Skip to content

Mace
Mace

Xiaomi / Mace

通知 107
Star 40
Fork 27
Mace
Mace
提交 320b509c 编写于 作者: L liyin
浏览文件
操作

Refactor model converter & encrypt process

上级 64419daa
隐藏空白更改
内联 并排
Showing with 1217 addition and 1295 deletion
.gitignore
浏览文件 @ 320b509c
...@@ -28,3 +28,4 @@ examples/android/macelibrary/src/main/cpp/include/ ...@@ -28,3 +28,4 @@ examples/android/macelibrary/src/main/cpp/include/
examples/android/macelibrary/src/main/cpp/lib/arm64-v8a/ examples/android/macelibrary/src/main/cpp/lib/arm64-v8a/
examples/android/macelibrary/src/main/jniLibs/arm64-v8a/ examples/android/macelibrary/src/main/jniLibs/arm64-v8a/
tools/python/py_proto/*_pb2.py
docs/user_guide/quantization_usage.rst
浏览文件 @ 320b509c
...@@ -62,7 +62,7 @@ MACE provides tools to do statistics with following steps: ...@@ -62,7 +62,7 @@ MACE provides tools to do statistics with following steps:
.. code:: sh .. code:: sh
python mace/python/tools/quantization/quantize_stat.py --log_file range_log > overall_range python tools/python/tools/quantize/quantize_stat.py --log_file range_log > overall_range
4. Convert quantized model (by setting `target_abis` to the final target abis, e.g., `armeabi-v7a`, 4. Convert quantized model (by setting `target_abis` to the final target abis, e.g., `armeabi-v7a`,
......
mace/codegen/BUILD.bazel
浏览文件 @ 320b509c
...@@ -9,8 +9,8 @@ load("//mace:mace.bzl", "encrypt_opencl_kernel_genrule", "mace_version_genrule") ...@@ -9,8 +9,8 @@ load("//mace:mace.bzl", "encrypt_opencl_kernel_genrule", "mace_version_genrule")
cc_library( cc_library(
name = "generated_models", name = "generated_models",
srcs = glob(["models/*/*.cc"]), srcs = glob(["models/**/*.cc"]),
hdrs = glob(["models/*/*.h"]), hdrs = glob(["models/**/*.h"]),
copts = [ copts = [
"-Werror", "-Werror",
"-Wextra", "-Wextra",
......
mace/python/tools/BUILD.bazel
浏览文件 @ 320b509c
py_library(
name = "quantization_lib",
srcs = [
"quantization/quantize_util.py",
],
srcs_version = "PY2AND3",
)
py_library(
name = "converter_lib",
srcs = [
"convert_util.py",
"converter_tool/base_converter.py",
"converter_tool/caffe_converter.py",
"converter_tool/hexagon_converter.py",
"converter_tool/onnx_converter.py",
"converter_tool/shape_inference.py",
"converter_tool/tensorflow_converter.py",
"converter_tool/apu_converter.py",
"converter_tool/transformer.py",
"graph_util.py",
],
srcs_version = "PY2AND3",
deps = [
":quantization_lib",
"//mace/proto:mace_py",
"//third_party/caffe:caffe_py",
],
)
py_library(
name = "model_saver_lib",
srcs = [
"model_saver.py",
],
srcs_version = "PY2AND3",
deps = [
"//mace/proto:mace_py",
],
)
py_binary(
name = "converter",
srcs = ["converter.py"],
srcs_version = "PY2AND3",
deps = [
":converter_lib",
":model_saver_lib",
"//mace/python/tools/visualization:visualization_lib",
"@six_archive//:six",
],
)
py_binary( py_binary(
name = "archive_static_lib", name = "archive_static_lib",
srcs = ["archive_static_lib.py"], srcs = ["archive_static_lib.py"],
......
mace/python/tools/binary_codegen.py 已删除 100644 → 0
浏览文件 @ 64419daa
# Copyright 2018 The MACE 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.
import argparse
import os
import sys
import struct
import jinja2
import numpy as np
import six
# python mace/python/tools/binary_codegen.py \
# --binary_dirs=${BIN_FILE} \
# --binary_file_name=mace_run.config \
# --output_path=${CODE_GEN_PATH} --variable_name=kTuningParamsData
FLAGS = None
def generate_cpp_source(binary_dirs, binary_file_name, variable_name):
data_map = {}
for binary_dir in binary_dirs.split(","):
binary_path = os.path.join(binary_dir, binary_file_name)
if not os.path.exists(binary_path):
continue
with open(binary_path, "rb") as f:
binary_array = np.fromfile(f, dtype=np.uint8)
six.print_("Generate binary from", binary_path)
idx = 0
size, = struct.unpack("Q", binary_array[idx:idx + 8])
idx += 8
for _ in six.moves.range(size):
key_size, = struct.unpack("i", binary_array[idx:idx + 4])
idx += 4
key, = struct.unpack(
str(key_size) + "s", binary_array[idx:idx + key_size])
idx += key_size
params_size, = struct.unpack("i", binary_array[idx:idx + 4])
idx += 4
data_map[key] = []
count = params_size / 4
params = struct.unpack(
str(count) + "i", binary_array[idx:idx + params_size])
for i in params:
data_map[key].append(i)
idx += params_size
env = jinja2.Environment(loader=jinja2.FileSystemLoader(sys.path[0]))
return env.get_template('str2vec_maps.cc.jinja2').render(
maps=data_map,
data_type='unsigned int',
variable_name=variable_name)
def tuning_param_codegen(binary_dirs,
binary_file_name,
output_path,
variable_name):
cpp_binary_source = generate_cpp_source(
binary_dirs, binary_file_name, variable_name)
if os.path.isfile(output_path):
os.remove(output_path)
with open(output_path, "w") as w_file:
w_file.write(cpp_binary_source)
def parse_args():
"""Parses command line arguments."""
parser = argparse.ArgumentParser()
parser.add_argument(
"--binary_dirs", type=str, default="", help="The binaries file path.")
parser.add_argument(
"--binary_file_name",
type=str,
default="mace_run.config",
help="The binary file name.")
parser.add_argument(
"--output_path",
type=str,
default="",
help="The path of generated C++ source file which contains the binary."
)
parser.add_argument(
"--variable_name",
type=str,
default="kTuningParamsData",
help="global variable name.")
return parser.parse_known_args()
if __name__ == '__main__':
FLAGS, unparsed = parse_args()
tuning_param_codegen(FLAGS.binary_dirs,
FLAGS.binary_file_name,
FLAGS.output_path,
FLAGS.variable_name)
mace/python/tools/convert_util.py 已删除 100644 → 0
浏览文件 @ 64419daa
# Copyright 2018 The MACE 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.
import enum
def mace_check(condition, msg):
if not condition:
raise Exception(msg)
def roundup_div4(value):
return int((value + 3) // 4)
class OpenCLBufferType(enum.Enum):
CONV2D_FILTER = 0
IN_OUT_CHANNEL = 1
ARGUMENT = 2
IN_OUT_HEIGHT = 3
IN_OUT_WIDTH = 4
WINOGRAD_FILTER = 5
DW_CONV2D_FILTER = 6
WEIGHT_HEIGHT = 7
WEIGHT_WIDTH = 8
def calculate_image_shape(buffer_type, shape, winograd_blk_size=0):
# keep the same with mace/kernel/opencl/helper.cc
image_shape = [0, 0]
if buffer_type == OpenCLBufferType.CONV2D_FILTER:
mace_check(len(shape) == 4, "Conv2D filter buffer should be 4D")
image_shape[0] = shape[1]
image_shape[1] = shape[2] * shape[3] * roundup_div4(shape[0])
elif buffer_type == OpenCLBufferType.IN_OUT_CHANNEL:
mace_check(len(shape) == 2 or len(shape) == 4,
"input/output buffer should be 2D|4D")
if len(shape) == 4:
image_shape[0] = roundup_div4(shape[3]) * shape[2]
image_shape[1] = shape[0] * shape[1]
elif len(shape) == 2:
image_shape[0] = roundup_div4(shape[1])
image_shape[1] = shape[0]
elif buffer_type == OpenCLBufferType.ARGUMENT:
mace_check(len(shape) == 1,
"Argument buffer should be 1D not " + str(shape))
image_shape[0] = roundup_div4(shape[0])
image_shape[1] = 1
elif buffer_type == OpenCLBufferType.IN_OUT_HEIGHT:
if len(shape) == 4:
image_shape[0] = shape[2] * shape[3]
image_shape[1] = shape[0] * roundup_div4(shape[1])
elif len(shape) == 2:
image_shape[0] = shape[0]
image_shape[1] = roundup_div4(shape[1])
elif buffer_type == OpenCLBufferType.IN_OUT_WIDTH:
mace_check(len(shape) == 4, "Input/output buffer should be 4D")
image_shape[0] = roundup_div4(shape[2]) * shape[3]
image_shape[1] = shape[0] * shape[1]
elif buffer_type == OpenCLBufferType.WINOGRAD_FILTER:
mace_check(len(shape) == 4, "Winograd filter buffer should be 4D")
image_shape[0] = roundup_div4(shape[1])
image_shape[1] = (shape[0] * (winograd_blk_size + 2)
* (winograd_blk_size + 2))
elif buffer_type == OpenCLBufferType.DW_CONV2D_FILTER:
mace_check(len(shape) == 4, "Winograd filter buffer should be 4D")
image_shape[0] = shape[0] * shape[2] * shape[3]
image_shape[1] = roundup_div4(shape[1])
elif buffer_type == OpenCLBufferType.WEIGHT_HEIGHT:
mace_check(len(shape) == 4, "Weight buffer should be 4D")
image_shape[0] = shape[1] * shape[2] * shape[3]
image_shape[1] = roundup_div4(shape[0])
elif buffer_type == OpenCLBufferType.WEIGHT_WIDTH:
mace_check(len(shape) == 4, "Weight buffer should be 4D")
image_shape[0] = roundup_div4(shape[1]) * shape[2] * shape[3]
image_shape[1] = shape[0]
else:
mace_check(False, "OpenCL Image do not support type "
+ str(buffer_type))
return image_shape
mace/python/tools/converter.py 已删除 100644 → 0
浏览文件 @ 64419daa
# Copyright 2018 The MACE 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.
import argparse
import sys
import hashlib
import os.path
import copy
import six
from mace.proto import mace_pb2
from mace.python.tools import model_saver
from mace.python.tools.converter_tool import base_converter as cvt
from mace.python.tools.converter_tool import transformer
from mace.python.tools.convert_util import mace_check
from mace.python.tools.visualization import visualize_model
# ./bazel-bin/mace/python/tools/tf_converter --model_file quantized_test.pb \
# --output quantized_test_dsp.pb \
# --runtime dsp \
# --input_dim input_node,1,28,28,3
FLAGS = None
device_type_map = {'cpu': cvt.DeviceType.CPU.value,
'gpu': cvt.DeviceType.GPU.value,
'dsp': cvt.DeviceType.HEXAGON.value,
'hta': cvt.DeviceType.HTA.value,
'apu': cvt.DeviceType.APU.value,
'cpu+gpu': cvt.DeviceType.CPU.value}
data_format_map = {
'NONE': cvt.DataFormat.NONE,
'NHWC': cvt.DataFormat.NHWC,
'NCHW': cvt.DataFormat.NCHW,
'OIHW': cvt.DataFormat.OIHW,
}
data_type_map = {
'float32': mace_pb2.DT_FLOAT,
'int32': mace_pb2.DT_INT32,
}
def parse_data_type(data_type, device_type):
if device_type == cvt.DeviceType.CPU.value or \
device_type == cvt.DeviceType.GPU.value:
if data_type == 'fp32_fp32':
return mace_pb2.DT_FLOAT
else:
return mace_pb2.DT_HALF
elif device_type == cvt.DeviceType.HEXAGON.value or \
device_type == cvt.DeviceType.HTA.value:
return mace_pb2.DT_FLOAT
elif device_type == cvt.DeviceType.APU.value:
return mace_pb2.DT_FLOAT
else:
print("Invalid device type: " + str(device_type))
def file_checksum(fname):
hash_func = hashlib.sha256()
with open(fname, "rb") as f:
for chunk in iter(lambda: f.read(4096), b""):
hash_func.update(chunk)
return hash_func.hexdigest()
def split_shape(shape):
if shape.strip() == "":
return []
else:
return shape.split(',')
def parse_int_array_from_str(ints_str):
return [int(i) for i in split_shape(ints_str)]
def parse_float_array_from_str(floats_str):
return [float(i) for i in floats_str.split(',')]
def transpose_shape(shape, dst_order):
t_shape = [0] * len(shape)
for i in range(len(shape)):
t_shape[i] = shape[dst_order[i]]
return t_shape
def main(unused_args):
if not os.path.isfile(FLAGS.model_file):
six.print_("Input graph file '" +
FLAGS.model_file +
"' does not exist!", file=sys.stderr)
sys.exit(-1)
model_checksum = file_checksum(FLAGS.model_file)
if FLAGS.model_checksum != "" and FLAGS.model_checksum != model_checksum:
six.print_("Model checksum mismatch: %s != %s" %
(model_checksum, FLAGS.model_checksum), file=sys.stderr)
sys.exit(-1)
weight_checksum = None
if FLAGS.platform == 'caffe':
if not os.path.isfile(FLAGS.weight_file):
six.print_("Input weight file '" + FLAGS.weight_file +
"' does not exist!", file=sys.stderr)
sys.exit(-1)
weight_checksum = file_checksum(FLAGS.weight_file)
if FLAGS.weight_checksum != "" and \
FLAGS.weight_checksum != weight_checksum:
six.print_("Weight checksum mismatch: %s != %s" %
(weight_checksum, FLAGS.weight_checksum),
file=sys.stderr)
sys.exit(-1)
if FLAGS.platform not in ['tensorflow', 'caffe', 'onnx']:
six.print_("platform %s is not supported." % FLAGS.platform,
file=sys.stderr)
sys.exit(-1)
if FLAGS.runtime not in ['cpu', 'gpu', 'dsp', 'hta', 'apu', 'cpu+gpu']:
six.print_("runtime %s is not supported." % FLAGS.runtime,
file=sys.stderr)
sys.exit(-1)
option = cvt.ConverterOption()
if FLAGS.graph_optimize_options:
option.transformer_option = FLAGS.graph_optimize_options.split(',')
option.winograd = FLAGS.winograd
option.quantize = FLAGS.quantize
option.quantize_large_weights = FLAGS.quantize_large_weights
option.quantize_range_file = FLAGS.quantize_range_file
option.change_concat_ranges = FLAGS.change_concat_ranges
option.cl_mem_type = FLAGS.cl_mem_type
option.device = device_type_map[FLAGS.runtime]
option.data_type = parse_data_type(FLAGS.data_type, option.device)
input_node_names = FLAGS.input_node.split(',')
input_data_types = FLAGS.input_data_types.split(',')
input_node_shapes = FLAGS.input_shape.split(':')
input_node_formats = FLAGS.input_data_formats.split(",")
if FLAGS.input_range:
input_node_ranges = FLAGS.input_range.split(':')
else:
input_node_ranges = []
if len(input_node_names) != len(input_node_shapes):
raise Exception('input node count and shape count do not match.')
