Batch size and epochs in TensorFlow - Model Pipeline Trace

Model Pipeline - Batch size and epochs

This pipeline shows how batch size and epochs affect training a simple neural network. Batch size controls how many samples the model sees before updating. Epochs control how many times the model sees the whole dataset.

Data Flow - 4 Stages

1Data Loading

1000 rows x 10 columns→Load dataset with 1000 samples and 10 features each→1000 rows x 10 columns

[[0.5, 1.2, ..., 0.3], [0.1, 0.4, ..., 0.9], ...]

↓

2Train/Test Split

1000 rows x 10 columns→Split data into 800 training and 200 testing samples→800 rows x 10 columns (train), 200 rows x 10 columns (test)

Train sample: [0.5, 1.2, ..., 0.3], Test sample: [0.2, 0.7, ..., 0.1]

↓

3Batching

800 rows x 10 columns→Divide training data into batches of 100 samples→8 batches x 100 rows x 10 columns

Batch 1: 100 samples, Batch 2: 100 samples, ...

↓

4Model Training

Batch of 100 rows x 10 columns→Train model on each batch, repeat for 5 epochs→Model weights updated after each batch

Batch 1 training updates weights, then Batch 2, ... repeat 5 times

Training Trace - Epoch by Epoch

Loss
0.7 |****
0.6 |*** 
0.5 |**  
0.4 |**  
0.3 |*   
0.2 |    
     1 2 3 4 5 Epochs

Epoch	Loss ↓	Accuracy ↑	Observation
1	0.65	0.60	Model starts learning with moderate loss and accuracy
2	0.45	0.75	Loss decreases and accuracy improves as model learns
3	0.35	0.82	Model continues to improve with more epochs
4	0.30	0.86	Loss decreases steadily, accuracy rises
5	0.28	0.88	Training converges with low loss and high accuracy

Prediction Trace - 3 Layers

Layer 1: Input Layer

Layer 2: Hidden Layer (ReLU)

Layer 3: Output Layer (Sigmoid)

Model Quiz - 3 Questions

What does increasing the batch size do during training?

A Increases the number of epochs

B Reduces the number of features

C Processes more samples before updating model weights

D Changes the model architecture

Key Insight

Batch size controls how many samples the model learns from before updating weights, affecting training speed and stability. Epochs control how many times the model sees the entire dataset, allowing it to improve gradually. Together, they balance learning quality and training time.