Post Go back to editing


Category: Hardware

There is a trained model based on the tensorflow framework: the KWS model (tiny/benchmark/training/keyword_spotting/trained_models provided by tinyML Perf at master · mlcommons/tiny (

I now want to deploy directly to the MAX78000 EVK. The following questions are available:

  1. Can I use Maxim's ai8x synthesizer to generate C code and deploy it to the MAX78000 directly using Maxim's ai8x synthesizer?

  2. For models created using the tensorflow standard framework, such as the tinyML Perf kws model above, what are the points to note when writing the corresponding YAML file? to meet the requirements of the AI8X synthesizer.

  • It seems to me that we are have same problem. I have searched all of the documentation belonging to Max78000 evaluation board for a way to convert and use a Tensorflow model with the accelerator. I also have a big problem understanding what to write in a yaml file if the conversion was possible. I have the model in Tensorflow, Tensorflow Lite and onnx format and the flow charts shows that onnx-format could be used but the documentation seems to lack information about it.

    I hopa that some engineer could enlighten me how to do the conversion.