for i in six.moves.range(len(input_node_names)):
input_node = cvt.NodeInfo()
input_node.name = input_node_names[i]
input_node.data_type = data_type_map[input_data_types[i]]
input_node.data_format = data_format_map[input_node_formats[i]]
input_node.shape = parse_int_array_from_str(input_node_shapes[i])
if input_node.data_format == cvt.DataFormat.NCHW and\
len(input_node.shape) == 4:
input_node.shape = transpose_shape(input_node.shape, [0, 2, 3, 1])
input_node.data_format = cvt.DataFormat.NHWC
if len(input_node_ranges) > i:
input_node.range = parse_float_array_from_str(input_node_ranges[i])
option.add_input_node(input_node)
output_node_names = FLAGS.output_node.split(',')
output_data_types = FLAGS.output_data_types.split(',')
output_node_shapes = FLAGS.output_shape.split(':')
output_node_formats = FLAGS.output_data_formats.split(",")
if len(output_node_names) != len(output_node_shapes):
raise Exception('output node count and shape count do not match.')
for i in six.moves.range(len(output_node_names)):
output_node = cvt.NodeInfo()
output_node.name = output_node_names[i]
output_node.data_type = data_type_map[output_data_types[i]]
output_node.data_format = data_format_map[output_node_formats[i]]
output_node.shape = parse_int_array_from_str(output_node_shapes[i])
if output_node.data_format == cvt.DataFormat.NCHW and\
len(output_node.shape) == 4:
output_node.shape = transpose_shape(output_node.shape,
[0, 2, 3, 1])
output_node.data_format = cvt.DataFormat.NHWC
option.add_output_node(output_node)
if FLAGS.check_node != '':
check_node_names = FLAGS.check_node.split(',')
check_node_shapes = FLAGS.check_shape.split(':')
if len(check_node_names) != len(check_node_shapes):
raise Exception('check node count and shape count do not match.')
for i in six.moves.range(len(check_node_names)):
check_node = cvt.NodeInfo()
check_node.name = check_node_names[i]
check_node.shape = parse_int_array_from_str(check_node_shapes[i])
option.add_check_node(check_node)
else:
option.check_nodes = option.output_nodes
option.build()
print("Transform model to one that can better run on device")
if FLAGS.platform == 'tensorflow':
from mace.python.tools.converter_tool import tensorflow_converter
converter = tensorflow_converter.TensorflowConverter(
option, FLAGS.model_file)
elif FLAGS.platform == 'caffe':
from mace.python.tools.converter_tool import caffe_converter
converter = caffe_converter.CaffeConverter(option,
FLAGS.model_file,
FLAGS.weight_file)
elif FLAGS.platform == 'onnx':
from mace.python.tools.converter_tool import onnx_converter
converter = onnx_converter.OnnxConverter(option, FLAGS.model_file)
else:
six.print_("Mace do not support platorm %s yet." % FLAGS.platform,
file=sys.stderr)
exit(1)
output_graph_def = converter.run()
mace_transformer = transformer.Transformer(
option, output_graph_def)
output_graph_def, quantize_activation_info = mace_transformer.run()
if option.device in [cvt.DeviceType.HEXAGON.value,
cvt.DeviceType.HTA.value]:
from mace.python.tools.converter_tool import hexagon_converter
converter = hexagon_converter.HexagonConverter(
option, output_graph_def, quantize_activation_info)
output_graph_def = converter.run()
elif FLAGS.runtime == 'apu':
if FLAGS.platform != 'tensorflow':
raise Exception('apu only support model from tensorflow')
from mace.python.tools.converter_tool import apu_converter
converter = apu_converter.ApuConverter(
option, output_graph_def, quantize_activation_info)
output_graph_def = converter.run()
try:
visualizer = visualize_model.ModelVisualizer(FLAGS.model_tag,
output_graph_def)
visualizer.save_html()
except: # noqa
print("Failed to visualize model:", sys.exc_info()[0])
model_saver.save_model(
option, output_graph_def, model_checksum, weight_checksum,
FLAGS.template_dir, FLAGS.obfuscate, FLAGS.model_tag,
FLAGS.output_dir,
FLAGS.embed_model_data,
FLAGS.winograd,
FLAGS.model_graph_format)
def str2bool(v):
if v.lower() in ('yes', 'true', 't', 'y', '1'):
return True
elif v.lower() in ('no', 'false', 'f', 'n', '0'):
return False
else:
raise argparse.ArgumentTypeError('Boolean value expected.')
def parse_args():
"""Parses command line arguments."""
parser = argparse.ArgumentParser()
parser.register("type", "bool", lambda v: v.lower() == "true")
parser.add_argument(
"--model_file",
type=str,
default="",
help="TensorFlow \'GraphDef\' file to load, "
"Onnx model file .onnx to load, "
"Caffe prototxt file to load.")
parser.add_argument(
"--weight_file", type=str, default="", help="Caffe data file to load.")
parser.add_argument(
"--model_checksum",
type=str,
default="",
help="Model file sha256 checksum")
parser.add_argument(
"--weight_checksum",
type=str,
default="",
help="Weight file sha256 checksum")
parser.add_argument(
"--output_dir",
type=str,
default="",
help="File to save the output graph to.")
parser.add_argument(
"--runtime", type=str, default="", help="Runtime: cpu/gpu/dsp/apu")
parser.add_argument(
"--input_node",
type=str,
default="input_node",
help="e.g., input_node")
parser.add_argument(
"--input_data_types",
type=str,
default="float32",
help="e.g., float32|int32")
parser.add_argument(
"--input_data_formats",
type=str,
default="NHWC",
help="e.g., NHWC,NONE")
parser.add_argument(
"--output_node", type=str, default="softmax", help="e.g., softmax")
parser.add_argument(
"--output_data_types",
type=str,
default="float32",
help="e.g., float32|int32")
parser.add_argument(
"--output_data_formats",
type=str,
default="NHWC",
help="e.g., NHWC,NONE")
parser.add_argument(
"--check_node", type=str, default="softmax", help="e.g., softmax")
parser.add_argument(
"--template_dir", type=str, default="", help="template path")
parser.add_argument(
"--obfuscate",
type=str2bool,
nargs='?',
const=False,
default=False,
help="obfuscate model names")
parser.add_argument(
"--model_tag",
type=str,
default="",
help="model tag for generated function and namespace")
parser.add_argument(
"--winograd",
type=int,
default=0,
help="Which version of winograd convolution to use. [2 | 4]")
parser.add_argument(
"--dsp_mode", type=int, default=0, help="dsp run mode, defalut=0")
parser.add_argument(
"--input_shape", type=str, default="", help="input shape.")
parser.add_argument(
"--input_range", type=str, default="", help="input range.")
parser.add_argument(
"--output_shape", type=str, default="", help="output shape.")
parser.add_argument(
"--check_shape", type=str, default="", help="check shape.")
parser.add_argument(
"--platform",
type=str,
default="tensorflow",
help="tensorflow/caffe/onnx")
parser.add_argument(
"--embed_model_data",
type=str2bool,
default=True,
help="embed model data.")
parser.add_argument(
"--model_graph_format",
type=str,
default="file",
help="[file|code] build models to code" +
"or `Protobuf` file.")
parser.add_argument(
"--data_type",
type=str,
default="fp16_fp32",
help="fp16_fp32/fp32_fp32")
parser.add_argument(
"--graph_optimize_options",
type=str,
default="",
help="graph optimize options")
parser.add_argument(
"--quantize",
type=str2bool,
nargs='?',
const=False,
default=False,
help="quantize model")
parser.add_argument(
"--quantize_large_weights",
type=str2bool,
nargs='?',
const=False,
default=False,
help="quantize large weights for compression")
parser.add_argument(
"--quantize_range_file",
type=str,
default="",
help="file path of quantize range for each tensor")
parser.add_argument(
"--change_concat_ranges",
type=str2bool,
nargs='?',
const=False,
default=False,
help="change ranges to use memcpy for quantized concat")
parser.add_argument(
"--cl_mem_type",
type=str,
default="image",
help="which memory type to use.[image|buffer]")
return parser.parse_known_args()
if __name__ == '__main__':
FLAGS, unparsed = parse_args()
main(unused_args=[sys.argv[0]] + unparsed)
mace/python/tools/encrypt_opencl_codegen.py
浏览文件 @ 320b509c
...@@ -55,7 +55,8 @@ def create_output_dir(dir_path): ...@@ -55,7 +55,8 @@ def create_output_dir(dir_path):
def write_cl_encrypted_kernel_to_file( def write_cl_encrypted_kernel_to_file(
encrypted_code_maps, template_path, output_path): encrypted_code_maps, template_path, output_path):
env = jinja2.Environment(loader=jinja2.FileSystemLoader(sys.path[0])) cwd = os.path.dirname(__file__)
env = jinja2.Environment(loader=jinja2.FileSystemLoader(cwd))
cl_encrypted_kernel = env.get_template(template_path).render( cl_encrypted_kernel = env.get_template(template_path).render(
tag='codegen', tag='codegen',
maps=encrypted_code_maps, maps=encrypted_code_maps,
......
mace/python/tools/graph_util.py 已删除 100644 → 0
浏览文件 @ 64419daa
# Copyright 2018 The MACE 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.
import tensorflow as tf
from mace.proto import mace_pb2
from collections import OrderedDict
def sort_tf_node(node, nodes_map, ordered_nodes_map):
if node.name not in ordered_nodes_map:
for input_tensor_name in node.input:
input_node_name = input_tensor_name.split(':')[
0] if ':' in input_tensor_name else input_tensor_name
if input_node_name not in nodes_map or \
input_node_name in ordered_nodes_map:
continue
input_node = nodes_map[input_node_name]
sort_tf_node(input_node, nodes_map, ordered_nodes_map)
ordered_nodes_map[node.name] = node
def sort_tf_graph(graph_def):
nodes_map = {}
ordered_nodes_map = OrderedDict()
for node in graph_def.node:
nodes_map[node.name] = node
for node in graph_def.node:
sort_tf_node(node, nodes_map, ordered_nodes_map)
sorted_graph = tf.GraphDef()
sorted_graph.node.extend([node for node in ordered_nodes_map.values()])
return sorted_graph
def sort_mace_node(node, nodes_map, ordered_nodes_map):
if node.name not in ordered_nodes_map:
for input_tensor_name in node.input:
input_node_name = input_tensor_name.split(':')[
0] if ':' in input_tensor_name else input_tensor_name
if input_node_name not in nodes_map or \
input_node_name in ordered_nodes_map:
continue
input_node = nodes_map[input_node_name]
sort_mace_node(input_node, nodes_map, ordered_nodes_map)
ordered_nodes_map[node.name] = node
def sort_mace_graph(graph_def, output_name):
nodes_map = {}
ordered_nodes_map = OrderedDict()
for node in graph_def.op:
nodes_map[node.name] = node
sort_mace_node(nodes_map[output_name], nodes_map, ordered_nodes_map)
del graph_def.op[:]
graph_def.op.extend([node for node in ordered_nodes_map.values()])
return graph_def
mace/python/tools/mace_engine_factory_codegen.py
浏览文件 @ 320b509c
...@@ -13,6 +13,7 @@ ...@@ -13,6 +13,7 @@
# limitations under the License. # limitations under the License.
import argparse import argparse
import os
from jinja2 import Environment, FileSystemLoader from jinja2 import Environment, FileSystemLoader
...@@ -20,11 +21,10 @@ from jinja2 import Environment, FileSystemLoader ...@@ -20,11 +21,10 @@ from jinja2 import Environment, FileSystemLoader
FLAGS = None FLAGS = None
def gen_mace_engine_factory(model_tags, template_dir, def gen_mace_engine_factory(model_tags, embed_model_data, output_dir):
embed_model_data, output_dir): cwd = os.path.dirname(__file__)
# Create the jinja2 environment.
j2_env = Environment( j2_env = Environment(
loader=FileSystemLoader(template_dir), trim_blocks=True) loader=FileSystemLoader(cwd), trim_blocks=True)
# generate mace_run BUILD file # generate mace_run BUILD file
template_name = 'mace_engine_factory.h.jinja2' template_name = 'mace_engine_factory.h.jinja2'
model_tags = list(model_tags) model_tags = list(model_tags)
......
