fake_quantize.py

import torch
from torch.nn import Module
from .observer import MovingAverageMinMaxObserver, HistogramObserver, MovingAveragePerChannelMinMaxObserver, _with_args
import re
from abc import ABC, abstractmethod

def _is_per_channel(qscheme: 'torch.qscheme') -> bool:
    return qscheme in [torch.per_channel_symmetric, torch.per_channel_affine]

def _is_per_tensor(qscheme: 'torch.qscheme') -> bool:
    return qscheme in [torch.per_tensor_symmetric, torch.per_tensor_affine]

class FakeQuantizeBase(ABC, Module):
    r""" Base fake quantize module
    Any fake quantize implementation should derive from this class.

    Concrete fake quantize module should follow the same API. In forward, they will update
    the statistics of the observed Tensor and fake quantize the input. They should also provide a
    `calculate_qparams` function that computes the quantization parameters given
    the collected statistics.

    """

    fake_quant_enabled: torch.Tensor
    observer_enabled: torch.Tensor

    def __init__(self):
        super().__init__()
        # fake_quant_enabled and observer_enabled are buffers to support their
        # replication in DDP. Data type is uint8 because NCCL does not support
        # bool tensors.
        self.register_buffer('fake_quant_enabled', torch.tensor([1], dtype=torch.uint8))
        self.register_buffer('observer_enabled', torch.tensor([1], dtype=torch.uint8))

    @abstractmethod
    def forward(self, x):
        pass

    @abstractmethod
    def calculate_qparams(self, **kwargs):
        pass

    @torch.jit.export
    def enable_fake_quant(self, enabled: bool = True) -> None:
        self.fake_quant_enabled[0] = 1 if enabled else 0

    @torch.jit.export
    def disable_fake_quant(self):
        self.enable_fake_quant(False)

    @torch.jit.export
    def enable_observer(self, enabled: bool = True) -> None:
        self.observer_enabled[0] = 1 if enabled else 0

    @torch.jit.export
    def disable_observer(self):
        self.enable_observer(False)

    with_args = classmethod(_with_args)

class FakeQuantize(FakeQuantizeBase):
    r""" Simulate the quantize and dequantize operations in training time.
    The output of this module is given by

    x_out = (clamp(round(x/scale + zero_point), quant_min, quant_max)-zero_point)*scale



    * :attr:`scale` defines the scale factor used for quantization.

    * :attr:`zero_point` specifies the quantized value to which 0 in floating point maps to

    * :attr:`quant_min` specifies the minimum allowable quantized value.

    * :attr:`quant_max` specifies the maximum allowable quantized value.

    * :attr:`fake_quant_enable` controls the application of fake quantization on tensors, note that
      statistics can still be updated.

    * :attr:`observer_enable` controls statistics collection on tensors

    * :attr:`dtype` specifies the quantized dtype that is being emulated with fake-quantization,
                    allowable values are torch.qint8 and torch.quint8. The values of quant_min and
                    quant_max should be chosen to be consistent with the dtype


    Args:
        observer (module): Module for observing statistics on input tensors and calculating scale
                           and zero-point.
        quant_min (int): The minimum allowable quantized value.
        quant_max (int): The maximum allowable quantized value.
        observer_kwargs (optional): Arguments for the observer module

    Attributes:
        observer (Module): User provided module that collects statistics on the input tensor and
                           provides a method to calculate scale and zero-point.

    """

    scale: torch.Tensor
    zero_point: torch.Tensor

    def __init__(self, observer=MovingAverageMinMaxObserver, quant_min=0, quant_max=255, **observer_kwargs):
        super().__init__()
        assert quant_min <= quant_max, \
            'quant_min must be less than or equal to quant_max'
        self.quant_min = quant_min
        self.quant_max = quant_max
        self.activation_post_process = observer(**observer_kwargs)
        assert torch.iinfo(self.activation_post_process.dtype).min <= quant_min, 'quant_min out of bound'
        assert quant_max <= torch.iinfo(self.activation_post_process.dtype).max, 'quant_max out of bound'
        self.register_buffer('scale', torch.tensor([1.0]))
        self.register_buffer('zero_point', torch.tensor([0]))
        self.dtype = self.activation_post_process.dtype
        self.qscheme = self.activation_post_process.qscheme
        self.ch_axis = self.activation_post_process.ch_axis \
            if hasattr(self.activation_post_process, 'ch_axis') else -1
        assert _is_per_channel(self.qscheme) or \
            _is_per_tensor(self.qscheme), \
            'Only per channel and per tensor quantization are supported in fake quantize' + \
            ' got qscheme: ' + str(self.qscheme)
        self.is_per_channel = _is_per_channel(self.qscheme)

