From 71be93f88bf1ab104f0b7b6f85bccf8392a06e2f Mon Sep 17 00:00:00 2001
From: Vijay Janapa Reddi <vj@eecs.harvard.edu>
Date: Mon, 15 Sep 2025 14:38:14 -0400
Subject: [PATCH] Improve 5 C's format: Use integrated code-comment style

Replace verbose bullet format with code-comment approach that:
- Integrates concepts directly with implementation preview
- Shows exactly where each principle applies in actual code
- Feels more natural and less academic
- Maintains educational value while respecting student time
- Bridges gap between understanding and coding

The code-comment style helps students see the connection between
concepts and implementation rather than treating them as separate
academic content.
---
 modules/source/02_tensor/tensor_dev.py | 56 ++++++++++----------------
 1 file changed, 22 insertions(+), 34 deletions(-)

diff --git a/modules/source/02_tensor/tensor_dev.py b/modules/source/02_tensor/tensor_dev.py
index ad768954..e560cec5 100644
--- a/modules/source/02_tensor/tensor_dev.py
+++ b/modules/source/02_tensor/tensor_dev.py
@@ -327,47 +327,35 @@ Let's implement our tensor foundation!
 
 # %% [markdown]
 """
-## 🧱 The 5 C's: Understanding Tensor Implementation
+### Building Your Tensor Implementation
 
-### 1️⃣ CONCEPT: What is a Tensor?
-A **Tensor** is a generalized mathematical object that represents scalars (0D), vectors (1D), matrices (2D), and higher-dimensional arrays (3D+). In machine learning, tensors are the fundamental data structure that flows through neural networks, carrying both data and gradients.
-
-### 2️⃣ CODE STRUCTURE: What We're Building
 ```python
+# CONCEPT: Tensors are N-dimensional arrays that carry data through neural networks
+# Think NumPy arrays with ML superpowers - same math, more ML-focused capabilities
+
 class Tensor:
-    def __init__(self, data):          # Create from data
-        self._data = np.array(data)    # Store as NumPy array
+    def __init__(self, data):
+        # CORE IMPLEMENTATION: Your tensor needs these fundamentals
+        # - Store N-dimensional data efficiently  
+        # - Support math operations: +, *, .sum(), .reshape()
+        # - Handle broadcasting (auto-shape matching)
+        # - Work with multiple data types (float32, int64, etc.)
+        pass
     
-    @property
-    def data(self):                    # Access underlying data
+    def __add__(self, other):
+        # REAL-WORLD CONNECTION: This mirrors torch.Tensor.__add__()
+        # Every + operation in PyTorch goes through similar logic
+        pass
     
-    def __add__(self, other):          # Enable: tensor + tensor
-    def __mul__(self, other):          # Enable: tensor * tensor
-    # ... more operations
+    def sum(self):
+        # WHY THIS MATTERS: Sum operations are everywhere in ML
+        # - Loss calculations (MSE, CrossEntropy)
+        # - Gradient accumulation  
+        # - Batch processing statistics
+        pass
 ```
 
-### 3️⃣ CONNECTIONS: Real-World Integration
-- **PyTorch equivalent**: `torch.Tensor` - same concept, optimized implementation
-- **TensorFlow equivalent**: `tf.Tensor` - distributed computing focus
-- **NumPy relationship**: We wrap `np.ndarray` with ML-specific operations
-- **Production usage**: Every neural network processes millions of tensors per second
-
-### 4️⃣ CONSTRAINTS: Implementation Requirements
-- **Data storage**: Must efficiently wrap NumPy arrays
-- **Type safety**: Handle different data types (float32, int32, etc.)
-- **Broadcasting**: Automatic shape handling for operations
-- **Memory efficiency**: Copy data only when necessary
-- **Operator support**: Enable natural mathematical notation (+, -, *, /)
-
-### 5️⃣ CONTEXT: Why This Matters in ML Systems
-Tensors are the **universal language** of machine learning:
-- **Neural networks**: All computations operate on tensors
-- **Automatic differentiation**: Gradients flow through tensor operations
-- **Hardware acceleration**: GPUs are optimized for tensor operations
-- **Distributed computing**: Tensors can be split across multiple devices
-- **Research foundation**: Every ML breakthrough builds on tensor mathematics
-
-**Understanding tensors deeply means understanding the foundation of all modern AI.**
+*This is the same concept as `torch.Tensor` - you're building the foundation that powers production ML frameworks.*
 """
 
 # %% nbgrader={"grade": false, "grade_id": "tensor-class", "locked": false, "schema_version": 3, "solution": true, "task": false}