提交 cbc3efdb 编写于 作者: S SunAhong1993

add densenet

上级 5bcd803c
...@@ -13,8 +13,7 @@ ...@@ -13,8 +13,7 @@
# limitations under the License. # limitations under the License.
def convert_prim(layer, indent=1, init_func=[], forward_func=[]): def gen_codes(code_list, indent=0):
def gen_codes(code_list, indent=0):
indent_blank = " " * indent indent_blank = " " * indent
codes = [] codes = []
for code_line in code_list: for code_line in code_list:
...@@ -24,12 +23,75 @@ def convert_prim(layer, indent=1, init_func=[], forward_func=[]): ...@@ -24,12 +23,75 @@ def convert_prim(layer, indent=1, init_func=[], forward_func=[]):
codes.append(indent_blank + code_line + '\n') codes.append(indent_blank + code_line + '\n')
return codes return codes
if layer.kernel == "prim.if":
def prim_add(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {} + {}".format(layer.outputs[0], layer.inputs["x"],
layer.inputs["y"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_add_(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {} + {} * {}".format(layer.outputs[0], layer.inputs["x"],
layer.attrs["alpha"], layer.inputs["y"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_and(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {} and {}".format(layer.outputs[0], layer.inputs["x"],
layer.inputs["y"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_append(layer, indent=1, init_func=[], forward_func=[]):
line = "{}.append({})".format(layer.inputs["list"], layer.inputs["element"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_assert(layer, indent=1, init_func=[], forward_func=[]):
if layer.attrs["type"] == "eq":
if isinstance(layer.attrs["value"], list):
s = ""
for v in layer.attrs["value"]:
s += "{} == {} or ".format(layer.attrs["key"], v)
if len(s) > 0:
s = s[:-4]
line = "assert {}, \'The {} must be {}!\'".format(
s, layer.attrs["key"], layer.attrs["value"])
else:
line = "assert {} == {}, \'The {} must be {}!\'".format(
layer.attrs["key"], layer.attrs["value"], layer.attrs["key"],
layer.attrs["value"])
else:
raise Exception("Not implement yet!")
forward_func.extend(gen_codes([line], indent=indent))
def prim_constant(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {}".format(layer.outputs[0], layer.attrs["value"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_eq(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {} == {}".format(layer.outputs[0], layer.inputs["x"],
layer.inputs["y"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_equal(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {}".format(layer.outputs[0], layer.inputs["input"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_exception(layer, indent=1, init_func=[], forward_func=[]):
line = "raise RaiseException({})".format(layer.inputs["input"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_if(layer, indent=1, init_func=[], forward_func=[]):
line = "if {} :".format(list(layer.inputs.values())[0]) line = "if {} :".format(list(layer.inputs.values())[0])
forward_func.extend(gen_codes([line], indent=indent)) forward_func.extend(gen_codes([line], indent=indent))
block = layer.blocks[0] block = layer.blocks[0]
b_init_lines, b_forward_lines = block.gen_dygraph_code( b_init_lines, b_forward_lines = block.gen_dygraph_code(indent=indent + 1)
indent=indent + 1)
init_func.extend(b_init_lines) init_func.extend(b_init_lines)
forward_func.extend(b_forward_lines) forward_func.extend(b_forward_lines)
block = layer.blocks[1] block = layer.blocks[1]
...@@ -40,117 +102,131 @@ def convert_prim(layer, indent=1, init_func=[], forward_func=[]): ...@@ -40,117 +102,131 @@ def convert_prim(layer, indent=1, init_func=[], forward_func=[]):
indent=indent + 1) indent=indent + 1)
init_func.extend(b_init_lines) init_func.extend(b_init_lines)
forward_func.extend(b_forward_lines) forward_func.extend(b_forward_lines)
return
elif layer.kernel == "prim.loop":
def prim_getitem(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {}[{}]".format(layer.outputs[0], layer.inputs["list"],
layer.inputs["index"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_gt(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {} > {}".format(layer.outputs[0], layer.inputs["x"],
layer.inputs["y"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_le(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {} <= {}".format(layer.outputs[0], layer.inputs["x"],
layer.inputs["y"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_len(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = len({})".format(layer.outputs[0], layer.inputs["input"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_lt(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {} < {}".format(layer.outputs[0], layer.inputs["x"],
layer.inputs["y"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_list(layer, indent=1, init_func=[], forward_func=[]):
inputs_list = list(layer.inputs.values())
inputs_str = ', '.join(inputs_list)
line = "{} = [{}]".format(layer.outputs[0], inputs_str)
forward_func.extend(gen_codes([line], indent=indent))
def prim_loop(layer, indent=1, init_func=[], forward_func=[]):
loop_range = list(layer.inputs.values())[0] loop_range = list(layer.inputs.values())[0]
