mirror of
https://github.com/luxfi/torus-ml.git
synced 2026-10-30 15:15:23 +00:00
80 lines
2.9 KiB
Python
80 lines
2.9 KiB
Python
# MIT License
|
|
#
|
|
# Copyright (c) 2019 Xilinx
|
|
#
|
|
# Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
# of this software and associated documentation files (the "Software"), to deal
|
|
# in the Software without restriction, including without limitation the rights
|
|
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
# copies of the Software, and to permit persons to whom the Software is
|
|
# furnished to do so, subject to the following conditions:
|
|
#
|
|
# The above copyright notice and this permission notice shall be included in all
|
|
# copies or substantial portions of the Software.
|
|
#
|
|
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
# SOFTWARE.
|
|
# Original file can be found at https://github.com/Xilinx/brevitas/blob/8c3d9de0113528cf6693c6474a13d802a66682c6/src/brevitas_examples/bnn_pynq/models/common.py
|
|
from brevitas.core.bit_width import BitWidthImplType
|
|
from brevitas.core.quant import QuantType
|
|
from brevitas.core.restrict_val import FloatToIntImplType, RestrictValueType
|
|
from brevitas.core.scaling import ScalingImplType
|
|
from brevitas.core.zero_point import ZeroZeroPoint
|
|
from brevitas.inject import ExtendedInjector
|
|
from brevitas.quant import Int8WeightPerTensorFloat
|
|
from brevitas.quant.solver import ActQuantSolver, WeightQuantSolver
|
|
from dependencies import value
|
|
|
|
|
|
class CommonQuant(ExtendedInjector):
|
|
bit_width_impl_type = BitWidthImplType.CONST
|
|
scaling_impl_type = ScalingImplType.CONST
|
|
restrict_scaling_type = RestrictValueType.FP
|
|
zero_point_impl = ZeroZeroPoint
|
|
float_to_int_impl_type = FloatToIntImplType.ROUND
|
|
scaling_per_output_channel = False
|
|
narrow_range = True
|
|
signed = True
|
|
|
|
@value
|
|
def quant_type(bit_width):
|
|
if bit_width is None:
|
|
return QuantType.FP
|
|
elif bit_width == 1:
|
|
return QuantType.BINARY
|
|
else:
|
|
return QuantType.INT
|
|
|
|
|
|
class CommonIntWeightPerTensorQuant(Int8WeightPerTensorFloat):
|
|
"""
|
|
Common per-tensor weight quantizer with bit-width set to None so that it is forced to be
|
|
specified by each layer.
|
|
"""
|
|
|
|
scaling_min_val = 2e-16
|
|
bit_width = None
|
|
|
|
|
|
class CommonIntWeightPerChannelQuant(CommonIntWeightPerTensorQuant):
|
|
"""
|
|
Common per-channel weight quantizer with bit-width set to None so that it is forced to be
|
|
specified by each layer.
|
|
"""
|
|
|
|
scaling_per_output_channel = True
|
|
|
|
|
|
class CommonWeightQuant(CommonQuant, WeightQuantSolver):
|
|
scaling_const = 1.0
|
|
|
|
|
|
class CommonActQuant(CommonQuant, ActQuantSolver):
|
|
min_val = -1.0
|
|
max_val = 1.0
|