Whisper in ONNX with key value caching pt. 2

As promised in the last post, I will cover the ONNX conversion of the full Whisper model, including key-value caching, in this post. The full model resists a straightforward ONNX conversion due to its reliance on hooks and branching control flow. In this post I will discuss how to patch the model to make it exportable.


Read more

Table of contents

#c++ #htmx #java #js #ml #polars #python #rust #testing

    2023

  1. 2022

  2. 2021

  3. 2020

  4. 2019

  5. 2018

  6. 2017

  7. 2016

  8. 2015

  9. 2014

  10. 2013