mace/python/tools/opencl_binary_codegen.py
浏览文件 @ 320b509c
...@@ -29,8 +29,9 @@ def generate_opencl_code(binary_file_name, load_func_name, size_func_name, ...@@ -29,8 +29,9 @@ def generate_opencl_code(binary_file_name, load_func_name, size_func_name,
with open(binary_file_name, 'rb') as f: with open(binary_file_name, 'rb') as f:
binary_array = np.fromfile(f, dtype=np.uint8) binary_array = np.fromfile(f, dtype=np.uint8)
cwd = os.path.dirname(__file__)
env = jinja2.Environment( env = jinja2.Environment(
loader=jinja2.FileSystemLoader(sys.path[0])) loader=jinja2.FileSystemLoader(cwd))
content = env.get_template('file_binary.cc.jinja2').render( content = env.get_template('file_binary.cc.jinja2').render(
data=binary_array, data=binary_array,
data_size=len(binary_array), data_size=len(binary_array),
......
mace/python/tools/quantization/quantize_util_test.py 已删除 100644 → 0
浏览文件 @ 64419daa
import unittest
import numpy as np
import quantize_util
class TestQuantize(unittest.TestCase):
def test_quantize_dequantize(self):
test_input = np.random.rand(20, 30) * 5
quantized_data = quantize_util.quantize(test_input)
dequantized_output = quantize_util.dequantize(quantized_data)
np.testing.assert_array_almost_equal(test_input, dequantized_output, 2)
if __name__ == '__main__':
unittest.main()
mace/python/tools/tf_ops_stats.py 已删除 100644 → 0
浏览文件 @ 64419daa
# Copyright 2018 The MACE 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.
import operator
import functools
import argparse
import sys
import copy
import six
import tensorflow as tf
from tensorflow import gfile
from tensorflow.core.framework import graph_pb2
from tensorflow.core.framework import tensor_shape_pb2
# ./bazel-bin/mace/python/tools/tf_ops_stats --input model.pb
FLAGS = None
def hist_inc(hist, key):
if key in hist:
hist[key] += 1
else:
hist[key] = 1
def to_int_list(long_list):
int_list = []
for value in long_list:
int_list.append(int(value))
return int_list
def add_shape_info(input_graph_def, input_nodes, input_shapes):
inputs_replaced_graph = graph_pb2.GraphDef()
for node in input_graph_def.node:
if node.name in input_nodes or node.name + ':0' in input_nodes:
if node.name in input_nodes:
idx = input_nodes.index(node.name)
else:
idx = input_nodes.index(node.name + ':0')
input_shape = input_shapes[idx]
print(input_shape)
placeholder_node = copy.deepcopy(node)
placeholder_node.attr.clear()
placeholder_node.attr['shape'].shape.dim.extend([
tensor_shape_pb2.TensorShapeProto.Dim(size=i)
for i in input_shape
])
placeholder_node.attr['dtype'].CopyFrom(node.attr['dtype'])
inputs_replaced_graph.node.extend([placeholder_node])
else:
inputs_replaced_graph.node.extend([copy.deepcopy(node)])
return inputs_replaced_graph
def main(unused_args):
if not FLAGS.input or not gfile.Exists(FLAGS.input):
print('Input graph file ' + FLAGS.input + ' does not exist!')
return -1
input_graph_def = tf.GraphDef()
with gfile.Open(FLAGS.input, 'rb') as f:
data = f.read()
input_graph_def.ParseFromString(data)
input_nodes = [x for x in FLAGS.input_tensors.split(',')]
input_shapes = []
if FLAGS.input_shapes != "":
input_shape_strs = [x for x in FLAGS.input_shapes.split(':')]
for shape_str in input_shape_strs:
input_shapes.extend([[int(x) for x in shape_str.split(',')]])
input_graph_def = add_shape_info(
input_graph_def, input_nodes, input_shapes)
with tf.Session() as session:
with session.graph.as_default() as graph:
tf.import_graph_def(input_graph_def, name='')
stats = {}
ops = graph.get_operations()
# extract kernel size for conv_2d
tensor_shapes = {}
tensor_values = {}
print("=========================consts============================")
for op in ops:
if op.type == 'Const':
for output in op.outputs:
tensor_name = output.name
tensor = output.eval()
tensor_shape = list(tensor.shape)
tensor_shapes[tensor_name] = tensor_shape
print("Const %s: %s, %d" %
(tensor_name, tensor_shape,
functools.reduce(operator.mul, tensor_shape, 1)))
if len(tensor_shape) == 1 and tensor_shape[0] < 10:
tensor_values[tensor_name] = list(tensor)
print("=========================ops============================")
for op in ops:
if op.type in ['Conv2D']:
padding = op.get_attr('padding')
strides = to_int_list(op.get_attr('strides'))
data_format = op.get_attr('data_format')
ksize = 'Unknown'
input = op.inputs[1]
input_name = input.name
if input_name.endswith('read:0'):
ksize = input.shape.as_list()
elif input_name in tensor_shapes:
ksize = tensor_shapes[input_name]
print(
'%s(padding=%s, strides=%s, ksize=%s, format=%s) %s => %s'
% (op.type, padding, strides, ksize, data_format,
op.inputs[0].shape, op.outputs[0].shape))
key = '%s(padding=%s, strides=%s, ksize=%s, format=%s)' % (
op.type, padding, strides, ksize, data_format)
hist_inc(stats, key)
elif op.type in ['FusedResizeAndPadConv2D']:
padding = op.get_attr('padding')
strides = to_int_list(op.get_attr('strides'))
resize_align_corners = op.get_attr('resize_align_corners')
ksize = 'Unknown'
for input in op.inputs:
input_name = input.name
if input_name.endswith(
'weights:0') and input_name in tensor_shapes:
ksize = tensor_shapes[input_name]
break
key = '%s(padding=%s, strides=%s, ksize=%s, ' \
'resize_align_corners=%s)' % (op.type, padding, strides,
ksize, resize_align_corners)
hist_inc(stats, key)
elif op.type in ['ResizeBilinear']:
align_corners = op.get_attr('align_corners')
size = 'Unknown'
for input in op.inputs:
input_name = input.name
if input_name.endswith(
'size:0') and input_name in tensor_values:
size = tensor_values[input_name]
break
key = '%s(size=%s, align_corners=%s)' % (op.type, size,
align_corners)
print(key)
hist_inc(stats, key)
elif op.type in ['AvgPool', 'MaxPool']:
padding = op.get_attr('padding')
strides = to_int_list(op.get_attr('strides'))
ksize = to_int_list(op.get_attr('ksize'))
data_format = op.get_attr('data_format')
key = '%s(padding=%s, strides=%s, ksize=%s)' % (op.type,
padding,
strides, ksize)
hist_inc(stats, key)
elif op.type in ['SpaceToBatchND', 'BatchToSpaceND']:
block_shape = 'Unknown'
for input in op.inputs:
input_name = input.name
if input_name.endswith(
'block_shape:0') and input_name in tensor_values:
block_shape = tensor_values[input_name]
break
paddings = 'Unknown'
for input in op.inputs:
input_name = input.name
if input_name.endswith(
'paddings:0') and input_name in tensor_values:
paddings = tensor_values[input_name]
break
crops = 'Unknown'
for input in op.inputs:
input_name = input.name
if input_name.endswith(
'crops:0') and input_name in tensor_values:
paddings = tensor_values[input_name]
break
if op.type == 'SpaceToBatchND':
key = '%s(block_shape=%s, paddings=%s)' % (op.type,
block_shape,
paddings)
else:
key = '%s(block_shape=%s, crops=%s)' % (op.type,
block_shape, crops)
print(key)
hist_inc(stats, key)
elif op.type == 'Pad':
paddings = 'Unknown'
for input in op.inputs:
input_name = input.name
if input_name.endswith(
'paddings:0') and input_name in tensor_values:
paddings = tensor_values[input_name]
break
key = '%s(paddings=%s)' % (op.type, paddings)
hist_inc(stats, key)
else:
hist_inc(stats, op.type)
print("=========================stats============================")
for key, value in sorted(six.iteritems(stats)):
print('%s: %d' % (key, value))
def parse_args():
"""Parses command line arguments."""
parser = argparse.ArgumentParser()
parser.add_argument(
'--input',
type=str,
default='',
help='TensorFlow \'GraphDef\' file to load.')
parser.add_argument(
'--input_tensors',
type=str,
default='',
help='input tensor names split by comma.')
parser.add_argument(
'--input_shapes',
type=str,
default='',
help='input tensor shapes split by colon and comma.')
return parser.parse_known_args()
if __name__ == '__main__':
FLAGS, unparsed = parse_args()
main(unused_args=[sys.argv[0]] + unparsed)
mace/python/tools/visualization/BUILD.bazel 已删除 100644 → 0
浏览文件 @ 64419daa
py_library(
name = "visualization_lib",
srcs = [
"visualize_model.py",
],
data = [
"index.html",
],
srcs_version = "PY2AND3",
visibility = ["//visibility:public"],
deps = [
"//mace/proto:mace_py",
],
)
tools/converter.py
浏览文件 @ 320b509c
...@@ -12,20 +12,25 @@ ...@@ -12,20 +12,25 @@
# See the License for the specific language governing permissions and # See the License for the specific language governing permissions and
# limitations under the License. # limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import argparse import argparse
import glob import glob
import sh import sh
import sys import sys
import time import time
import yaml import yaml
from enum import Enum
import six
import sh_commands import sh_commands
from enum import Enum
sys.path.insert(0, "tools/python") # noqa
from common import * from common import *
from device import DeviceWrapper, DeviceManager from device import DeviceWrapper, DeviceManager
from utils import config_parser
import convert
import encrypt
################################ ################################
# set environment # set environment
...@@ -711,111 +716,6 @@ def print_configuration(configs): ...@@ -711,111 +716,6 @@ def print_configuration(configs):
MaceLogger.summary(StringFormatter.table(header, data, title)) MaceLogger.summary(StringFormatter.table(header, data, title))
def convert_model(configs, cl_mem_type):
# Remove previous output dirs
library_name = configs[YAMLKeyword.library_name]
if not os.path.exists(BUILD_OUTPUT_DIR):
os.makedirs(BUILD_OUTPUT_DIR)
elif os.path.exists(os.path.join(BUILD_OUTPUT_DIR, library_name)):
sh.rm("-rf", os.path.join(BUILD_OUTPUT_DIR, library_name))
os.makedirs(os.path.join(BUILD_OUTPUT_DIR, library_name))
if not os.path.exists(BUILD_DOWNLOADS_DIR):
os.makedirs(BUILD_DOWNLOADS_DIR)
model_output_dir = \
'%s/%s/%s' % (BUILD_OUTPUT_DIR, library_name, MODEL_OUTPUT_DIR_NAME)
model_header_dir = \
'%s/%s/%s' % (BUILD_OUTPUT_DIR, library_name, MODEL_HEADER_DIR_PATH)
# clear output dir
if os.path.exists(model_output_dir):
sh.rm("-rf", model_output_dir)
os.makedirs(model_output_dir)
if os.path.exists(model_header_dir):
sh.rm("-rf", model_header_dir)
embed_model_data = \
configs[YAMLKeyword.model_data_format] == ModelFormat.code
if os.path.exists(MODEL_CODEGEN_DIR):
sh.rm("-rf", MODEL_CODEGEN_DIR)
if os.path.exists(ENGINE_CODEGEN_DIR):
sh.rm("-rf", ENGINE_CODEGEN_DIR)
if configs[YAMLKeyword.model_graph_format] == ModelFormat.code:
os.makedirs(model_header_dir)
sh_commands.gen_mace_engine_factory_source(
configs[YAMLKeyword.models].keys(),
embed_model_data)
sh.cp("-f", glob.glob("mace/codegen/engine/*.h"),
model_header_dir)
for model_name in configs[YAMLKeyword.models]:
MaceLogger.header(
StringFormatter.block("Convert %s model" % model_name))
model_config = configs[YAMLKeyword.models][model_name]
runtime = model_config[YAMLKeyword.runtime]
if cl_mem_type:
model_config[YAMLKeyword.cl_mem_type] = cl_mem_type
else:
model_config[YAMLKeyword.cl_mem_type] = "image"
data_type = model_config[YAMLKeyword.data_type]
# TODO(liuqi): support multiple subgraphs
subgraphs = model_config[YAMLKeyword.subgraphs]
model_codegen_dir = "%s/%s" % (MODEL_CODEGEN_DIR, model_name)
sh_commands.gen_model_code(
model_codegen_dir,
model_config[YAMLKeyword.platform],
model_config[YAMLKeyword.model_file_path],
model_config[YAMLKeyword.weight_file_path],
model_config[YAMLKeyword.model_sha256_checksum],
model_config[YAMLKeyword.weight_sha256_checksum],
",".join(subgraphs[0][YAMLKeyword.input_tensors]),
",".join(subgraphs[0][YAMLKeyword.input_data_types]),
",".join(subgraphs[0][YAMLKeyword.input_data_formats]),
",".join(subgraphs[0][YAMLKeyword.output_tensors]),
",".join(subgraphs[0][YAMLKeyword.output_data_types]),
