Export NeurEco Compression model with GUI#

By default, NeurEco saves models in its binary format .ednn.
A NeurEco embed license allows to export .ednn models to the following formats.
NeurEco Tabular export formats#

Format

Precision

Description

FMU

double

The Functional Mock-up Interface (or FMI) defines a standardized interface to be used in computer simulations to develop complex cyber-physical systems. More details are available at these pages: https://fmi-standard.org/, and https://en.wikipedia.org/wiki/Functional_Mock-up_Interface

ONNX

double, float, float16

The Open Neural Network Exchange (ONNX) is an open-source artificial intelligence ecosystem of technology companies and research organizations that establish open standards for representing machine learning algorithms and software tools to promote innovation and collaboration in the AI sector. More details are available at these pages: https://onnx.ai, and https://en.wikipedia.org/wiki/Open_Neural_Network_Exchange

C format

double or float

generates a header file containing a C representation of the neural network inside a single procedure.

VBA format

double or float

generates a visual basic macro representing the neural network for the use from Excel files.
**Export** tab for **Compression** solution

Export tab for Compression solution#

To export a model in GUI:

  • Switch to Export tab

  • If applicable, choose Exported precision

  • Click on the button with a logo of the chosen format and choose a name and a destination of the model to save

Note

By default the last model in the checkpoint is exported.
Use the checkpoint slider on the bottom to choose any intermediate model and then export it in a chosen format.

Note

When exporting the model under a name file_name, its compression and decompession blocks are saved as well under the names file_name_compression and file_name_decompression. These files contain the NeurEco Regression models and can be used as usual (see Tabular Regression).

Note

See Select a model from a checkpoint and improve it to give the intermediate model a boost of final learning.