qat.py 22.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14
#   Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

G
guofei 已提交
15
import collections
16 17 18
import logging
import numpy as np
import sys
19
import os
20
import paddle
21 22 23
from paddle.fluid import dygraph, core, framework
from paddle.fluid.executor import Executor
from paddle.fluid.dygraph.io import INFER_MODEL_SUFFIX, INFER_PARAMS_SUFFIX
G
guofei 已提交
24 25
from paddle.nn import Linear, Conv2D, Conv2DTranspose, MaxPool2D, MaxPool1D, BatchNorm1D, BatchNorm2D, BatchNorm3D
from paddle.fluid.dygraph.nn import BatchNorm, Pool2D
26
from paddle.fluid.io import load_inference_model, save_inference_model
27
from paddle.nn.layer.activation import ReLU, LeakyReLU, Sigmoid, ReLU6, Tanh, Softmax, PReLU, Swish
28 29
from paddle.fluid.log_helper import get_logger
from . import quant_nn
30
from .. import quantization_pass
31

32
__all__ = ['ImperativeQuantAware', 'ImperativeCalcOutScale']
33 34 35 36

_logger = get_logger(
    __name__, logging.INFO, fmt='%(asctime)s-%(levelname)s: %(message)s')

37 38 39 40 41 42 43 44 45 46 47 48 49
_op_real_in_out_name = {
    "conv2d": [["Input", "Filter"], ["Output"]],
    "conv2d_transpose": [["Input", "Filter"], ["Output"]],
    "pool2d": [["X"], ["Out"]],
    "elementwise_add": [["X", "Y"], ["Out"]],
    "softmax": [["X"], ["Out"]],
    "relu": [["X"], ["Out"]],
    "relu6": [["X"], ["Out"]],
    "leaky_relu": [["X"], ["Out"]],
    "prelu": [["X"], ["Out"]],
    "tanh": [["X"], ["Out"]],
    "batch_norm": [["X"], ["Y"]],
    "sigmoid": [["X"], ["Out"]],
50
    "swish": [["X"], ["Out"]],
51 52
}

53 54 55 56 57 58 59 60 61 62 63 64 65

class ImperativeQuantAware(object):
    """
    Add the fake quant logic for given quantizable layers, namely add the quant_dequant
    computational logic both for activation inputs and weight inputs.
    """

    def __init__(self,
                 weight_bits=8,
                 activation_bits=8,
                 weight_quantize_type='abs_max',
                 activation_quantize_type='moving_average_abs_max',
                 moving_rate=0.9,
66 67 68 69 70
                 quantizable_layer_type=['Conv2D', 'Linear'],
                 weight_preprocess_layer=None,
                 act_preprocess_layer=None,
                 weight_quantize_layer=None,
                 act_quantize_layer=None):
71
        r"""
72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91
        The constructor for ImperativeQuantAware.

        Args:
            weight_bits(int): quantization bit number for weights,
                whereas the bias is not quantized.
            activation_bits(int): quantization bit number for activations.
            weight_quantize_type(str): quantization type for weights,
                which supports 'abs_max' now. The 'moving_average_abs_max'
                usually is not used for weights, since weights are fixed once the
                model is well trained.
            activation_quantize_type(str): quantization type for activations,
                which supports 'abs_max' and 'moving_average_abs_max' now.
                If using 'abs_max' mode, the quantization scale will be calculated
                dynamically each step in both training and testing period. If using
                'moving_average_abs_max', the static quantization scale will be calculated
                during training and used in inference.
            moving_rate(float): the parameter for 'moving_average_abs_max' quantization.
            quantizable_op_type(list[str]): List the type of layers that will be quantized. 
                Default is ['Conv2D', 'Linear']. The quantizable_op_type in
                QuantizationFreezePass and ConvertToInt8Pass must be the same as this.
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113
            weight_preprocess_layer(paddle.nn.Layer, optional): A paddle Layer that defines how to preprocess
                weight before quantization. Using this can quickly test if user's
                preprocess method works or not. The input is non-quantized
                weight and function returns processed weight to be quantized.
                If None, the weight will be quantized directly. Default is None.
            act_preprocess_layer(paddle.nn.Layer, optional): A paddle Layer that defines how to preprocess
                activation before quantization. Using this can quickly test if user's
                preprocess method works or not. The input is non-quantized
                activation and function returns processed activation to be quantized.
                If None, the activation will be quantized directly. Default is None.
            weight_quantize_layer(paddle.nn.Layer, optional): A paddle Layer that defines how to quantize weight.
                Using this can quickly test if user's quantization method works or not.
                In this layer, user should both define quantization method and
                dequantization method, that is, the function's input is non-quantized
                weight and returns dequantized weight. If None, will use
                quantization op defined by 'weight_quantize_type'. Default is None.
            act_quantize_layer(paddle.nn.Layer, optional): A paddle Layer that defines how to quantize activation.
                Using this can quickly test if user's quantization method works or not.
                In this layer, user should both define quantization method and
                dequantization method, that is, the function's input is non-quantized
                activation and returns dequantized activation. If None, will use
                quantization op defined by 'activation_quantize_type'. Default is None.
114