",".join(subgraphs[0][YAMLKeyword.output_data_formats]),
",".join(subgraphs[0][YAMLKeyword.check_tensors]),
runtime,
model_name,
":".join(subgraphs[0][YAMLKeyword.input_shapes]),
":".join(subgraphs[0][YAMLKeyword.input_ranges]),
":".join(subgraphs[0][YAMLKeyword.output_shapes]),
":".join(subgraphs[0][YAMLKeyword.check_shapes]),
model_config[YAMLKeyword.nnlib_graph_mode],
embed_model_data,
model_config[YAMLKeyword.winograd],
model_config[YAMLKeyword.quantize],
model_config[YAMLKeyword.quantize_large_weights],
model_config[YAMLKeyword.quantize_range_file],
model_config[YAMLKeyword.change_concat_ranges],
model_config[YAMLKeyword.obfuscate],
configs[YAMLKeyword.model_graph_format],
data_type,
model_config[YAMLKeyword.cl_mem_type],
",".join(model_config.get(YAMLKeyword.graph_optimize_options, [])))
if configs[YAMLKeyword.model_graph_format] == ModelFormat.file:
sh.mv("-f",
'%s/%s.pb' % (model_codegen_dir, model_name),
model_output_dir)
sh.mv("-f",
'%s/%s.data' % (model_codegen_dir, model_name),
model_output_dir)
else:
if not embed_model_data:
sh.mv("-f",
'%s/%s.data' % (model_codegen_dir, model_name),
model_output_dir)
sh.cp("-f", glob.glob("mace/codegen/models/*/*.h"),
model_header_dir)
MaceLogger.summary(
StringFormatter.block("Model %s converted" % model_name))
def build_model_lib(configs, address_sanitizer, debug_mode): def build_model_lib(configs, address_sanitizer, debug_mode):
MaceLogger.header(StringFormatter.block("Building model library")) MaceLogger.header(StringFormatter.block("Building model library"))
...@@ -863,13 +763,85 @@ def print_library_summary(configs): ...@@ -863,13 +763,85 @@ def print_library_summary(configs):
def convert_func(flags): def convert_func(flags):
configs = format_model_config(flags) configs = config_parser.parse(flags.config)
library_name = configs[YAMLKeyword.library_name]
if not os.path.exists(BUILD_OUTPUT_DIR):
os.makedirs(BUILD_OUTPUT_DIR)
elif os.path.exists(os.path.join(BUILD_OUTPUT_DIR, library_name)):
sh.rm("-rf", os.path.join(BUILD_OUTPUT_DIR, library_name))
os.makedirs(os.path.join(BUILD_OUTPUT_DIR, library_name))
if not os.path.exists(BUILD_DOWNLOADS_DIR):
os.makedirs(BUILD_DOWNLOADS_DIR)
print_configuration(configs) model_output_dir = \
'%s/%s/%s' % (BUILD_OUTPUT_DIR, library_name, MODEL_OUTPUT_DIR_NAME)
model_header_dir = \
'%s/%s/%s' % (BUILD_OUTPUT_DIR, library_name, MODEL_HEADER_DIR_PATH)
# clear output dir
if os.path.exists(model_output_dir):
sh.rm("-rf", model_output_dir)
os.makedirs(model_output_dir)
if os.path.exists(model_header_dir):
sh.rm("-rf", model_header_dir)
convert_model(configs, flags.cl_mem_type) if os.path.exists(MODEL_CODEGEN_DIR):
sh.rm("-rf", MODEL_CODEGEN_DIR)
if os.path.exists(ENGINE_CODEGEN_DIR):
sh.rm("-rf", ENGINE_CODEGEN_DIR)
if configs[YAMLKeyword.model_graph_format] == ModelFormat.code: if flags.model_data_format:
model_data_format = flags.model_data_format
else:
model_data_format = configs.get(YAMLKeyword.model_data_format,
"file")
embed_model_data = model_data_format == ModelFormat.code
if flags.model_graph_format:
model_graph_format = flags.model_graph_format
else:
model_graph_format = configs.get(YAMLKeyword.model_graph_format,
"file")
if model_graph_format == ModelFormat.code:
os.makedirs(model_header_dir)
sh_commands.gen_mace_engine_factory_source(
configs[YAMLKeyword.models].keys(),
embed_model_data)
sh.cp("-f", glob.glob("mace/codegen/engine/*.h"),
model_header_dir)
convert.convert(configs, MODEL_CODEGEN_DIR)
for model_name, model_config in configs[YAMLKeyword.models].items():
model_codegen_dir = "%s/%s" % (MODEL_CODEGEN_DIR, model_name)
encrypt.encrypt(model_name,
"%s/%s.pb" % (model_codegen_dir, model_name),
"%s/%s.data" % (model_codegen_dir, model_name),
model_config[YAMLKeyword.runtime],
model_codegen_dir,
bool(model_config[YAMLKeyword.obfuscate]))
if model_graph_format == ModelFormat.file:
sh.mv("-f",
'%s/file/%s.pb' % (model_codegen_dir, model_name),
model_output_dir)
sh.mv("-f",
'%s/file/%s.data' % (model_codegen_dir, model_name),
model_output_dir)
sh.rm("-rf", '%s/code' % model_codegen_dir)
else:
if not embed_model_data:
sh.mv("-f",
'%s/file/%s.data' % (model_codegen_dir, model_name),
model_output_dir)
sh.rm('%s/code/tensor_data.cc' % model_codegen_dir)
sh.cp("-f", glob.glob("mace/codegen/models/*/code/*.h"),
model_header_dir)
MaceLogger.summary(
StringFormatter.block("Model %s converted" % model_name))
if model_graph_format == ModelFormat.code:
build_model_lib(configs, flags.address_sanitizer, flags.debug_mode) build_model_lib(configs, flags.address_sanitizer, flags.debug_mode)
print_library_summary(configs) print_library_summary(configs)
...@@ -1047,11 +1019,6 @@ def parse_args(): ...@@ -1047,11 +1019,6 @@ def parse_args():
'convert', 'convert',
parents=[all_type_parent_parser, convert_run_parent_parser], parents=[all_type_parent_parser, convert_run_parent_parser],
help='convert to mace model (file or code)') help='convert to mace model (file or code)')
convert.add_argument(
"--cl_mem_type",
type=str,
default=None,
help="Which type of OpenCL memory type to use [image | buffer].")
convert.set_defaults(func=convert_func) convert.set_defaults(func=convert_func)
run = subparsers.add_parser( run = subparsers.add_parser(
......
tools/python/convert.py 0 → 100644
浏览文件 @ 320b509c
# Copyright 2019 The MACE 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.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import argparse
import os
import sys
import numpy as np
from utils import config_parser
from utils import util
from utils.util import mace_check
from py_proto import mace_pb2
from transform import base_converter as cvt
from transform import transformer
from visualize import visualize_model
device_type_map = {'cpu': cvt.DeviceType.CPU.value,
'gpu': cvt.DeviceType.GPU.value,
'dsp': cvt.DeviceType.HEXAGON.value,
'hta': cvt.DeviceType.HTA.value,
'apu': cvt.DeviceType.APU.value,
'cpu+gpu': cvt.DeviceType.CPU.value}
data_format_map = {
'NONE': cvt.DataFormat.NONE,
'NHWC': cvt.DataFormat.NHWC,
'NCHW': cvt.DataFormat.NCHW,
'OIHW': cvt.DataFormat.OIHW,
}
data_type_map = {
'float32': mace_pb2.DT_FLOAT,
'int32': mace_pb2.DT_INT32,
}
def parse_data_type(data_type, quantize):
if quantize or data_type == 'fp32_fp32':
return mace_pb2.DT_FLOAT
else:
return mace_pb2.DT_HALF
def split_shape(shape):
if shape.strip() == "":
return []
else:
return shape.split(',')
def parse_int_array_from_str(ints_str):
return [int(i) for i in split_shape(ints_str)]
def parse_float_array_from_str(floats_str):
return [float(i) for i in floats_str.split(',')]
def transpose_shape(shape, dst_order):
t_shape = [0] * len(shape)
for i in range(len(shape)):
t_shape[i] = shape[dst_order[i]]
return t_shape
def to_list(x):
if isinstance(x, list):
return x
else:
return [x]
def separate_params(mace_model):
tensors = mace_model.tensors
params = mace_pb2.NetDef()
params.tensors.extend(tensors)
model = mace_model
del model.tensors[:]
return model, params
def convert(conf, output):
if not os.path.exists(output):
os.mkdir(output)
for model_name, model_conf in conf["models"].items():
model_output = output + "/" + model_name
if not os.path.exists(model_output):
os.mkdir(model_output)
subgraph = model_conf["subgraphs"][0]
del model_conf["subgraphs"]
model_conf.update(subgraph)
model_file = util.download_or_get_file(model_conf["model_file_path"],
model_conf[
"model_sha256_checksum"],
model_output)
model_conf["model_file_path"] = model_file
if "weight_file_path" in model_conf:
weight_file = util.download_or_get_file(
model_conf["weight_file_path"],
model_conf["weight_sha256_checksum"], model_output)
model_conf["weight_file_path"] = weight_file
mace_model = convert_model(model_conf)
try:
visualizer = visualize_model.ModelVisualizer(model_name,
mace_model,
model_output)
visualizer.save_html()
except: # noqa
print("Failed to visualize model:", sys.exc_info()[0])
model, params = merge_params(mace_model)
output_model_file = model_output + "/" + model_name + ".pb"
output_params_file = model_output + "/" + model_name + ".data"
with open(output_model_file, "wb") as f:
f.write(model.SerializeToString())
with open(output_params_file, "wb") as f:
f.write(bytearray(params))
with open(output_model_file + "_txt", "w") as f:
f.write(str(model))
def convert_model(conf):
print(conf)
platform = conf["platform"]
mace_check(platform in ['tensorflow', 'caffe', 'onnx'],
"platform not supported")
runtime = conf["runtime"]
mace_check(
runtime in ['cpu', 'gpu', 'dsp', 'hta', 'apu', 'cpu+gpu'],
"runtime not supported")
option = cvt.ConverterOption()
if "graph_optimize_options" in conf:
option.transformer_option = conf["graph_optimize_options"].split(',')
option.winograd = conf.get("winograd", 0)
option.quantize = bool(conf.get("quantize", 0))
option.quantize_large_weights = bool(conf.get("quantize_large_weights", 0))
option.quantize_range_file = conf.get("quantize_range_file", "")
option.change_concat_ranges = bool(conf.get("change_concat_ranges", 0))
option.cl_mem_type = conf.get("cl_mem_type", "image")
option.device = device_type_map[conf.get("runtime", "cpu")]
option.data_type = parse_data_type(conf.get("data_type", "fp16_fp32"),
option.quantize)
input_tensors = to_list(conf["input_tensors"])
input_shapes = [parse_int_array_from_str(shape) for shape in
to_list(conf["input_shapes"])]
mace_check(len(input_tensors) == len(input_shapes),
"input node count and shape count do not match")
input_count = len(input_tensors)
input_data_types = [data_type_map[dt] for dt in
to_list(conf.get("input_data_types",
["float32"] * input_count))]
input_data_formats = [data_format_map[df] for df in
to_list(conf.get("input_data_formats",
["NHWC"] * input_count))]
input_ranges = [parse_float_array_from_str(r) for r in
to_list(conf.get("input_ranges",
["-1.0,1.0"] * input_count))]
for i in range(len(input_tensors)):
input_node = cvt.NodeInfo()
input_node.name = input_tensors[i]
input_node.shape = input_shapes[i]
input_node.data_type = input_data_types[i]
input_node.data_format = input_data_formats[i]
if (input_node.data_format == cvt.DataFormat.NCHW and len(
input_node.shape) == 4):
input_node.shape = transpose_shape(input_node.shape, [0, 2, 3, 1])
input_node.data_format = cvt.DataFormat.NHWC
input_node.range = input_ranges[i]
option.add_input_node(input_node)
output_tensors = to_list(conf["output_tensors"])
output_shapes = [parse_int_array_from_str(shape) for shape in
to_list(conf["output_shapes"])]
mace_check(len(output_tensors) == len(output_shapes),
"output node count and shape count do not match")
output_count = len(output_tensors)
output_data_types = [data_type_map[dt] for dt in
to_list(conf.get("output_data_types",
["float32"] * output_count))]
output_data_formats = [data_format_map[df] for df in
to_list(conf.get("output_data_formats",
["NHWC"] * output_count))]
for i in range(len(output_tensors)):
output_node = cvt.NodeInfo()
output_node.name = output_tensors[i]
output_node.shape = output_shapes[i]
output_node.data_type = output_data_types[i]
output_node.data_format = output_data_formats[i]
if output_node.data_format == cvt.DataFormat.NCHW and len(
output_node.shape) == 4:
output_node.shape = transpose_shape(output_node.shape,
[0, 2, 3, 1])
output_node.data_format = cvt.DataFormat.NHWC
option.add_output_node(output_node)
if "check_node" in conf:
check_node_names = to_list(conf["check_node"])
check_node_shapes = [parse_int_array_from_str(shape) for shape in
to_list(conf["check_shape"])]
mace_check(len(check_node_names) == len(check_node_shapes),
"check node count and shape count do not match.")
