量化转换

量化卷积。

类

Conv2d	QuantConv2d 的别名
QuantConv2d	量化的2D卷积。
Conv3d	QuantConv3d 的别名
QuantConv3d	量化的3D卷积。
Conv1d	QuantConv1d 的别名
QuantConv1d	量化一维卷积。
ConvTranspose1d	QuantConvTranspose1d 的别名
ConvTranspose2d	QuantConvTranspose2d 的别名
ConvTranspose3d	QuantConvTranspose3d 的别名
QuantConvTranspose1d	量化的一维转置卷积。
QuantConvTranspose2d	量化的2D转置卷积。
QuantConvTranspose3d	量化的3D转置卷积。

Conv1d

QuantConv1d 的别名

Conv2d

QuantConv2d 的别名

Conv3d

QuantConv3d 的别名

ConvTranspose1d

QuantConvTranspose1d 的别名

ConvTranspose2d

QuantConvTranspose2d 的别名

ConvTranspose3d

QuantConvTranspose3d 的别名

class QuantConv1d

基础类：_LegacyQuantLinearConvBaseMixin, Conv1d

量化一维卷积。

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, trt_high_precision_dtype='Float', calibrator='max')

class QuantConv2d

基础类: _LegacyQuantLinearConvBaseMixin, Conv2d

量化的2D卷积。

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, trt_high_precision_dtype='Float', calibrator='max')

class QuantConv3d

基础类: _LegacyQuantLinearConvBaseMixin, Conv3d

量化的3D卷积。

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, trt_high_precision_dtype='Float', calibrator='max')

class QuantConvTranspose1d

基础类: _LegacyQuantLinearConvBaseMixin, ConvTranspose1d

量化的一维转置卷积。

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, trt_high_precision_dtype='Float', calibrator='max')

class QuantConvTranspose2d

基础类: _LegacyQuantLinearConvBaseMixin, ConvTranspose2d

量化的2D转置卷积。

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, trt_high_precision_dtype='Float', calibrator='max')

class QuantConvTranspose3d

基础类: _LegacyQuantLinearConvBaseMixin, ConvTranspose3d

量化的3D转置卷积。

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, trt_high_precision_dtype='Float', calibrator='max')