Source code for

import torch
import as nnqd
import as nni

__all__ = [

[docs]class LinearReLU(nnqd.Linear): r""" A LinearReLU module fused from Linear and ReLU modules that can be used for dynamic quantization. Supports both, FP16 and INT8 quantization. We adopt the same interface as :class:``. Attributes: Same as Examples:: >>> # xdoctest: +SKIP >>> m = nn.intrinsic.quantized.dynamic.LinearReLU(20, 30) >>> input = torch.randn(128, 20) >>> output = m(input) >>> print(output.size()) torch.Size([128, 30]) """ _FLOAT_MODULE = nni.LinearReLU # type: ignore[assignment] def __init__(self, in_features, out_features, bias=True, dtype=torch.qint8): super().__init__(in_features, out_features, bias, dtype) def forward(self, x: torch.Tensor) -> torch.Tensor: if self._packed_params.dtype == torch.qint8: # TODO check if we should set reduce_rage = True by default here Y = torch.ops.quantized.linear_relu_dynamic( x, self._packed_params._packed_params, reduce_range=True) elif self._packed_params.dtype == torch.float16: Y = torch.ops.quantized.linear_relu_dynamic_fp16( x, self._packed_params._packed_params) else: raise RuntimeError('Unsupported dtype on dynamic quantized linear relu!') return def _get_name(self): return 'DynamicQuantizedLinearReLU' @classmethod def from_float(cls, mod): return super(LinearReLU, cls).from_float(mod) @classmethod def from_reference(cls, ref_qlinear_relu): return super().from_reference(ref_qlinear_relu[0])


Access comprehensive developer documentation for PyTorch

View Docs


Get in-depth tutorials for beginners and advanced developers

View Tutorials


Find development resources and get your questions answered

View Resources