for i in range(len(check_node_names)):
check_node = cvt.NodeInfo()
check_node.name = check_node_names[i]
check_node.shape = check_node_shapes[i]
option.add_check_node(check_node)
else:
option.check_nodes = option.output_nodes
option.build()
print("Transform model to one that can better run on device")
if platform == 'tensorflow':
from transform import tensorflow_converter
converter = tensorflow_converter.TensorflowConverter(
option, conf["model_file_path"])
elif platform == 'caffe':
from transform import caffe_converter
converter = caffe_converter.CaffeConverter(option,
conf["model_file_path"],
conf["weight_file_path"])
elif platform == 'onnx':
from transform import onnx_converter
converter = onnx_converter.OnnxConverter(option,
conf["model_file_path"])
else:
mace_check(False, "Mace do not support platorm %s yet." % platform)
output_graph_def = converter.run()
mace_transformer = transformer.Transformer(
option, output_graph_def)
output_graph_def, quantize_activation_info = mace_transformer.run()
if option.device in [cvt.DeviceType.HEXAGON.value,
cvt.DeviceType.HTA.value]:
from transform import hexagon_converter
converter = hexagon_converter.HexagonConverter(
option, output_graph_def, quantize_activation_info)
output_graph_def = converter.run()
elif runtime == 'apu':
mace_check(platform == "tensorflow",
"apu only support model from tensorflow")
from transform import apu_converter
converter = apu_converter.ApuConverter(
option, output_graph_def, quantize_activation_info)
output_graph_def = converter.run()
return output_graph_def
def merge_params(net_def):
def tensor_to_bytes(tensor):
if tensor.data_type == mace_pb2.DT_HALF:
data = bytearray(
np.array(tensor.float_data).astype(np.float16).tobytes())
elif tensor.data_type == mace_pb2.DT_FLOAT:
data = bytearray(
np.array(tensor.float_data).astype(np.float32).tobytes())
elif tensor.data_type == mace_pb2.DT_INT32:
data = bytearray(
np.array(tensor.int32_data).astype(np.int32).tobytes())
elif tensor.data_type == mace_pb2.DT_UINT8:
data = bytearray(
np.array(tensor.int32_data).astype(np.uint8).tolist())
elif tensor.data_type == mace_pb2.DT_FLOAT16:
data = bytearray(
np.array(tensor.float_data).astype(np.float16).tobytes())
else:
raise Exception('Tensor data type %s not supported' %
tensor.data_type)
return data
model_data = []
offset = 0
for tensor in net_def.tensors:
raw_data = tensor_to_bytes(tensor)
if tensor.data_type != mace_pb2.DT_UINT8 and offset % 4 != 0:
padding = 4 - offset % 4
model_data.extend(bytearray([0] * padding))
offset += padding
tensor.offset = offset
model_data.extend(raw_data)
offset += len(raw_data)
for tensor in net_def.tensors:
if tensor.data_type == mace_pb2.DT_FLOAT \
or tensor.data_type == mace_pb2.DT_HALF \
or tensor.data_type == mace_pb2.DT_FLOAT16:
del tensor.float_data[:]
elif tensor.data_type == mace_pb2.DT_INT32:
del tensor.int32_data[:]
elif tensor.data_type == mace_pb2.DT_UINT8:
del tensor.int32_data[:]
return net_def, model_data
def parse_args():
parser = argparse.ArgumentParser()
parser.add_argument(
'--config',
type=str,
default="",
required=True,
help="the path of model yaml configuration file.")
parser.add_argument(
'--output',
type=str,
default=".",
help="output dir")
flgs, _ = parser.parse_known_args()
return flgs
if __name__ == '__main__':
flags = parse_args()
conf = config_parser.parse(flags.config)
convert(conf, flags.output)
mace/python/tools/model_saver.py → tools/python/encrypt.py
浏览文件 @ 320b509c
# Copyright 2018 The MACE Authors. All Rights Reserved. # Copyright 2019 The MACE Authors. All Rights Reserved.
# #
# Licensed under the Apache License, Version 2.0 (the "License"); # Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License. # you may not use this file except in compliance with the License.
...@@ -12,31 +12,28 @@ ...@@ -12,31 +12,28 @@
# See the License for the specific language governing permissions and # See the License for the specific language governing permissions and
# limitations under the License. # limitations under the License.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import argparse
import datetime import datetime
import os import os
import six
import uuid
import numpy as np
import hashlib import hashlib
from enum import Enum
from mace.proto import mace_pb2
from mace.python.tools.converter_tool import base_converter as cvt
from mace.python.tools.convert_util import mace_check
from jinja2 import Environment, FileSystemLoader from jinja2 import Environment, FileSystemLoader
from py_proto import mace_pb2
from utils import util
from transform import base_converter as cvt
from utils.util import mace_check
from utils.config_parser import CPP_KEYWORDS
GENERATED_NAME = set() GENERATED_NAME = set()
class ModelFormat(object):
file = "file"
code = "code"
def generate_obfuscated_name(namespace, name): def generate_obfuscated_name(namespace, name):
md5 = hashlib.md5() md5 = hashlib.md5()
md5.update(six.b(namespace)) md5.update(namespace)
md5.update(six.b(name)) md5.update(name)
md5_digest = md5.hexdigest() md5_digest = md5.hexdigest()
name = md5_digest[:8] name = md5_digest[:8]
...@@ -76,19 +73,23 @@ def generate_in_out_map(ops, tensor_map): ...@@ -76,19 +73,23 @@ def generate_in_out_map(ops, tensor_map):
return in_out_map return in_out_map
def obfuscate_name(option, net_def): def stringfy(value):
return ', '.join('"{0}"'.format(w) for w in value)
def obfuscate_name(model):
input_nodes = set() input_nodes = set()
for name in option.input_nodes: for input_node in model.input_info:
input_nodes.add(name) input_nodes.add(input_node.name)
output_nodes = set() output_nodes = set()
for name in option.output_nodes: for output_node in model.output_info:
output_nodes.add(name) output_nodes.add(output_node.name)
tensor_map = generate_tensor_map(net_def.tensors) tensor_map = generate_tensor_map(model.tensors)
in_out_map = generate_in_out_map(net_def.op, tensor_map) in_out_map = generate_in_out_map(model.op, tensor_map)
for t in net_def.tensors: for t in model.tensors:
if t.name not in input_nodes and t.name not in output_nodes: if t.name not in input_nodes and t.name not in output_nodes:
t.name = tensor_map[t.name] t.name = tensor_map[t.name]
for op in net_def.op: for op in model.op:
for i in range(len(op.input)): for i in range(len(op.input)):
if op.input[i] not in input_nodes: if op.input[i] not in input_nodes:
op.input[i] = in_out_map[op.input[i]] op.input[i] = in_out_map[op.input[i]]
...@@ -97,201 +98,140 @@ def obfuscate_name(option, net_def): ...@@ -97,201 +98,140 @@ def obfuscate_name(option, net_def):
op.output[i] = in_out_map[op.output[i]] op.output[i] = in_out_map[op.output[i]]
def stringfy(value): def save_model_to_code(namespace, model, params, model_checksum,
return ', '.join('"{0}"'.format(w) for w in value) params_checksum, device, output):
if not os.path.exists(output):
os.mkdir(output)
class TensorInfo: cwd = os.path.dirname(__file__)
def __init__(self, id, tensor):
self.id = id
self.data_type = tensor.data_type
if tensor.data_type == mace_pb2.DT_HALF:
self.data = bytearray(
np.array(tensor.float_data).astype(np.float16).tobytes())
elif tensor.data_type == mace_pb2.DT_FLOAT:
self.data = bytearray(
np.array(tensor.float_data).astype(np.float32).tobytes())
elif tensor.data_type == mace_pb2.DT_INT32:
self.data = bytearray(
np.array(tensor.int32_data).astype(np.int32).tobytes())
elif tensor.data_type == mace_pb2.DT_UINT8:
self.data = bytearray(
np.array(tensor.int32_data).astype(np.uint8).tolist())
elif tensor.data_type == mace_pb2.DT_FLOAT16:
self.data = bytearray(
np.array(tensor.float_data).astype(np.float16).tobytes())
else:
raise Exception('Tensor data type %s not supported' %
tensor.data_type)
def update_tensor_infos(net_def, data_type):
offset = 0
counter = 0
tensor_infos = []
for tensor in net_def.tensors:
if tensor.data_type == mace_pb2.DT_FLOAT:
tensor.data_type = data_type
# Add offset and data_size
tensor_info = TensorInfo(counter, tensor)
tensor_infos.append(tensor_info)
# align
if tensor_info.data_type != mace_pb2.DT_UINT8 and offset % 4 != 0:
padding = 4 - offset % 4
offset += padding
if tensor.data_type == mace_pb2.DT_FLOAT \
or tensor.data_type == mace_pb2.DT_HALF \
or tensor.data_type == mace_pb2.DT_FLOAT16:
tensor.data_size = len(tensor.float_data)
elif tensor.data_type == mace_pb2.DT_INT32:
tensor.data_size = len(tensor.int32_data)
elif tensor.data_type == mace_pb2.DT_UINT8:
tensor.data_size = len(tensor.int32_data)
tensor.offset = offset
offset += len(tensor_info.data)
counter += 1
def extract_model_data(net_def):
model_data = []
offset = 0
counter = 0
for tensor in net_def.tensors:
tensor_info = TensorInfo(counter, tensor)
# align
mace_check(offset <= tensor.offset,
"Current offset should be <= tensor.offset")
if offset < tensor.offset:
model_data.extend(bytearray([0] * (tensor.offset - offset)))
offset = tensor.offset
model_data.extend(tensor_info.data)
offset += len(tensor_info.data)
counter += 1
return model_data
def save_model_data(net_def, model_tag, output_dir):
model_data = extract_model_data(net_def)
# generate tensor data
with open(output_dir + model_tag + '.data', "wb") as f:
f.write(bytearray(model_data))
def save_model_to_proto(net_def, model_tag, output_dir):
for tensor in net_def.tensors:
if tensor.data_type == mace_pb2.DT_FLOAT \
or tensor.data_type == mace_pb2.DT_HALF \
or tensor.data_type == mace_pb2.DT_FLOAT16:
del tensor.float_data[:]
elif tensor.data_type == mace_pb2.DT_INT32:
del tensor.int32_data[:]
elif tensor.data_type == mace_pb2.DT_UINT8:
del tensor.int32_data[:]
proto_file_path = output_dir + model_tag + '.pb'
with open(proto_file_path, "wb") as f:
f.write(net_def.SerializeToString())
with open(proto_file_path + '_txt', "w") as f:
f.write(str(net_def))
return proto_file_path
def save_model_to_code(net_def, model_tag, device,
template_dir, output_dir, embed_model_data,
model_checksum, weight_checksum,
obfuscate, winograd_conv):
# Create the jinja2 environment.
j2_env = Environment( j2_env = Environment(
loader=FileSystemLoader(template_dir), trim_blocks=True) loader=FileSystemLoader(cwd + "/template"), trim_blocks=True)
j2_env.filters['stringfy'] = stringfy j2_env.filters["stringfy"] = stringfy
# generate tensor source files
template_name = 'tensor_source.jinja2'
template_name = "tensor_source.jinja2"
counter = 0 counter = 0
for tensor in net_def.tensors: for tensor in model.tensors:
tensor_info = TensorInfo(counter, tensor)
# convert tensor # convert tensor
source = j2_env.get_template(template_name).render( source = j2_env.get_template(template_name).render(
tensor_info=tensor_info,
tensor=tensor, tensor=tensor,
tag=model_tag, tensor_id=counter,
tag=namespace,
) )
with open(output_dir + 'tensor' + str(counter) + '.cc', "w") as f: with open(output + "/tensor" + str(counter) + ".cc", "w") as f:
f.write(source) f.write(source)
counter += 1 counter += 1
# generate tensor data template_name = "tensor_data.jinja2"
if embed_model_data: source = j2_env.get_template(template_name).render(
model_data = extract_model_data(net_def) tag=namespace,
template_name = 'tensor_data.jinja2' model_data_size=len(params),
source = j2_env.get_template(template_name).render( model_data=params)
tag=model_tag, with open(output + "/tensor_data.cc", "w") as f:
model_data_size=len(model_data), f.write(source)
model_data=model_data)
with open(output_dir + 'tensor_data' + '.cc', "w") as f:
f.write(source)
# generate op source files template_name = "operator.jinja2"
template_name = 'operator.jinja2'
counter = 0 counter = 0
op_size = len(net_def.op) op_size = len(model.op)
try:
device = cvt.DeviceType[device.upper()]
except: # noqa
if device.upper == "DSP":
device = cvt.DeviceType.HEXAGON
else:
device = cvt.DeviceType.CPU
for start in range(0, op_size, 10): for start in range(0, op_size, 10):
source = j2_env.get_template(template_name).render( source = j2_env.get_template(template_name).render(
start=start, start=start,
end=min(start + 10, op_size), end=min(start + 10, op_size),
net=net_def, net=model,
tag=model_tag, tag=namespace,
device=device, device=device,
) )
with open(output_dir + 'op' + str(counter) + '.cc', "w") as f: with open(output + "/op" + str(counter) + ".cc", "w") as f:
f.write(source) f.write(source)
counter += 1 counter += 1
# generate model source files # generate model source files
build_time = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S') build_time = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
template_name = 'model.jinja2' template_name = "model.jinja2"
checksum = model_checksum checksum = "{},{}".format(model_checksum, params_checksum)
if weight_checksum is not None:
checksum = "{},{}".format(model_checksum, weight_checksum)
source = j2_env.get_template(template_name).render( source = j2_env.get_template(template_name).render(
net=net_def, net=model,
tag=model_tag, tag=namespace,
obfuscate=obfuscate,
embed_model_data=embed_model_data,
winograd_conv=winograd_conv,
checksum=checksum, checksum=checksum,
build_time=build_time) build_time=build_time)
with open(output_dir + 'model.cc', "w") as f: with open(output + "/model.cc", "w") as f:
f.write(source) f.write(source)
# generate model header file
template_name = 'model_header.jinja2' template_name = 'model_header.jinja2'
source = j2_env.get_template(template_name).render(tag=model_tag, ) source = j2_env.get_template(template_name).render(tag=namespace, )
with open(output_dir + model_tag + '.h', "w") as f: with open(output + "/" + namespace + '.h', "w") as f:
f.write(source) f.write(source)
def save_model(option, net_def, model_checksum, weight_checksum, template_dir, def save_model_to_file(model_name, model, params, output):
obfuscate, model_tag, output_dir, embed_model_data, if not os.path.exists(output):
winograd_conv, model_graph_format): os.mkdir(output)
if obfuscate: with open(output + "/" + model_name + ".pb", "wb") as f:
obfuscate_name(option, net_def) f.write(model.SerializeToString())
with open(output + "/" + model_name + ".data", "wb") as f:
output_dir = output_dir + '/' f.write(params)
net_def.data_type = option.data_type
# update tensor type
update_tensor_infos(net_def, option.data_type) def encrypt(model_name, model_file, params_file, device, output,
is_obfuscate=False):
if model_graph_format == ModelFormat.file or not embed_model_data: model_checksum = util.file_checksum(model_file)
save_model_data(net_def, model_tag, output_dir) params_checksum = util.file_checksum(params_file)
if model_graph_format == ModelFormat.file: with open(model_file, "rb") as model_file:
save_model_to_proto(net_def, model_tag, output_dir) with open(params_file, "rb") as params_file:
else: model = mace_pb2.NetDef()
save_model_to_code(net_def, model_tag, option.device, model.ParseFromString(model_file.read())
template_dir, output_dir, embed_model_data, params = bytearray(params_file.read())
model_checksum, weight_checksum,
obfuscate, winograd_conv) if is_obfuscate:
obfuscate_name(model)
save_model_to_file(model_name, model, params, output + "/file/")
save_model_to_code(model_name, model, params, model_checksum,
params_checksum, device, output + "/code/")
def parse_args():
parser = argparse.ArgumentParser()
parser.add_argument(
'--model_name',
type=str,
help="the namespace of gernerated code")
parser.add_argument(
'--model_file',
type=str,
help="model file")
parser.add_argument(
'--params_file',
type=str,
help="params file")
parser.add_argument(
'--device',
type=str,
default='cpu',
help="cpu/gpu/hexagon/hta/apu")
parser.add_argument(
'--output',
type=str,
default=".",
help="output dir")
parser.add_argument(
"--obfuscate",
action="store_true",
help="obfuscate model names")
flgs, _ = parser.parse_known_args()
mace_check(flags.model_name not in CPP_KEYWORDS, "model name cannot be cpp"
"keywords")
return flgs
if __name__ == '__main__':
flags = parse_args()
encrypt(flags.model_name, flags.model_file, flags.params_file,
flags.device, flags.output, flags.obfuscate)
tools/experimental/utils/util.py → tools/python/py_proto/__init__.py
浏览文件 @ 320b509c
...@@ -15,3 +15,15 @@ ...@@ -15,3 +15,15 @@
from __future__ import absolute_import from __future__ import absolute_import
from __future__ import division from __future__ import division
from __future__ import print_function from __future__ import print_function
import os
from utils import device
cwd = os.path.dirname(__file__)
# TODO: Remove bazel deps
device.execute("bazel build //mace/proto:mace_py")
device.execute("cp -f bazel-genfiles/mace/proto/mace_pb2.py %s" % cwd)
device.execute("bazel build //third_party/caffe:caffe_py")
device.execute("cp -f bazel-genfiles/third_party/caffe/caffe_pb2.py %s" % cwd)
mace/python/tools/quantization/quantize_stat.py → tools/python/quantize/quantize_stat.py
浏览文件 @ 320b509c
# Copyright 2019 The MACE 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.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import argparse import argparse
import numpy as np import numpy as np
......
mace/python/tools/quantization/quantize_util.py → tools/python/quantize/quantize_util.py
浏览文件 @ 320b509c
# Copyright 2019 The MACE 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.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import numpy as np import numpy as np
import math import math
from mace.python.tools.converter_tool.base_converter import DeviceType from transform.base_converter import DeviceType
class QuantizedData(object): class QuantizedData(object):
...@@ -126,7 +144,8 @@ def cal_multiplier_and_shift(scale): ...@@ -126,7 +144,8 @@ def cal_multiplier_and_shift(scale):
def quantize_with_scale_and_zero(data, scale, zero): def quantize_with_scale_and_zero(data, scale, zero):
output = np.round(zero + data / scale).astype(np.int32) np_data = np.array(data).astype(float)
output = np.round(zero + np_data / scale).astype(np.int32)
quantized_data = QuantizedData() quantized_data = QuantizedData()
quantized_data.data = output quantized_data.data = output
quantized_data.scale = scale quantized_data.scale = scale
...@@ -140,7 +159,8 @@ def quantize(data, device, non_zero): ...@@ -140,7 +159,8 @@ def quantize(data, device, non_zero):
in_max = np_data.max() in_max = np_data.max()
scale, zero, out_min, out_max = adjust_range(in_min, in_max, device, scale, zero, out_min, out_max = adjust_range(in_min, in_max, device,
non_zero=non_zero) non_zero=non_zero)
output = np.clip((np.round(zero + data / scale).astype(np.int32)), 0, 255) output = np.clip((np.round(zero + np_data / scale).astype(np.int32)),
0, 255)
quantized_data = QuantizedData() quantized_data = QuantizedData()
quantized_data.data = output quantized_data.data = output
......
tools/experimental/utils/config_parser.py → tools/python/quantize/quantize_util_test.py
浏览文件 @ 320b509c
...@@ -12,32 +12,19 @@ ...@@ -12,32 +12,19 @@
# See the License for the specific language governing permissions and # See the License for the specific language governing permissions and
# limitations under the License. # limitations under the License.
from __future__ import absolute_import import unittest
from __future__ import division import numpy as np
from __future__ import print_function import quantize.quantize_util
import re
import os
import yaml
class TestQuantize(unittest.TestCase):
def sanitize_load(s): def test_quantize_dequantize(self):
# do not let yaml parse ON/OFF to boolean test_input = np.random.rand(20, 30) * 5
for w in ["ON", "OFF", "on", "off"]: quantized_data = quantize_util.quantize(test_input)
s = re.sub(r":\s+" + w, r": '" + w + "'", s) dequantized_output = quantize_util.dequantize(quantized_data)
np.testing.assert_array_almost_equal(test_input, dequantized_output, 2)
# sub ${} to env value
s = re.sub(r"\${(\w+)}", lambda x: os.environ[x.group(1)], s)
return yaml.load(s)
if __name__ == '__main__':
def parse(path): unittest.main()
with open(path) as f:
config = sanitize_load(f.read())
return config
def parse_device_info(path):
conf = parse(path)
return conf["devices"]
mace/python/tools/model.jinja2 → tools/python/template/model.jinja2
浏览文件 @ 320b509c
...@@ -166,10 +166,5 @@ const std::string ModelBuildTime() { ...@@ -166,10 +166,5 @@ const std::string ModelBuildTime() {
return {{ build_time|tojson }}; return {{ build_time|tojson }};
} }
const std::string ModelBuildOptions() {
return {{ "obfuscate: {}, embed_model_data: {}, winograd: {}"
.format(obfuscate, embed_model_data, winograd_conv)|tojson }};
}
} // namespace {{tag}} } // namespace {{tag}}
} // namespace mace } // namespace mace
tools/python/template/operator.jinja2 0 → 100644
浏览文件 @ 320b509c
// Copyright 2018 The MACE 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.
// This is a generated file. DO NOT EDIT!
#include <vector>
#include <string>
#include "mace/proto/mace.pb.h"
#include "mace/public/mace.h"
#include "mace/port/env.h"
#include "mace/utils/logging.h"
namespace mace {
namespace {
void UpdateOp(mace::OperatorDef *op,
const std::string &name,
const std::string &type,
const std::vector<std::string> &inputs,
const std::vector<std::string> &outputs,
const std::vector<mace::DataType> &output_types,
uint32_t node_id,
const std::vector<int> &mem_ids) {
op->set_name(name);
op->set_type(type);
op->set_node_id(node_id);
op->mutable_input()->Reserve(inputs.size());
for (auto input : inputs) {
op->add_input(input);
}
op->mutable_output()->Reserve(outputs.size());
for (auto output : outputs) {
op->add_output(output);
}
op->mutable_output_type()->Reserve(output_types.size());
for (auto output_type : output_types) {
op->add_output_type(output_type);
}
op->mutable_mem_id()->Reserve(mem_ids.size());
for (auto mem_id : mem_ids) {
op->add_mem_id(mem_id);
}
}
} // namespace
} // namespace mace
namespace mace {
namespace {{tag}} {
{% for i in range(start, end) %}
void CreateOperator{{i}}(mace::OperatorDef *op) {
MACE_LATENCY_LOGGER(2, "Create operator {{ net.op[i].name }}");
mace::Argument *arg = nullptr;
op->mutable_arg()->Reserve({{ net.op[i].arg|length }});
{% for arg in net.op[i].arg %}
arg = op->add_arg();
arg->set_name({{ arg.name|tojson }});
{%- if arg.HasField('f') %}
arg->set_f({{ arg.f }});
{%- endif %}
{%- if arg.HasField('i') %}
arg->set_i({{ arg.i }});
{%- endif %}
{%- if arg.HasField('s') %}
arg->set_s({{ arg.s.decode('utf-8')|tojson }});
{%- endif %}
{% if arg.floats|length > 0 %}
arg->mutable_floats()->Reserve({{ arg.floats|length }});
{% for float_value in arg.floats %}
arg->add_floats({{ float_value }});
{% endfor %}
{% endif %}
{% if arg.ints|length > 0 %}
arg->mutable_ints()->Reserve({{ arg.ints|length }});
{% for int_value in arg.ints %}
arg->add_ints({{ int_value }});
{% endfor %}
{% endif %}
{% endfor %}
{% if net.op[i].output_shape|length > 0 %}
op->mutable_output_shape()->Reserve({{ net.op[i].output_shape|length }});
{% for shape in net.op[i].output_shape %}
{% if shape.dims|length > 0 %}
{
mace::OutputShape *output_shape = op->add_output_shape();
output_shape->mutable_dims()->Reserve({{ shape.dims|length }});
{% for dim in shape.dims %}
output_shape->add_dims({{ dim }});
{% endfor %}
}
{% endif %}
{% endfor %}
{% endif %}
std::vector<int> output_types_int({ {{ net.op[i].output_type | join(', ') }} });
std::vector<mace::DataType> output_types({{ net.op[i].output_type | length }});
for (int k = 0; k < {{ net.op[i].output_type | length }}; ++k) {
output_types[k] = static_cast<mace::DataType>(output_types_int[k]);
}
UpdateOp(op, {{ net.op[i].name|tojson }}, {{ net.op[i].type|tojson}},
{ {{ net.op[i].input|stringfy }} },
{ {{ net.op[i].output|stringfy }} },
output_types,
{{ net.op[i].node_id }},
{ {{ net.op[i].mem_id | join(', ') }} });
op->mutable_quantize_info()->Reserve({{ net.op[i].quantize_info | length }});
{% for j in range(net.op[i].quantize_info|length) %}
auto quantize_info{{j}} = op->add_quantize_info();
quantize_info{{j}}->set_scale({{ net.op[i].quantize_info[j].scale }});
quantize_info{{j}}->set_zero_point({{ net.op[i].quantize_info[j].zero_point }});
quantize_info{{j}}->set_minval({{ net.op[i].quantize_info[j].minval }});
quantize_info{{j}}->set_maxval({{ net.op[i].quantize_info[j].maxval }});
{% endfor %}
{% if device == 3 %}
op->set_padding({{ net.op[i].padding }});
{% if net.op[i].node_input | length > 0 %}
std::vector<int> input_node_ids({ {{ net.op[i].node_input | map(attribute='node_id') | join(', ') }} });
std::vector<int> input_output_ports({ {{ net.op[i].node_input | map(attribute='output_port') | join(', ')}} });
mace::NodeInput *node_input = nullptr;
op->mutable_node_input()->Reserve({{ net.op[i].node_input|length }});
for (size_t i = 0; i < {{ net.op[i].node_input|length }}; ++i) {
node_input = op->add_node_input();
node_input->set_node_id(input_node_ids[i]);
node_input->set_output_port(input_output_ports[i]);
}
{% endif %}
{% if net.op[i].out_max_byte_size | length > 0 %}
std::vector<int> out_max_byte_sizes {{ net.op[i].out_max_byte_size | replace('[', '{') | replace(']', '}') }};
op->mutable_out_max_byte_size()->Reserve({{ net.op[i].out_max_byte_size|length }});
for (size_t i = 0; i < {{ net.op[i].out_max_byte_size|length }}; ++i) {
op->add_out_max_byte_size(out_max_byte_sizes[i]);
}
{% endif %}
{% endif %}
{% if device == 5 %}
{% if net.op[i].node_input | length > 0 %}
std::vector<int> input_node_ids({ {{ net.op[i].node_input | map(attribute='node_id') | join(', ') }} });
mace::NodeInput *node_input = nullptr;
op->mutable_node_input()->Reserve({{ net.op[i].node_input|length }});
for (size_t i = 0; i < {{ net.op[i].node_input|length }}; ++i) {
node_input = op->add_node_input();
node_input->set_node_id(input_node_ids[i]);
}
{% endif %}
{% endif %}
}
{% endfor %}
} // namespace {{tag}}
} // namespace mace
mace/python/tools/tensor_source.jinja2 → tools/python/template/tensor_source.jinja2
浏览文件 @ 320b509c
...@@ -22,7 +22,7 @@ ...@@ -22,7 +22,7 @@
namespace mace { namespace mace {
namespace {{tag}} { namespace {{tag}} {
void CreateTensor{{tensor_info.id}}(mace::ConstTensor *const_tensor) { void CreateTensor{{tensor_id}}(mace::ConstTensor *const_tensor) {
MACE_LATENCY_LOGGER(2, "Create tensor {{ tensor.name }}"); MACE_LATENCY_LOGGER(2, "Create tensor {{ tensor.name }}");
const_tensor->set_name({{ tensor.name|tojson }}); const_tensor->set_name({{ tensor.name|tojson }});
const_tensor->set_offset({{ tensor.offset }}); const_tensor->set_offset({{ tensor.offset }});
...@@ -30,7 +30,7 @@ void CreateTensor{{tensor_info.id}}(mace::ConstTensor *const_tensor) { ...@@ -30,7 +30,7 @@ void CreateTensor{{tensor_info.id}}(mace::ConstTensor *const_tensor) {
{% for dim in tensor.dims %} {% for dim in tensor.dims %}
const_tensor->add_dims({{ dim }}); const_tensor->add_dims({{ dim }});
{% endfor %} {% endfor %}
const_tensor->set_data_type(static_cast<DataType>({{ tensor_info.data_type }})); const_tensor->set_data_type(static_cast<DataType>({{ tensor.data_type }}));
const_tensor->set_node_id({{ tensor.node_id }}); const_tensor->set_node_id({{ tensor.node_id }});
const_tensor->set_scale({{ tensor.scale }}); const_tensor->set_scale({{ tensor.scale }});
const_tensor->set_zero_point({{ tensor.zero_point }}); const_tensor->set_zero_point({{ tensor.zero_point }});
......
