecated
|
|
implementation_status: "stable" # stable, beta, alpha, experimental
|
|
|
|
# Learning Information
|
|
learning_objectives:
|
|
- "Understand why activation functions are essential for neural networks"
|
|
- "Implement the three most important activation functions: ReLU, Sigmoid, and Tanh"
|
|
- "Test functions with various inputs to understand their behavior"
|
|
- "Grasp the mathematical properties that make each function useful"
|
|
|
|
key_concepts:
|
|
- "Nonlinearity"
|
|
- "Activation functions"
|
|
- "Mathematical properties"
|
|
- "Function composition"
|
|
- "Neural network foundations"
|
|
|
|
# Dependencies
|
|
dependencies:
|
|
prerequisites: ["setup", "tensor"]
|
|
builds_on: ["tensor"]
|
|
enables: ["layers", "networks"]
|
|
|
|
# Educational Metadata
|
|
difficulty: "beginner" # beginner, intermediate, advanced
|
|
estimated_time: "2-3 hours"
|
|
pedagogical_pattern: "Build → Use → Understand"
|
|
|
|
# Implementation Details
|
|
components:
|
|
- name: "ReLU"
|
|
type: "function"
|
|
description: "Rectified Linear Unit: f(x) = max(0, x)"
|
|
status: "complete"
|
|
|
|
- name: "Sigmoid"
|
|
type: "function"
|
|
description: "Sigmoid function: f(x) = 1 / (1 + e^(-x))"
|
|
status: "complete"
|
|
|
|
- name: "Tanh"
|
|
type: "function"
|
|
description: "Hyperbolic tangent: f(x) = tanh(x)"
|
|
status: "complete"
|
|
|
|
# Package Export Information
|
|
exports_to: "tinytorch.core.activations"
|
|
export_directive: "core.activations"
|
|
|
|
# Testing Information
|
|
test_coverage: "comprehensive" # comprehensive, partial, minimal, none
|
|
test_count: 12
|
|
test_categories:
|
|
- "ReLU function behavior"
|
|
- "Sigmoid function behavior"
|
|
- "Tanh function behavior"
|
|
- "Edge cases and numerical stability"
|
|
- "Function properties"
|
|
- "Visualization tests"
|
|
|
|
# File Structure
|
|
required_files:
|
|
- "activations_dev.py"
|
|
- "activations_dev.ipynb"
|
|
- "tests/test_activations.py"
|
|
- "README.md"
|
|
|
|
# Systems Focus
|
|
systems_concepts:
|
|
- "Numerical stability"
|
|
- "Mathematical implementation"
|
|
- "Function design patterns"
|
|
- "Visualization techniques"
|
|
- "Testing mathematical functions"
|
|
|
|
# Real-world Applications
|
|
applications:
|
|
- "Neural network nonlinearity"
|
|
- "Hidden layer transformations"
|
|
- "Output layer activations"
|
|
- "Gradient flow control"
|
|
|
|
# Next Steps
|
|
next_modules: ["layers", "networks"]
|
|
completion_criteria:
|
|
- "All tests pass"
|
|
- "Can visualize activation functions"
|
|
- "Understand nonlinearity importance"
|
|
- "Ready for layer composition" |