115 116 117 118 119 120
        Note:
            If user sets attribute 'skip_quant' to a Layer that support dynamic quantization and sets
            it to true, the layer would not be quantized during training. If this attribute is not sets
            or the attribute is false, the Layer would be qunatized in training.

        Examples 1:
121 122
        .. code-block:: python

123
            import paddle
124 125
            from paddle.fluid.contrib.slim.quantization \
                import ImperativeQuantAware
126
            from paddle.vision.models \
127 128 129 130 131 132 133 134 135 136
                import resnet
            
            model = resnet.resnet50(pretrained=True)

            imperative_qat = ImperativeQuantAware(
                weight_quantize_type='abs_max',
                activation_quantize_type='moving_average_abs_max')
            
            # Add the fake quant logical.
            # The original model will be rewrite.
137
            # The outscale of outputs in supportted layers would be calculated.
138 139 140 141 142 143
            imperative_qat.quantize(model)

            # Fine-tune the quantized model
            # ...
            
            # Save quant model for the inference.
144
            imperative_qat.save_quantized_model(
145 146 147 148 149
                layer=model,
                model_path="./resnet50_qat",
                input_spec=[
                    paddle.static.InputSpec(
                    shape=[None, 3, 224, 224], dtype='float32')])
150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192

        Examples 2:
        .. code-block:: python

            import paddle
            from paddle.fluid.contrib.slim.quantization \
                import ImperativeQuantAware

            class ImperativeModel(paddle.nn.Layer):
                def __init__(self):
                    super(ImperativeModel, self).__init__()
                    # self.linear_0 would skip the quantization.
                    self.linear_0 = paddle.nn.Linear(784, 400)
                    self.linear_0.skip_quant = True

                    # self.linear_1 would not skip the quantization.
                    self.linear_1 = paddle.nn.Linear(400, 10)
                    self.linear_1.skip_quant = False

                def forward(self, inputs):
                    x = self.linear_0(inputs)
                    x = self.linear_1(inputs)
                    return x

            model = ImperativeModel()
            imperative_qat = ImperativeQuantAware(
                weight_quantize_type='abs_max',
                activation_quantize_type='moving_average_abs_max')

            # Add the fake quant logical.
            # The original model will be rewrite.
            #
            # There is only one Layer(self.linear1) would be added the
            # fake quant logical.
            imperative_qat.quantize(model)

            # Fine-tune the quantized model
            # ...

            # Save quant model for the inference.
            imperative_qat.save_quantized_model(
                layer=model,
                model_path="./imperative_model_qat")
193 194 195 196 197
        """
        super(ImperativeQuantAware, self).__init__()
        self._weight_bits = weight_bits
        self._activation_bits = activation_bits
        self._moving_rate = moving_rate
H
huangxu96 已提交
198 199
        self._activation_quantize_type = activation_quantize_type
        self._weight_quantize_type = weight_quantize_type
200

201 202 203 204
        self._weight_pre_layer = weight_preprocess_layer
        self._act_pre_layer = act_preprocess_layer
        self._weight_quant_layer = weight_quantize_layer
        self._act_quant_layer = act_quantize_layer
205
        self._out_scale = ImperativeCalcOutScale()
206 207 208 209 210 211 212 213 214

        t_check = lambda method: method is None or issubclass(method, dygraph.layers.Layer)
        assert t_check(
            self._weight_pre_layer), "weight_preprocess should be nn.Layer"
        assert t_check(self._act_pre_layer), "act_preprocess should be nn.Layer"
        assert t_check(
            self._weight_quant_layer), "weight_quantize should be nn.Layer"
        assert t_check(self._act_quant_layer), "act_quantize should be nn.Layer"

H
huangxu96 已提交
215 216 217 218 219 220
        quant_type = {
            'abs_max', 'moving_average_abs_max', 'channel_wise_abs_max'
        }

        assert activation_quantize_type != 'channel_wise_abs_max', \
            "The activation quantization type does not support 'channel_wise_abs_max'."
221 222 223 224 225 226 227 228
        if activation_quantize_type not in quant_type:
            raise ValueError(
                "Unknown activation_quantize_type : '%s'. It can only be "
                "'abs_max' or 'moving_average_abs_max' now." %
                (str(activation_quantize_type)))
        if weight_quantize_type not in quant_type:
            raise ValueError(
                "Unknown weight_quantize_type: '%s'. It can only be "
H
huangxu96 已提交
229 230
                "'abs_max' or 'moving_average_abs_max' or 'channel_wise_abs_max' now."
                % (str(weight_quantize_type)))
231 232 233 234 235 236 237 238 239 240 241 242 243 244

        self._quant_layers_map = {'Conv2D': Conv2D, 'Linear': Linear}
        self._quantizable_layer_type = tuple(
            self._quant_layers_map[layer]
            if layer in self._quant_layers_map else layer
            for layer in quantizable_layer_type)
        for layer in self._quantizable_layer_type:
            assert not isinstance(
                layer, str), "{} is unspported to be quantized.".format(layer)

    def quantize(self, model):
        """
        According to weights' and activations' quantization types, the model will be added some fake
        quant ops, such as fake_quantize_dequantize_moving_average_abs_max, fake_quantize_dequantize_abs_max
245
        and so on. At the same time, the out_scale value of outputs would be calculated.
246 247 248 249 250 251 252 253 254

        Args:
            model(fluid.dygraph.Layer): the model to be quantized.
        Returns:
            None
        """
        for name, layer in model.named_sublayers():
            if not isinstance(layer, self._quantizable_layer_type):
                continue
255 256 257
            if hasattr(layer, "skip_quant") and layer.skip_quant == True:
                continue

258 259 260 261 262 263 264 265 266
            scopes = name.split('.')
            target = scopes[-1]
            obj = model
            parent = model
            for i in range(len(scopes) - 1):
                obj = getattr(parent, scopes[i])
                parent = obj

            quant_layer = self._get_quantized_counterpart(layer)
G
guofei 已提交
267
            setattr(quant_layer, "layer_name", layer.full_name())
268 269
            setattr(obj, target, quant_layer)

270 271
        self._out_scale.calc_out_scale(model)

272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
    def _get_quantized_counterpart(self, layer):
        quant_layers = tuple(self._quant_layers_map.values())
        quantized_counterpart = tuple('Quantized' + k
                                      for k in self._quant_layers_map.keys())

        predicate = lambda value: isinstance(layer, value)
        index_generator = (i for i, v in enumerate(quant_layers)
                           if predicate(v))

        try:
            index = next(index_generator)
        except StopIteration:
            _logger.fatal("The layer {} is unsupported to be quantized.".format(
                layer.full_name()))
            sys.exit(-1)

        quantized_layer = quant_nn.__dict__[quantized_counterpart[index]](
            layer, self._weight_bits, self._activation_bits, self._moving_rate,
290 291 292
            self._weight_quantize_type, self._activation_quantize_type,
            self._weight_pre_layer, self._act_pre_layer,
            self._weight_quant_layer, self._act_quant_layer)
293
        return quantized_layer
294

295 296 297
    def save_quantized_model(self, layer, path, input_spec=None, **config):
        self._out_scale.save_quantized_model(layer, path, input_spec, **config)

298 299

class ImperativeCalcOutScale(object):
300
    def __init__(self, moving_rate=0.9):
301 302 303 304 305 306 307 308 309
        """
        Add the logic of calculating and setting output quantization scales of some layers.
        These output quantization scales may be used by tensorRT or some other inference engines.