    @torch.jit.export
    def calculate_qparams(self):
        return self.activation_post_process.calculate_qparams()

    def forward(self, X):
        if self.observer_enabled[0] == 1:
            self.activation_post_process(X.detach())
            _scale, _zero_point = self.calculate_qparams()
            _scale, _zero_point = _scale.to(self.scale.device), _zero_point.to(self.zero_point.device)
            self.scale.resize_(_scale.shape)
            self.scale.copy_(_scale)
            self.zero_point.resize_(_zero_point.shape)
            self.zero_point.copy_(_zero_point)

        if self.fake_quant_enabled[0] == 1:
            if self.is_per_channel:
                X = torch.fake_quantize_per_channel_affine(
                    X, self.scale, self.zero_point,
                    self.ch_axis, self.quant_min, self.quant_max)
            else:
                X = torch.fake_quantize_per_tensor_affine(
                    X, float(self.scale), int(self.zero_point),
                    self.quant_min, self.quant_max)
        return X

    @torch.jit.export
    def extra_repr(self):
        return 'fake_quant_enabled={}, observer_enabled={}, ' \
               'quant_min={}, quant_max={}, dtype={}, qscheme={}, ch_axis={}, ' \
               'scale={}, zero_point={}'.format(
                   self.fake_quant_enabled, self.observer_enabled,
                   self.quant_min, self.quant_max,
                   self.dtype, self.qscheme, self.ch_axis, self.scale, self.zero_point)

    def _save_to_state_dict(self, destination, prefix, keep_vars):
        # We cannot currently register scalar values as buffers, so need to manually
        # specify serialization here.
        super(FakeQuantize, self)._save_to_state_dict(destination, prefix, keep_vars)
        destination[prefix + 'scale'] = self.scale
        destination[prefix + 'zero_point'] = self.zero_point

    def _load_from_state_dict(self, state_dict, prefix, local_metadata, strict,
                              missing_keys, unexpected_keys, error_msgs):
        # Removing this function throws an error that the the size of the loaded tensor does not match the original size
        # i.e., These buffers start out with numel 0 and become numel 1 once they have their first forward pass.
        local_state = ['scale', 'zero_point']
        for name in local_state:
            key = prefix + name
            if key in state_dict:
                val = state_dict[key]
                # Custom handling to allow loading scale and zero_point
                # of size N into uninitialized buffers of size 0. The
                # buffers are resized here, and the values are copied in
                # the default state_dict loading code of the parent.
                if name == 'scale':
                    self.scale.resize_(val.shape)
                else:
                    assert name == 'zero_point'
                    self.zero_point.resize_(val.shape)
                # For torchscript module we need to update the attributes here since we do not
                # call the `_load_from_state_dict` function defined module.py
                if torch.jit.is_scripting():
                    if name == 'scale':
                        self.scale.copy_(val)
                    else:
                        assert name == 'zero_point'
                        self.zero_point.copy_(val)
            elif strict:
                missing_keys.append(key)
        super(FakeQuantize, self)._load_from_state_dict(state_dict, prefix, local_metadata, strict,
                                                        missing_keys, unexpected_keys, error_msgs)

class FixedQParamsFakeQuantize(FakeQuantizeBase):
    """ Simulate quantize and dequantize with fixed quantization
    parameters in training time. Only per tensor quantization
    is supported.
    Args:
        `scale` (float): fixed scale for the fake quantize module
        `zero_point` (int): fixed zero point for the fake quantize module
        `dtype`, `qscheme`, `quant_min`, `quant_max`
    """

