Train/val/test split in PyTorch - Model Pipeline Trace

Model Pipeline - Train/val/test split

This pipeline splits the dataset into three parts: training, validation, and testing. The training set teaches the model, the validation set helps tune it, and the test set checks how well it learned.

Data Flow - 2 Stages

1Original Dataset

1000 rows x 10 columns→Start with full dataset→1000 rows x 10 columns

[[5.1, 3.5, ..., 0], [4.9, 3.0, ..., 1], ...]

↓

2Train/Val/Test Split

1000 rows x 10 columns→Split dataset into 70% train, 15% val, 15% test→Train: 700 rows x 10 columns, Val: 150 rows x 10 columns, Test: 150 rows x 10 columns

Train sample: [5.1, 3.5, ..., 0], Val sample: [6.7, 3.1, ..., 1], Test sample: [5.9, 3.0, ..., 0]

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	0.85	0.60	Model starts learning with moderate loss and accuracy.
2	0.65	0.72	Loss decreases and accuracy improves as model learns.
3	0.50	0.80	Model shows good progress with lower loss and higher accuracy.
4	0.40	0.85	Training continues to improve with steady loss decrease.
5	0.35	0.88	Model converges with low loss and high accuracy.

Prediction Trace - 3 Layers

Layer 1: Input Sample

Layer 2: Model Forward Pass

Layer 3: Prediction

Model Quiz - 3 Questions

Why do we split data into train, validation, and test sets?

A To train, tune, and fairly evaluate the model

B To increase dataset size

C To reduce model complexity

D To speed up training only

Key Insight

Splitting data into train, validation, and test sets helps the model learn well, tune hyperparameters, and fairly measure performance on unseen data.