Advanced level · Module 6: Deep Learning Capstone · Lesson 2 of 5
Training and Model Selection
Train candidate models and select one that can meet all targets.
In this lesson you will
- Train and compare candidate models
- Select a model with deployment in mind
Key concepts
transfer learningexperiment trackingcandidate comparisonselection rationale
The lesson
Start from pre-trained models suited to the task and device. Track every experiment.
Compare candidates on validation accuracy, size and estimated latency.
Select the model most likely to meet all targets after compression.
Step by step
- Train two or three candidates.
- Track and compare them.
- Estimate their deployed size and speed.
- Select and justify one.
Practice activity
Complete milestone 2: trained candidates and a selection.
Expected result: A justified selection with tracked experiments.
Check your understanding
-
Why estimate deployed size before selecting?
Show answer
The model must fit the device after compression.
-
Why track experiments?
Show answer
To compare candidates and reproduce the chosen one.
Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.