update_params¶
Update the parameters embedded into a generated .tflite model file.
Additional Documentation¶
Usage¶
Usage: mltk update_params [OPTIONS] <model>
Update the parameters of a previously trained model
This updates the metadata of a previously trained .tflite model.
The parameters are taken from either the given MltkModel's python script
or the given "params" .json/.yaml file. Additional int/float/str parameters
can also be
given on the command line (see examples below).
NOTE: The .tflite metadata is only modified.
The weights and model structure of the .tflite file are NOT modified.
For more details see:
https://siliconlabs.github.io/mltk/docs/guides/model_parameters
----------
Examples
----------
# Update the tflite_micro_speech model parameters with any modifications
# made in mltk/models/tflite_micro/tflite_micro_speech.py
# The associated model archive is updated with the updated .tflite
mltk update_params tflite_micro_speech
# Update my_model.tflite with parameters in my_params.json
# Also update the model description
mltk update_params my_model.tflite --params my_params.json --description "My
model is great!"
# Update my_model.tflite with additional params on the command-line
mltk update_params my_model.tflite my_custom_param="some value"
led_period_ms=43
Arguments
* model <model> One of the following:
- Name of trained MLTK model
- Path to trained model's archive (.mltk.zip)
- Path to .tflite model file
[default: None]
[required]
Options
--params -p <params path> Optional path to .json or .yaml file
contains parameters to add to given
model
[default: None]
--description -d <description> Optional description to add to the
generated .tflite
[default: None]
--verbose -v Enable verbose console logs
--output -o <path> One of the following:
- Path to generated output .tflite
file
- Directory where output .tflite is
generated
- If omitted, .tflite is generated in
the MLTK model's log directory and
the model archive is updated (if
applicable)
[default: None]
--accelerator -a <accelerator> Optional accelerator to use when
calculating the "runtime_memory_size"
model parameter. If omitted then use
the CMSIS kernels
[default: None]
--device -d If provided, program the updated
.tflite to end of the flash memory of
the the connected device.
Supported apps (e.g. model_profiler,
audio_classifier, etc) will use this
.tflite instead of the default model.
This allows for making changes to the
model without re-building the
firmware application.
If this option is provided, then the
device must be locally connected
--help Show this message and exit.