Tensor shapes and reshaping in TensorFlow - Model Pipeline Trace

Model Pipeline - Tensor shapes and reshaping

This pipeline shows how tensors (multi-dimensional arrays) change shape during a simple reshaping operation. It helps understand how data can be rearranged without changing its content, which is important for feeding data into machine learning models.

Data Flow - 3 Stages

1Input tensor

1 row x 12 columns→Create a 1D tensor with 12 elements→1 row x 12 columns

[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12]

↓

2Reshape to 3 rows x 4 columns

1 row x 12 columns→Reshape tensor to 2D with 3 rows and 4 columns→3 rows x 4 columns

[[1, 2, 3, 4], [5, 6, 7, 8], [9, 10, 11, 12]]

↓

3Reshape to 2 rows x 2 columns x 3 depth

3 rows x 4 columns→Reshape tensor to 3D with shape (2, 2, 3)→2 rows x 2 columns x 3 depth

[[[1, 2, 3], [4, 5, 6]], [[7, 8, 9], [10, 11, 12]]]

Training Trace - Epoch by Epoch

Loss
1.0 |*       
0.8 | *      
0.6 |  *     
0.4 |   *    
0.2 |    *   
0.0 +--------
     1 2 3 4 5 Epochs

Epoch	Loss ↓	Accuracy ↑	Observation
1	0.9	0.30	Initial loss is high, accuracy is low as model starts learning
2	0.7	0.50	Loss decreases, accuracy improves as model learns patterns
3	0.5	0.70	Loss continues to decrease, accuracy rises steadily
4	0.3	0.85	Model is learning well, loss low and accuracy high
5	0.2	0.90	Training converges with low loss and high accuracy

Prediction Trace - 3 Layers

Layer 1: Input tensor

Layer 2: Reshape to (3,4)

Layer 3: Reshape to (2,2,3)

Model Quiz - 3 Questions

What happens to the total number of elements when reshaping a tensor?

A It doubles

B It stays the same

C It halves

D It becomes zero

Key Insight

Reshaping tensors changes their shape but not the total number of elements. This is crucial for preparing data to fit model input requirements without losing information. During training, a good model shows decreasing loss and increasing accuracy, indicating it is learning well.