Create 'hello world' tflite model
Overview
The new version of spamcheck
requires the tflite
(tensorflow lite) bindings for python.
To try and avoid introducing the entire and massive tensorflow
package to Omnibus GitLab, we need a tflite
based model to test against.
Technical context
The full tensorflow
is resource intensive at runtime, so we do not want to have to introduce it to Omnibus GitLab.
Based on research, the most efficient way to test the tflite
python bindings would be with a tflite
model.
The Distribution team needs assistance to identify or create a basic tflite
model to use when testing the tflite
python
bindings package build.
Deliverables
- Find or create a
tflite
model that can be used to test thetflite
python
bindings in a system wheretflite
python
has been installed viapip
. - Write documentation for how to use the
tflite
based model that can be added as thedefinition of done
for Add python-tflite software config to Omnibus Gi... (#8415) • Unassigned
Definition of done / testing
- A
tflite
based model is created or identified. - Clear documentation added to Add python-tflite software config to Omnibus Gi... (#8415) • Unassigned for how to use the model for testing a
tflite
python
binding.