        Args:
            moving_rate(float): The decay coefficient of moving average. The default value is 0.9.
        """
        super(ImperativeCalcOutScale, self).__init__()
        self._moving_rate = moving_rate
310
        self._out_scale_layer_type_list = (
G
guofei 已提交
311 312 313
            BatchNorm, BatchNorm1D, BatchNorm2D, BatchNorm3D, Conv2D,
            Conv2DTranspose, LeakyReLU, Linear, PReLU, Pool2D, MaxPool1D,
            MaxPool2D, ReLU, ReLU6, Sigmoid, Softmax, Tanh, Swish)
314
        self._register_hook_handle_list = []
G
guofei 已提交
315
        self._out_scale_dict = collections.OrderedDict()
316 317 318 319 320 321 322 323 324 325 326 327 328 329

    def calc_out_scale(self, model):
        """
        Insert the `moving_average_abs_max_scale` op to calculate output scale of Specific layers in model.

        Args:
            model(fluid.dygraph.Layer): The target model which would be calculate the output quantization scale.

        Returns:
            None
        """
        assert isinstance(
            model, dygraph.Layer), "model must be the instance of dygraph.Layer"
        for _, layer in model.named_sublayers():
330
            if not isinstance(layer, self._out_scale_layer_type_list):
G
guofei 已提交
331 332
                if 'quantized_' not in layer.full_name():
                    continue
333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361
            forward_post_hook_handle = layer.register_forward_post_hook(
                self._forward_post_hook)
            self._register_hook_handle_list.append(forward_post_hook_handle)

    def save_quantized_model(self, layer, path, input_spec=None, **config):
        """
        Save the quantized model for the inference.

        Args:
            layer (Layer): The Layer to be saved.
            path (str): The path prefix to save model. The format is ``dirname/file_prefix`` or ``file_prefix``.
            input_spec (list[InputSpec|Tensor], optional): Describes the input of the saved model's forward 
                method, which can be described by InputSpec or example Tensor. If None, all input variables of 
                the original Layer's forward method would be the inputs of the saved model. Default None.
            **configs (dict, optional): Other save configuration options for compatibility. We do not 
                recommend using these configurations, they may be removed in the future. If not necessary, 
                DO NOT use them. Default None.
                The following options are currently supported:
                (1) output_spec (list[Tensor]): Selects the output targets of the saved model.
                By default, all return variables of original Layer's forward method are kept as the 
                output of the saved model. If the provided ``output_spec`` list is not all output variables, 
                the saved model will be pruned according to the given ``output_spec`` list. 

        Returns:
            None
        """

        assert isinstance(
            layer, dygraph.Layer), "model must be the instance of dygraph.Layer"
362
        is_dynamic_mode = False
363 364 365 366 367 368 369 370
        with dygraph.guard():
            layer.eval()
            for handle in self._register_hook_handle_list:
                handle.remove()
            for key in self._out_scale_dict:
                self._out_scale_dict[key] = float(self._out_scale_dict[key]
                                                  .numpy())

371 372 373 374
        if paddle.in_dynamic_mode():
            is_dynamic_mode = True
            paddle.enable_static()

G
guofei 已提交
375 376
        paddle.jit.save(layer=layer, path=path, input_spec=input_spec, **config)

377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397
        if core.is_compiled_with_cuda():
            place = core.CUDAPlace(0)
        else:
            place = core.CPUPlace()
        exe = Executor(place)

        file_prefix = os.path.basename(path)
        dirname = os.path.dirname(path)
        model_filename = file_prefix + INFER_MODEL_SUFFIX
        params_filename = file_prefix + INFER_PARAMS_SUFFIX

        [inference_program, feed_target_names, fetch_targets] = (
            load_inference_model(
                dirname=dirname,
                executor=exe,
                model_filename=model_filename,
                params_filename=params_filename))

