Advanced level · Module 6: Deep Learning Capstone · Lesson 3 of 5
Compression and On-Device Testing
Export, compress and run the model on the target device, measuring real performance.
In this lesson you will
- Compress and export the selected model
- Measure on-device accuracy, size and latency
Key concepts
exportquantisationon-device benchmarktarget check
The lesson
Export and compress the selected model, checking accuracy after each step.
Run it on the target phone or edge device offline and measure size, latency and memory.
Compare with the targets and iterate if needed.
Step by step
- Export and quantise the model.
- Check accuracy after compression.
- Benchmark on the device.
- Iterate until targets are met.
Practice activity
Complete milestone 3: an on-device model meeting the targets.
Expected result: Measured on-device results compared with the targets.
Check your understanding
-
What must be checked after each compression step?
Show answer
Accuracy.
-
Where should latency be measured?
Show answer
On the target device.
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.