Skip to content

torch.nn.qat.rst

Latest commit

 

History

History
21 lines (14 loc) · 368 Bytes

torch.nn.qat.rst

File metadata and controls

21 lines (14 loc) · 368 Bytes

torch.nn.qat

This module implements versions of the key nn modules Conv2d() and Linear() which run in FP32 but with rounding applied to simulate the effect of INT8 quantization.

.. automodule:: torch.nn.qat

Conv2d

.. autoclass:: Conv2d
    :members:

Linear

.. autoclass:: Linear
    :members: