Skip to main content
Ctrl+K
Logo image Logo image

Site Navigation

  • Installation
  • Tutorials
  • API
  • Examples Gallery
  • Notebooks Gallery
  • Other pages
  • Blog Gallery

Site Navigation

  • Installation
  • Tutorials
  • API
  • Examples Gallery
  • Notebooks Gallery
  • Other pages
  • Blog Gallery

Links

  • Blog
  • Index
  • Module

Contents

  • Installation
  • Tutorials
  • API
  • Examples Gallery
  • Notebooks Gallery
  • Other pages
  • Blog Gallery

Information

  • FAQ
  • Glossary
  • README
  • Changes
  • License

Previous topic

Fast runtime with onnxruntime

Next topic

Issues when switching to float

Show Source

Complex Scenarios and discepancies#

Discrepencies may happen. Let’s see some unexpected cases.

Dealing with discrepancies#

  • Issues when switching to float
  • Funny discrepancies

Unexpected issues#

  • TfIdf and sparse matrices
  • Convert a pipeline with a LightGBM regressor
  • Batch predictions vs one-off predictions
  • Batch predictions without parallelization
  • Dealing with discrepancies (tf-idf)

previous

Fast runtime with onnxruntime

next

Issues when switching to float

On this page
  • Dealing with discrepancies
  • Unexpected issues

Show Source

Built with the PyData Sphinx Theme 0.12.0.