mace/python/tools/converter_tool/apu_converter.py → tools/python/transform/apu_converter.py
浏览文件 @ 320b509c
...@@ -17,19 +17,18 @@ import numpy as np ...@@ -17,19 +17,18 @@ import numpy as np
from enum import Enum from enum import Enum
from operator import mul from operator import mul
from mace.proto import mace_pb2 from py_proto import mace_pb2
from mace.python.tools.converter_tool import base_converter from transform import base_converter
from mace.python.tools.converter_tool.base_converter import ConverterUtil from transform.base_converter import ConverterUtil
from mace.python.tools.converter_tool.base_converter import EltwiseType from transform.base_converter import EltwiseType
from mace.python.tools.converter_tool.base_converter import MaceKeyword from transform.base_converter import MaceKeyword
from mace.python.tools.converter_tool.base_converter import MaceOp from transform.base_converter import MaceOp
from mace.python.tools.converter_tool.base_converter import PaddingMode from transform.base_converter import PaddingMode
from mace.python.tools.converter_tool.base_converter import PoolingType from transform.base_converter import PoolingType
from mace.python.tools.converter_tool.base_converter import ReduceType from transform.base_converter import ReduceType
from mace.python.tools.converter_tool.base_converter import DataFormat from transform.base_converter import DataFormat
from mace.python.tools.converter_tool.base_converter import FrameworkType from transform.base_converter import FrameworkType
from mace.python.tools.convert_util import mace_check from utils.util import mace_check
from mace.python.tools import graph_util
ApuSupportedOps = [ ApuSupportedOps = [
......
mace/python/tools/converter_tool/base_converter.py → tools/python/transform/base_converter.py
浏览文件 @ 320b509c
...@@ -15,7 +15,7 @@ ...@@ -15,7 +15,7 @@
from enum import Enum from enum import Enum
from mace.proto import mace_pb2 from py_proto import mace_pb2
class DeviceType(Enum): class DeviceType(Enum):
......
mace/python/tools/converter_tool/caffe_converter.py → tools/python/transform/caffe_converter.py
浏览文件 @ 320b509c
...@@ -19,20 +19,20 @@ import numpy as np ...@@ -19,20 +19,20 @@ import numpy as np
import six import six
import google.protobuf.text_format import google.protobuf.text_format
from mace.proto import mace_pb2 from py_proto import mace_pb2
from mace.python.tools.converter_tool import base_converter from transform import base_converter
from mace.python.tools.converter_tool import shape_inference from transform import shape_inference
from mace.python.tools.converter_tool.base_converter import PoolingType from transform.base_converter import PoolingType
from mace.python.tools.converter_tool.base_converter import ActivationType from transform.base_converter import ActivationType
from mace.python.tools.converter_tool.base_converter import EltwiseType from transform.base_converter import EltwiseType
from mace.python.tools.converter_tool.base_converter import FrameworkType from transform.base_converter import FrameworkType
from mace.python.tools.converter_tool.base_converter import DataFormat from transform.base_converter import DataFormat
from mace.python.tools.converter_tool.base_converter import MaceOp from transform.base_converter import MaceOp
from mace.python.tools.converter_tool.base_converter import MaceKeyword from transform.base_converter import MaceKeyword
from mace.python.tools.converter_tool.base_converter import ConverterUtil from transform.base_converter import ConverterUtil
from mace.python.tools.convert_util import mace_check from utils.util import mace_check
from third_party.caffe import caffe_pb2 from py_proto import caffe_pb2
caffe_group_str = 'group' caffe_group_str = 'group'
caffe_kernel_h_str = 'kernel_h' caffe_kernel_h_str = 'kernel_h'
......
mace/python/tools/converter_tool/hexagon_converter.py → tools/python/transform/hexagon_converter.py
浏览文件 @ 320b509c
...@@ -17,18 +17,17 @@ import numpy as np ...@@ -17,18 +17,17 @@ import numpy as np
from enum import Enum from enum import Enum
from operator import mul from operator import mul
from mace.proto import mace_pb2 from py_proto import mace_pb2
from mace.python.tools.converter_tool import base_converter from transform import base_converter
from mace.python.tools.converter_tool.base_converter import ConverterUtil from transform.base_converter import ConverterUtil
from mace.python.tools.converter_tool.base_converter import DeviceType from transform.base_converter import DeviceType
from mace.python.tools.converter_tool.base_converter import EltwiseType from transform.base_converter import EltwiseType
from mace.python.tools.converter_tool.base_converter import MaceKeyword from transform.base_converter import MaceKeyword
from mace.python.tools.converter_tool.base_converter import MaceOp from transform.base_converter import MaceOp
from mace.python.tools.converter_tool.base_converter import PaddingMode from transform.base_converter import PaddingMode
from mace.python.tools.converter_tool.base_converter import PoolingType from transform.base_converter import PoolingType
from mace.python.tools.converter_tool.base_converter import ReduceType from transform.base_converter import ReduceType
from mace.python.tools.convert_util import mace_check from utils.util import mace_check
from mace.python.tools import graph_util
from six.moves import reduce from six.moves import reduce
...@@ -144,23 +143,18 @@ class HexagonConverter(base_converter.ConverterInterface): ...@@ -144,23 +143,18 @@ class HexagonConverter(base_converter.ConverterInterface):
return self._model return self._model
def add_port_for_tensors(self, tensors):
for i in range(len(tensors)):
if ':' not in tensors[i]:
node_name = tensors[i]
tensors[i] += ':0'
if node_name in self._quantize_activation_info:
self._quantize_activation_info[tensors[i]] = \
self._quantize_activation_info[node_name]
def convert_ops(self): def convert_ops(self):
print("Convert mace graph to hexagon.") print("Convert mace graph to hexagon.")
for op in self._model.op: for op in self._model.op:
if not self._hexagon_ops.has_op(op.type): if not self._hexagon_ops.has_op(op.type):
raise Exception('Unsupported op: ', op) raise Exception('Unsupported op: ', op)
for i in range(len(op.input)):
self.add_port_for_tensors(op.input) if ':' not in op.input[i]:
self.add_port_for_tensors(op.output) node_name = op.input[i]
op.input[i] += ':0'
if node_name in self._quantize_activation_info:
self._quantize_activation_info[op.input[i]] = \
self._quantize_activation_info[node_name]
if op.type == MaceOp.Conv2D.name \ if op.type == MaceOp.Conv2D.name \
or op.type == MaceOp.DepthwiseConv2d.name: or op.type == MaceOp.DepthwiseConv2d.name:
...@@ -488,15 +482,13 @@ class HexagonConverter(base_converter.ConverterInterface): ...@@ -488,15 +482,13 @@ class HexagonConverter(base_converter.ConverterInterface):
for tensor in self._model.tensors: for tensor in self._model.tensors:
tensor.node_id = node_id_counter tensor.node_id = node_id_counter
node_id_counter += 1 node_id_counter += 1
node_id_map[tensor.name] = tensor.node_id tensor_op, port = get_op_and_port_from_tensor(tensor.name)
node_id_map[tensor_op] = tensor.node_id
print("Hexagon op:") print("Hexagon op:")
index = 0 index = 0
for op in self._model.op: for op in self._model.op:
op.node_id = node_id_counter op.node_id = node_id_counter
node_id_counter += 1
for output in op.output:
node_id_map[output] = op.node_id
if op.type not in [HexagonOp.QuantizeINPUT_f_to_8, if op.type not in [HexagonOp.QuantizeINPUT_f_to_8,
HexagonOp.DequantizeOUTPUT_8tof.name]: HexagonOp.DequantizeOUTPUT_8tof.name]:
index_str = str(index) index_str = str(index)
...@@ -505,10 +497,11 @@ class HexagonConverter(base_converter.ConverterInterface): ...@@ -505,10 +497,11 @@ class HexagonConverter(base_converter.ConverterInterface):
index_str = '' index_str = ''
print('Op: %s (%s, node_id:%d, index:%s)' % print('Op: %s (%s, node_id:%d, index:%s)' %
(op.name, op.type, op.node_id, index_str)) (op.name, op.type, op.node_id, index_str))
node_id_counter += 1
node_id_map[op.name] = op.node_id
for ipt in op.input: for ipt in op.input:
op_name, port = get_op_and_port_from_tensor(ipt) op_name, port = get_op_and_port_from_tensor(ipt)
tensor_name = ipt if port == 0 else op_name + ':0' node_id = node_id_map[op_name]
node_id = node_id_map[tensor_name]
node_input = op.node_input.add() node_input = op.node_input.add()
node_input.node_id = node_id node_input.node_id = node_id
node_input.output_port = int(port) node_input.output_port = int(port)
mace/python/tools/converter_tool/onnx_converter.py → tools/python/transform/onnx_converter.py
浏览文件 @ 320b509c
...@@ -17,20 +17,20 @@ import sys ...@@ -17,20 +17,20 @@ import sys
from enum import Enum from enum import Enum
import six import six
from mace.proto import mace_pb2 from py_proto import mace_pb2
from mace.python.tools.converter_tool import base_converter from transform import base_converter
from mace.python.tools.converter_tool.base_converter import PoolingType from transform.base_converter import PoolingType
from mace.python.tools.converter_tool.base_converter import PaddingMode from transform.base_converter import PaddingMode
from mace.python.tools.converter_tool.base_converter import ActivationType from transform.base_converter import ActivationType
from mace.python.tools.converter_tool.base_converter import EltwiseType from transform.base_converter import EltwiseType
from mace.python.tools.converter_tool.base_converter import ReduceType from transform.base_converter import ReduceType
from mace.python.tools.converter_tool.base_converter import FrameworkType from transform.base_converter import FrameworkType
from mace.python.tools.converter_tool.base_converter import RoundMode from transform.base_converter import RoundMode
from mace.python.tools.converter_tool.base_converter import DataFormat from transform.base_converter import DataFormat
from mace.python.tools.converter_tool.base_converter import MaceOp from transform.base_converter import MaceOp
from mace.python.tools.converter_tool.base_converter import MaceKeyword from transform.base_converter import MaceKeyword
from mace.python.tools.converter_tool.base_converter import ConverterUtil from transform.base_converter import ConverterUtil
from mace.python.tools.convert_util import mace_check from utils.util import mace_check
import numpy as np import numpy as np
......
mace/python/tools/converter_tool/shape_inference.py → tools/python/transform/shape_inference.py
浏览文件 @ 320b509c
...@@ -18,12 +18,12 @@ import math ...@@ -18,12 +18,12 @@ import math
import numpy as np import numpy as np
import six import six
from mace.python.tools.converter_tool.transformer import Transformer from transform.transformer import Transformer
from mace.python.tools.converter_tool.base_converter import DataFormat from transform.base_converter import DataFormat
from mace.python.tools.converter_tool.base_converter import MaceOp from transform.base_converter import MaceOp
from mace.python.tools.converter_tool.base_converter import MaceKeyword from transform.base_converter import MaceKeyword
from mace.python.tools.converter_tool.base_converter import ConverterUtil from transform.base_converter import ConverterUtil
from mace.python.tools.convert_util import mace_check from utils.util import mace_check
class ShapeInference(object): class ShapeInference(object):
......
mace/python/tools/converter_tool/tensorflow_converter.py → tools/python/transform/tensorflow_converter.py
浏览文件 @ 320b509c
...@@ -19,20 +19,20 @@ import six ...@@ -19,20 +19,20 @@ import six
import tensorflow as tf import tensorflow as tf
from enum import Enum from enum import Enum
from mace.proto import mace_pb2 from py_proto import mace_pb2
from mace.python.tools.converter_tool import base_converter from transform import base_converter
from mace.python.tools.converter_tool.base_converter import PoolingType from transform.base_converter import PoolingType
from mace.python.tools.converter_tool.base_converter import PaddingMode from transform.base_converter import PaddingMode
from mace.python.tools.converter_tool.base_converter import ActivationType from transform.base_converter import ActivationType
from mace.python.tools.converter_tool.base_converter import EltwiseType from transform.base_converter import EltwiseType
from mace.python.tools.converter_tool.base_converter import PadType from transform.base_converter import PadType
from mace.python.tools.converter_tool.base_converter import FrameworkType from transform.base_converter import FrameworkType
from mace.python.tools.converter_tool.base_converter import ReduceType from transform.base_converter import ReduceType
from mace.python.tools.converter_tool.base_converter import DataFormat from transform.base_converter import DataFormat
from mace.python.tools.converter_tool.base_converter import MaceOp from transform.base_converter import MaceOp
from mace.python.tools.converter_tool.base_converter import MaceKeyword from transform.base_converter import MaceKeyword
from mace.python.tools.converter_tool.base_converter import ConverterUtil from transform.base_converter import ConverterUtil
from mace.python.tools.convert_util import mace_check from utils.util import mace_check
from tensorflow.core.framework import tensor_shape_pb2 from tensorflow.core.framework import tensor_shape_pb2
from tensorflow.tools.graph_transforms import TransformGraph from tensorflow.tools.graph_transforms import TransformGraph
......
mace/python/tools/converter_tool/transformer.py → tools/python/transform/transformer.py
浏览文件 @ 320b509c
...@@ -18,22 +18,22 @@ import re ...@@ -18,22 +18,22 @@ import re
import numpy as np import numpy as np
import six import six
from mace.proto import mace_pb2 from py_proto import mace_pb2
from mace.python.tools.converter_tool import base_converter from transform import base_converter
from mace.python.tools.converter_tool.base_converter import ConverterUtil from transform.base_converter import ConverterUtil
from mace.python.tools.converter_tool.base_converter import DataFormat from transform.base_converter import DataFormat
from mace.python.tools.converter_tool.base_converter import DeviceType from transform.base_converter import DeviceType
from mace.python.tools.converter_tool.base_converter import EltwiseType from transform.base_converter import EltwiseType
from mace.python.tools.converter_tool.base_converter import FrameworkType from transform.base_converter import FrameworkType
from mace.python.tools.converter_tool.base_converter import MaceKeyword from transform.base_converter import MaceKeyword
from mace.python.tools.converter_tool.base_converter import MaceOp from transform.base_converter import MaceOp
from mace.python.tools.converter_tool.base_converter import MaceFixedDataFormatOps # noqa from transform.base_converter import MaceFixedDataFormatOps # noqa
from mace.python.tools.converter_tool.base_converter import MaceTransposableDataFormatOps # noqa from transform.base_converter import MaceTransposableDataFormatOps # noqa
from mace.python.tools.converter_tool.base_converter import PaddingMode from transform.base_converter import PaddingMode
from mace.python.tools.converter_tool.base_converter import ReduceType from transform.base_converter import ReduceType
from mace.python.tools.converter_tool.base_converter import TransformerRule from transform.base_converter import TransformerRule
from mace.python.tools.convert_util import mace_check from quantize import quantize_util
from mace.python.tools.quantization import quantize_util from utils.util import mace_check
class Transformer(base_converter.ConverterInterface): class Transformer(base_converter.ConverterInterface):
...@@ -1309,12 +1309,27 @@ class Transformer(base_converter.ConverterInterface): ...@@ -1309,12 +1309,27 @@ class Transformer(base_converter.ConverterInterface):
return False return False
def update_float_op_data_type(self): def update_float_op_data_type(self):
if self._option.quantize:
return
print("update op with float data type") print("update op with float data type")
net = self._model net = self._model
data_type = self._option.data_type data_type = self._option.data_type
net.data_type = data_type
for tensor in net.tensors:
if tensor.data_type == mace_pb2.DT_FLOAT:
tensor.data_type = data_type
if tensor.data_type == mace_pb2.DT_FLOAT \
or tensor.data_type == mace_pb2.DT_HALF \
or tensor.data_type == mace_pb2.DT_FLOAT16:
tensor.data_size = len(tensor.float_data)
elif tensor.data_type == mace_pb2.DT_INT32:
tensor.data_size = len(tensor.int32_data)
elif tensor.data_type == mace_pb2.DT_UINT8:
tensor.data_size = len(tensor.int32_data)
if self._option.quantize:
return
for op in net.op: for op in net.op:
data_type_arg = ConverterUtil.get_arg( data_type_arg = ConverterUtil.get_arg(
op, MaceKeyword.mace_op_data_type_str) op, MaceKeyword.mace_op_data_type_str)
...@@ -1732,6 +1747,8 @@ class Transformer(base_converter.ConverterInterface): ...@@ -1732,6 +1747,8 @@ class Transformer(base_converter.ConverterInterface):
self._quantize_activation_info[op.input[0]] = quantize_info self._quantize_activation_info[op.input[0]] = quantize_info
# for add -> fakequant pattern # for add -> fakequant pattern
self._quantize_activation_info[op.output[0]] = quantize_info self._quantize_activation_info[op.output[0]] = quantize_info
print(op.input[0], op.output[0])
op.type = MaceOp.Identity.name op.type = MaceOp.Identity.name
return False return False
...@@ -1857,7 +1874,8 @@ class Transformer(base_converter.ConverterInterface): ...@@ -1857,7 +1874,8 @@ class Transformer(base_converter.ConverterInterface):
self.copy_quantize_info( self.copy_quantize_info(
op, self._quantize_activation_info[new_input_name]) op, self._quantize_activation_info[new_input_name])
else: else:
self.copy_quantize_info(op, producer_op.quantize_info[0]) self.copy_quantize_info(op,
producer_op.quantize_info[0])
self._quantize_activation_info[op.output[0]] = \ self._quantize_activation_info[op.output[0]] = \
op.quantize_info[0] op.quantize_info[0]
elif (op.type == MaceOp.Concat.name elif (op.type == MaceOp.Concat.name
......
tools/python/utils/config_parser.py 0 → 100644
浏览文件 @ 320b509c
# Copyright 2019 The MACE 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.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import re
import os
import yaml
CPP_KEYWORDS = [
'alignas', 'alignof', 'and', 'and_eq', 'asm', 'atomic_cancel',
'atomic_commit', 'atomic_noexcept', 'auto', 'bitand', 'bitor',
'bool', 'break', 'case', 'catch', 'char', 'char16_t', 'char32_t',
'class', 'compl', 'concept', 'const', 'constexpr', 'const_cast',
'continue', 'co_await', 'co_return', 'co_yield', 'decltype', 'default',
'delete', 'do', 'double', 'dynamic_cast', 'else', 'enum', 'explicit',
'export', 'extern', 'false', 'float', 'for', 'friend', 'goto', 'if',
'import', 'inline', 'int', 'long', 'module', 'mutable', 'namespace',
'new', 'noexcept', 'not', 'not_eq', 'nullptr', 'operator', 'or', 'or_eq',
'private', 'protected', 'public', 'register', 'reinterpret_cast',
'requires', 'return', 'short', 'signed', 'sizeof', 'static',
'static_assert', 'static_cast', 'struct', 'switch', 'synchronized',
'template', 'this', 'thread_local', 'throw', 'true', 'try', 'typedef',
'typeid', 'typename', 'union', 'unsigned', 'using', 'virtual', 'void',
'volatile', 'wchar_t', 'while', 'xor', 'xor_eq', 'override', 'final',
'transaction_safe', 'transaction_safe_dynamic', 'if', 'elif', 'else',
'endif', 'defined', 'ifdef', 'ifndef', 'define', 'undef', 'include',
'line', 'error', 'pragma',
]
def sanitize_load(s):
# do not let yaml parse ON/OFF to boolean
for w in ["ON", "OFF", "on", "off"]:
s = re.sub(r":\s+" + w, r": '" + w + "'", s)
# sub ${} to env value
s = re.sub(r"\${(\w+)}", lambda x: os.environ[x.group(1)], s)
return yaml.load(s)
def parse(path):
with open(path) as f:
config = sanitize_load(f.read())
return config
def parse_device_info(path):
conf = parse(path)
return conf["devices"]
tools/python/utils/util.py 0 → 100644
浏览文件 @ 320b509c
# Copyright 2019 The MACE 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.
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import inspect
import hashlib
import os
import urllib
from utils import device
################################
# log
################################
class CMDColors:
PURPLE = '\033[95m'
BLUE = '\033[94m'
GREEN = '\033[92m'
YELLOW = '\033[93m'
RED = '\033[91m'
ENDC = '\033[0m'
BOLD = '\033[1m'
UNDERLINE = '\033[4m'
def get_frame_info(level=2):
caller_frame = inspect.stack()[level]
info = inspect.getframeinfo(caller_frame[0])
return info.filename + ':' + str(info.lineno) + ': '
class MaceLogger:
@staticmethod
def header(message):
print(CMDColors.PURPLE + message + CMDColors.ENDC)
@staticmethod
def summary(message):
print(CMDColors.GREEN + message + CMDColors.ENDC)
@staticmethod
def info(message):
print(get_frame_info() + message)
@staticmethod
def warning(message):
print(CMDColors.YELLOW + 'WARNING: ' + get_frame_info() + message
+ CMDColors.ENDC)
@staticmethod
def error(message):
print(CMDColors.RED + 'ERROR: ' + get_frame_info() + message
+ CMDColors.ENDC)
exit(1)
def mace_check(condition, message):
if not condition:
MaceLogger.error(message)
################################
# file
################################
def file_checksum(fname):
hash_func = hashlib.sha256()
with open(fname, "rb") as f:
for chunk in iter(lambda: f.read(4096), b""):
hash_func.update(chunk)
return hash_func.hexdigest()
def download_or_get_file(file,
sha256_checksum,
output_dir):
model_file = output_dir + "/" + sha256_checksum + ".pb"
if file.startswith("http://") or file.startswith("https://"):
if not os.path.exists(model_file) or file_checksum(
model_file) != sha256_checksum:
MaceLogger.info("Downloading file %s, please wait ..." % file)
urllib.urlretrieve(file, model_file)
MaceLogger.info("Model downloaded successfully.")
else:
device.execute("cp %s %s" % (file, model_file))
return model_file
mace/python/tools/visualization/visualize_model.py → tools/python/visualize/visualize_model.py
浏览文件 @ 320b509c
import os
import json import json
import numpy as np import numpy as np
...@@ -19,8 +20,8 @@ class NPEncoder(json.JSONEncoder): ...@@ -19,8 +20,8 @@ class NPEncoder(json.JSONEncoder):
class ModelVisualizer(object): class ModelVisualizer(object):
def __init__(self, model_name, proto): def __init__(self, model_name, proto, output_dir):
self._output_file = "build/%s_index.html" % model_name self._output_file = "%s/%s_index.html" % (output_dir, model_name)
self._proto = proto self._proto = proto
def render_html(self): def render_html(self):
...@@ -82,7 +83,8 @@ class ModelVisualizer(object): ...@@ -82,7 +83,8 @@ class ModelVisualizer(object):
json_msg = json.dumps(json_obj, cls=NPEncoder) json_msg = json.dumps(json_obj, cls=NPEncoder)
with open("mace/python/tools/visualization/index.html") as f: cwd = os.path.dirname(__file__)
with open(cwd + "/index.html") as f:
html = f.read() html = f.read()
return html % json_msg return html % json_msg
......
tools/sh_commands.py
浏览文件 @ 320b509c
...@@ -368,7 +368,6 @@ def gen_mace_engine_factory_source(model_tags, ...@@ -368,7 +368,6 @@ def gen_mace_engine_factory_source(model_tags,
sh.mkdir("-p", codegen_tools_dir) sh.mkdir("-p", codegen_tools_dir)
gen_mace_engine_factory( gen_mace_engine_factory(
model_tags, model_tags,
"mace/python/tools",
embed_model_data, embed_model_data,
codegen_tools_dir) codegen_tools_dir)
six.print_("Generate mace engine creator source done!\n") six.print_("Generate mace engine creator source done!\n")
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或