quant_conv — Model Optimizer 0.31.0 (original) (raw)
Quantized convolution.
Classes
Conv1d | alias of QuantConv1d |
---|---|
Conv2d | alias of QuantConv2d |
Conv3d | alias of QuantConv3d |
ConvTranspose1d | alias of QuantConvTranspose1d |
ConvTranspose2d | alias of QuantConvTranspose2d |
ConvTranspose3d | alias of QuantConvTranspose3d |
QuantConv1d | Quantized 1D convolution. |
QuantConv2d | Quantized 2D convolution. |
QuantConv3d | Quantized 3D convolution. |
QuantConvTranspose1d | Quantized 1D transposed convolution. |
QuantConvTranspose2d | Quantized 2D transposed convolution. |
QuantConvTranspose3d | Quantized 3D transposed convolution. |
Conv1d
alias of QuantConv1d
Conv2d
alias of QuantConv2d
Conv3d
alias of QuantConv3d
ConvTranspose1d
alias of QuantConvTranspose1d
ConvTranspose2d
alias of QuantConvTranspose2d
ConvTranspose3d
alias of QuantConvTranspose3d
class QuantConv1d
Bases: _LegacyQuantLinearConvBaseMixin
, Conv1d
Quantized 1D convolution.
default_quant_desc_weight = QuantizerAttributeConfig(enable=True, num_bits=8, axis=0, fake_quant=True, unsigned=False, narrow_range=False, learn_amax=False, type='static', block_sizes=None, bias=None, trt_high_precision_dtype='Float', calibrator='max', rotate=False)
class QuantConv2d
Bases: _LegacyQuantLinearConvBaseMixin
, Conv2d
Quantized 2D convolution.
default_quant_desc_weight = QuantizerAttributeConfig(enable=True, num_bits=8, axis=0, fake_quant=True, unsigned=False, narrow_range=False, learn_amax=False, type='static', block_sizes=None, bias=None, trt_high_precision_dtype='Float', calibrator='max', rotate=False)
class QuantConv3d
Bases: _LegacyQuantLinearConvBaseMixin
, Conv3d
Quantized 3D convolution.
default_quant_desc_weight = QuantizerAttributeConfig(enable=True, num_bits=8, axis=0, fake_quant=True, unsigned=False, narrow_range=False, learn_amax=False, type='static', block_sizes=None, bias=None, trt_high_precision_dtype='Float', calibrator='max', rotate=False)
class QuantConvTranspose1d
Bases: _LegacyQuantLinearConvBaseMixin
, ConvTranspose1d
Quantized 1D transposed convolution.
default_quant_desc_weight = QuantizerAttributeConfig(enable=True, num_bits=8, axis=0, fake_quant=True, unsigned=False, narrow_range=False, learn_amax=False, type='static', block_sizes=None, bias=None, trt_high_precision_dtype='Float', calibrator='max', rotate=False)
class QuantConvTranspose2d
Bases: _LegacyQuantLinearConvBaseMixin
, ConvTranspose2d
Quantized 2D transposed convolution.
default_quant_desc_weight = QuantizerAttributeConfig(enable=True, num_bits=8, axis=0, fake_quant=True, unsigned=False, narrow_range=False, learn_amax=False, type='static', block_sizes=None, bias=None, trt_high_precision_dtype='Float', calibrator='max', rotate=False)
class QuantConvTranspose3d
Bases: _LegacyQuantLinearConvBaseMixin
, ConvTranspose3d
Quantized 3D transposed convolution.
default_quant_desc_weight = QuantizerAttributeConfig(enable=True, num_bits=8, axis=0, fake_quant=True, unsigned=False, narrow_range=False, learn_amax=False, type='static', block_sizes=None, bias=None, trt_high_precision_dtype='Float', calibrator='max', rotate=False)