        # Traverse all ops in the program and find out the op matching
        # the Layer in the dynamic graph.
        layer_var_dict = {}
G
guofei 已提交
398 399
        ops_list = [key for key, _ in self._out_scale_dict.items()]
        op_count = 0
400 401 402
        for block in inference_program.blocks:
            for op in block.ops:
                if op.type in _op_real_in_out_name:
G
guofei 已提交
403 404 405
                    if op.type in ["batch_norm", "pool2d"]:
                        if op.type == "pool2d" and op.attr(
                                "pooling_type") != "max":
406
                            continue
G
guofei 已提交
407 408
                        op_count = self.op_match(op, ops_list, op_count)
                        if op_count >= len(ops_list):
409
                            continue
G
guofei 已提交
410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442
                        op._set_attr('out_threshold',
                                     self._out_scale_dict[ops_list[op_count]])
                        op_count += 1
                    else:
                        output_var_names = quantization_pass._get_op_output_var_names(
                            op)
                        for output_var_name in output_var_names:
                            output_var_tensor = block.var(output_var_name)
                            if output_var_tensor.dtype not in [
                                    core.VarDesc.VarType.FP64,
                                    core.VarDesc.VarType.FP32
                            ]:
                                continue
                            # Because the Layer in dygraph may correspond to multiple ops
                            # in static program after being saved. To ensure correctness,
                            # the outscale collected for output of dygraph Layer can only
                            # be set to the last op in the corresponding ops in static program.
                            #
                            # We can judge the execution order of the ops which corresponding
                            # to dygraph Layer by the name of output. And use dict to save
                            # the corresponding relationship between the dygraph Layer and the
                            # static graph op that needs to set the outscale attribute.
                            if '.' not in output_var_name:
                                continue
                            dynamic_layer_name, var_name_suffix = output_var_name.split(
                                ".")
                            if dynamic_layer_name in layer_var_dict:
                                if layer_var_dict[dynamic_layer_name][
                                        0] < var_name_suffix:
                                    layer_var_dict[dynamic_layer_name] = [
                                        var_name_suffix, op
                                    ]
                            else:
443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468
                                layer_var_dict[dynamic_layer_name] = [
                                    var_name_suffix, op
                                ]

        # Because the naming styles of static and dynamic graph are different,
        # in order to avoid mistakes, we unify the name here.
        for (layer_name, var_name_op_list) in layer_var_dict.items():
            if 'prelu' in layer_name:
                layer_name = layer_name.replace('prelu', 'p_re_lu')
            if 'relu' in layer_name:
                layer_name = layer_name.replace('relu', 're_lu')
            if layer_name not in self._out_scale_dict:
                continue
            var_name_op_list[1]._set_attr('out_threshold',
                                          self._out_scale_dict[layer_name])

        # Save the processed program.
        save_inference_model(
            dirname=dirname,
            feeded_var_names=feed_target_names,
            target_vars=fetch_targets,
            executor=exe,
            main_program=inference_program.clone(),
            model_filename=model_filename,
            params_filename=params_filename)

469 470 471
        if is_dynamic_mode:
            paddle.disable_static()

G
guofei 已提交
472 473 474 475 476 477 478 479
    def op_match(self, op, ops_list, op_count):
        while op_count < len(ops_list) and op.type not in ops_list[op_count]:
            op_count += 1
        while op_count < len(ops_list) and op.type is "pool2d" and op.attr(
                "pooling_type") != "max":
            op_count += 1
        return op_count

480 481
    def _forward_post_hook(self, layer, input, output):
        assert isinstance(
482
            output, (core.VarBase, framework.Variable)
483 484 485 486 487 488 489 490 491
        ), "Multiple outputs are not currently supported in ImperativeOutScale."
        if output.dtype not in [
                core.VarDesc.VarType.FP32, core.VarDesc.VarType.FP64
        ]:
            return
        if not hasattr(layer, "_out_scale"):
            layer._out_scale = quant_nn.MovingAverageAbsMaxScale(
                output.name, self._moving_rate, output.dtype)
        scale_out = layer._out_scale(output)
G
guofei 已提交
492 493 494 495 496
        if hasattr(layer, 'layer_name'):
            layer_name = layer.layer_name
        else:
            layer_name = layer.full_name()
        self._out_scale_dict[layer_name] = scale_out