.. _f-orttraininghelper:

module ``utils.orttraining_helper``
===================================







Short summary
+++++++++++++

module ``onnxcustom.utils.orttraining_helper``

ONNX manipulations to help build ONNX gradient graphs.


:githublink:`%|py|6`






Functions
+++++++++

+-------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+
| function                                                                                  | truncated documentation                                                                                               |
+===========================================================================================+=======================================================================================================================+
| :func:`_loss_elastic <onnxcustom.utils.orttraining_helper._loss_elastic>`                 | Implements mixture of losses l1 and l2.                                                                               |
+-------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+
| :func:`_loss_l1 <onnxcustom.utils.orttraining_helper._loss_l1>`                           | Implements loss l1.                                                                                                   |
+-------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+
| :func:`_loss_l2 <onnxcustom.utils.orttraining_helper._loss_l2>`                           | Implements loss l2.                                                                                                   |
+-------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+
| :func:`_loss_log <onnxcustom.utils.orttraining_helper._loss_log>`                         | This only works for a binary classification. The log loss is `'log(yt, yp) = (1-yt)\log(1-yp) - yt\log(yp)`, this ... |
+-------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+
| :func:`_rewrite_op_no_grad <onnxcustom.utils.orttraining_helper._rewrite_op_no_grad>`     | Rewrites operators with no gradient.                                                                                  |
+-------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+
| :func:`_unique_name <onnxcustom.utils.orttraining_helper._unique_name>`                   | Returns a name different from any name in *existing_names*.                                                           |
+-------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+
| :func:`add_loss_output <onnxcustom.utils.orttraining_helper.add_loss_output>`             | Modifies an ONNX graph to add operators to score and allow training.                                                  |
+-------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+
| :func:`get_train_initializer <onnxcustom.utils.orttraining_helper.get_train_initializer>` | Returns the list of initializers to train.                                                                            |
+-------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+
| :func:`penalty_loss_onnx <onnxcustom.utils.orttraining_helper.penalty_loss_onnx>`         | Returns onnx nodes to compute :math:`|w| \alpha + w^2 \beta` where :math:`\alpha=l1` and :math:`\beta=l2`.            |
+-------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------------------------------------------------+








Documentation
+++++++++++++

.. automodule:: onnxcustom.utils.orttraining_helper
    :members:
    :special-members: __init__
    :show-inheritance: