No .tflite file got following Custom Scikit Learn ML block example for Edge Impulse

Blog
#1

Good morning,

I followed the Custom Scikit Learn ML block example for Edge Impulse( GitHub - edgeimpulse/example-custom-ml-block-scikit: Scikit learn block) in order to get the .tflite files as mentioned in the final step 10. But i just got the .pkl model. When i opened the train.py there is no conversion function ( convert_to_tf_lite) as writen in the keras custom ml block example.

Could you please tell me what shall i do to get the outputs as mentioned in step10 of this example.
Thanks an advance

#2

Thanks for reporting @aydi

@cward do you know if anything changed here, was there any change to the convert_to_tf_lite function?

Best

Eoin

#3

@aydi

I can see this block was just updated 20hrs ago and no longer outputs the tflite file. You should do a new pull / build and deploy and only expect to see the pkl file.

Best

Eoin

#4

The .pkl file will be automatically converted to tflite server side.

1 Like
#5

I mean if i can generate the tflite file from the .pkl file as in keras custom block where they use convert_to_tf_lite function in the train.py file where its possible to add to edge impulse as a customblock.

i generate my .pkl file and i want to add this block to edge impulse because i want to test my code in openmv h7 board afterwards.

#6

Follow this section of the README, then the block will be in Edge Impulse and you can create a deployment for your choice of target.