    scale: torch.Tensor
    zero_point: torch.Tensor

    def __init__(self,
                 scale,
                 zero_point,
                 dtype=torch.quint8,
                 qscheme=torch.per_tensor_affine,
                 quant_min=0,
                 quant_max=255):
        super().__init__()
        assert quant_min <= quant_max, 'quant_min should be less than or equal to quant_max'
        self.quant_min = quant_min
        self.quant_max = quant_max
        self.register_buffer('scale', torch.tensor([scale]))
        self.register_buffer('zero_point', torch.tensor([zero_point]))
        self.dtype = dtype
        self.qscheme = qscheme
        assert _is_per_tensor(self.qscheme), 'Only per tensor quantization is supported' + \
            ' FixedQParamsFakeQuantize module, got qscheme:' + str(self.qscheme)

    def forward(self, X):
        if self.fake_quant_enabled[0] == 1:
            X = torch.fake_quantize_per_tensor_affine(X, float(self.scale),
                                                      int(self.zero_point), self.quant_min,
                                                      self.quant_max)
        return X

    @torch.jit.export
    def calculate_qparams(self):
        return self.scale, self.zero_point

    @torch.jit.export
    def extra_repr(self):
        return 'fake_quant_enabled={}, observer_enabled={}, scale={}, zero_point={}, ' \
               'dtype={}, quant_min={}, quant_max={}, qscheme={}'.format(
                   self.fake_quant_enabled, self.observer_enabled,
                   self.scale, self.zero_point, self.dtype,
                   self.quant_min, self.quant_max, self.qscheme)


default_fake_quant = FakeQuantize.with_args(observer=MovingAverageMinMaxObserver, quant_min=0, quant_max=255,
                                            dtype=torch.quint8, qscheme=torch.per_tensor_affine, reduce_range=True)
default_weight_fake_quant = FakeQuantize.with_args(observer=MovingAverageMinMaxObserver, quant_min=-128, quant_max=127,
                                                   dtype=torch.qint8, qscheme=torch.per_tensor_symmetric, reduce_range=False)

# TODO(future PR): remove these defaults and enforce activation functions
# to explicitly specify their output range
default_symmetric_fixed_qparams_fake_quant = FixedQParamsFakeQuantize.with_args(
    scale=2.0 / 256.0, zero_point=128, dtype=torch.quint8, quant_min=0, quant_max=255)
default_affine_fixed_qparams_fake_quant = FixedQParamsFakeQuantize.with_args(
    scale=1.0 / 256.0, zero_point=0, dtype=torch.quint8, quant_min=0, quant_max=255)

default_per_channel_weight_fake_quant = FakeQuantize.with_args(observer=MovingAveragePerChannelMinMaxObserver,
                                                               quant_min=-128,
                                                               quant_max=127,
                                                               dtype=torch.qint8,
                                                               qscheme=torch.per_channel_symmetric,
                                                               reduce_range=False,
                                                               ch_axis=0)
default_histogram_fake_quant = FakeQuantize.with_args(observer=HistogramObserver,
                                                      quant_min=0,
                                                      quant_max=255,
                                                      dtype=torch.quint8,
                                                      qscheme=torch.per_tensor_affine,
                                                      reduce_range=True)

def _is_fake_quant_script_module(mod):
    ''' Returns true if given mod is an instance of FakeQuantize script module.
    '''
    if isinstance(mod, torch.jit.RecursiveScriptModule):
        # qualified name looks like '__torch__.torch.quantization.fake_quantize.___torch_mangle_2.FakeQuantize'
        suffix = mod._c.qualified_name.split('.', 1)[1]
        name = re.sub(r'\.___torch_mangle_\d+', '', suffix)
        return name == 'torch.quantization.fake_quantize.FakeQuantize'
    return False

def disable_fake_quant(mod):
    if isinstance(mod, FakeQuantizeBase) or _is_fake_quant_script_module(mod):
        mod.disable_fake_quant()

def enable_fake_quant(mod):
    if isinstance(mod, FakeQuantizeBase) or _is_fake_quant_script_module(mod):
        mod.enable_fake_quant()

def disable_observer(mod):
    if isinstance(mod, FakeQuantizeBase) or _is_fake_quant_script_module(mod):
        mod.disable_observer()

def enable_observer(mod):
    if isinstance(mod, FakeQuantizeBase) or _is_fake_quant_script_module(mod):
        mod.enable_observer()