| 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: |