You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

linear.py 1.5 kB

123456789101112131415161718192021222324252627282930313233343536373839
  1. # MegEngine is Licensed under the Apache License, Version 2.0 (the "License")
  2. #
  3. # Copyright (c) 2014-2020 Megvii Inc. All rights reserved.
  4. #
  5. # Unless required by applicable law or agreed to in writing,
  6. # software distributed under the License is distributed on an
  7. # "AS IS" BASIS, WITHOUT ARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  8. from ...quantization.utils import fake_quant_bias
  9. from .. import linear as Float
  10. from .module import QATModule
  11. class Linear(Float.Linear, QATModule):
  12. r"""
  13. A :class:`~.QATModule` version of :class:`~.module.linear.Linear`.
  14. Could be applied with :class:`~.Observer` and :class:`~.FakeQuantize`.
  15. :param in_features: size of each input sample.
  16. :param out_features: size of each output sample.
  17. :param bias: If set to ``False``, the layer will not learn an additive bias.
  18. Default: True
  19. """
  20. def forward(self, x):
  21. w_qat = self.apply_quant_weight(self.weight)
  22. b_qat = fake_quant_bias(self.bias, x, w_qat)
  23. return self.apply_quant_activation(self._calc_linear(x, w_qat, b_qat))
  24. @classmethod
  25. def from_float_module(cls, float_module: Float.Linear):
  26. r"""
  27. Return a :class:`~.QATModule` instance converted from
  28. a float :class:`~.Module` instance.
  29. """
  30. qmod = cls(float_module.in_features, float_module.out_features)
  31. qmod.weight = float_module.weight
  32. qmod.bias = float_module.bias
  33. return qmod

MegEngine 安装包中集成了使用 GPU 运行代码所需的 CUDA 环境,不用区分 CPU 和 GPU 版。 如果想要运行 GPU 程序,请确保机器本身配有 GPU 硬件设备并安装好驱动。 如果你想体验在云端 GPU 算力平台进行深度学习开发的感觉,欢迎访问 MegStudio 平台