if list(layer.inputs.values())[0] is None: if list(layer.inputs.values())[0] is None:
loop_range = str(layer.attrs[list(layer.inputs.keys())[0]]) loop_range = str(layer.attrs[list(layer.inputs.keys())[0]])
line = "for {} in range({}):".format(layer.outputs[1], loop_range) line = "for {} in range({}):".format(layer.outputs[1], loop_range)
forward_func.extend(gen_codes([line], indent=indent)) forward_func.extend(gen_codes([line], indent=indent))
block = layer.blocks[0] block = layer.blocks[0]
b_init_lines, b_forward_lines = block.gen_dygraph_code( b_init_lines, b_forward_lines = block.gen_dygraph_code(indent=indent + 1)
indent=indent + 1)
init_func.extend(b_init_lines) init_func.extend(b_init_lines)
forward_func.extend(b_forward_lines) forward_func.extend(b_forward_lines)
return
elif layer.kernel == "prim.equal":
line = "{} = {}".format(layer.outputs[0], def prim_min(layer, indent=1, init_func=[], forward_func=[]):
list(layer.inputs.values())[0]) line = "{} = min({})".format(layer.outputs[0], layer.inputs["input"])
elif layer.kernel == "prim.constant": forward_func.extend(gen_codes([line], indent=indent))
line = "{} = {}".format(layer.outputs[0], layer.attrs["value"])
elif layer.kernel == "prim.list":
def prim_mul(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {} * {}".format(layer.outputs[0], layer.inputs["x"],
layer.inputs["y"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_ne(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {} < {}".format(layer.outputs[0], layer.inputs["x"],
layer.inputs["y"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_neg(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = -{}".format(layer.outputs[0], layer.inputs["input"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_not(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = not {}".format(layer.outputs[0], layer.inputs["input"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_requires_grad(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = not {}.stop_gradient".format(layer.outputs[0],
layer.inputs["input"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_select(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {}[".format(layer.outputs[0], layer.inputs["input"])
for dim in range(layer.attrs["dim"]):
line += ":, "
line += (layer.inputs["index"] + "]")
forward_func.extend(gen_codes([line], indent=indent))
def prim_shape(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {}.shape".format(layer.outputs[0], layer.inputs["input"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_slice(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {}[{}: {}: {}]".format(
layer.outputs[0], layer.inputs["input"], layer.inputs["start"],
layer.inputs["end"], layer.inputs["step"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_sub(layer, indent=1, init_func=[], forward_func=[]):
line = "{} = {} - {}".format(layer.outputs[0], layer.inputs["x"],
layer.inputs["y"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_tuple(layer, indent=1, init_func=[], forward_func=[]):
inputs_list = list(layer.inputs.values()) inputs_list = list(layer.inputs.values())
for i, input in enumerate(inputs_list):
if input is None:
inputs_list[i] = str(layer.attrs[list(layer.inputs.keys())[i]])
inputs_str = ', '.join(inputs_list) inputs_str = ', '.join(inputs_list)
line = "{} = [{}]".format(layer.outputs[0], inputs_str) line = "{} = ({})".format(layer.outputs[0], inputs_str)
elif layer.kernel == "prim.exception":
exception = list(layer.inputs.values())[0]
if list(layer.inputs.values())[0] is None:
exception = str(layer.attrs[list(layer.inputs.keys())[0]])
line = "raise RaiseException({})".format(exception)
elif layer.kernel == "prim.min":
line = "{} = min({})".format(layer.outputs[0],
list(layer.inputs.values())[0])
elif layer.kernel == "prim.add_":
line = "{} = {} + {} * {}".format(layer.outputs[0],
list(layer.inputs.values())[0],
layer.attrs["alpha"],
list(layer.inputs.values())[1])
elif layer.kernel == "prim.append":
line = "{} = {}.append({})".format(layer.outputs[0],
list(layer.inputs.values())[0],
list(layer.inputs.values())[1])
elif layer.kernel == "prim.shape":
line = "{} = {}.shape".format(layer.outputs[0],
list(layer.inputs.values())[0])
elif layer.kernel == "prim.len":
line = "{} = len({})".format(layer.outputs[0],
list(layer.inputs.values())[0])
elif layer.kernel == "prim.eq":
line = "{} = {} == {}".format(layer.outputs[0],
list(layer.inputs.values())[0],
list(layer.inputs.values())[1])
elif layer.kernel == "prim.assert":
if layer.attrs["type"] == "eq":
if isinstance(layer.attrs["value"], list):
s = ""
for v in layer.attrs["value"]:
s += "{} == {} or ".format(layer.attrs["key"], v)
if len(s) > 0:
s = s[:-4]
line = "assert {}, \'The {} must be {}!\'".format(
s, layer.attrs["key"], layer.attrs["value"])
else:
line = "assert {} == {}, \'The {} must be {}!\'".format(
layer.attrs["key"], layer.attrs["value"],
layer.attrs["key"], layer.attrs["value"])
else:
raise Exception("Not implement yet!")
elif layer.kernel == "prim.getitem":
item0 = list(layer.inputs.values())[0]
if list(layer.inputs.values())[0] is None:
item0 = str(layer.attrs[list(layer.inputs.keys())[0]])
item1 = list(layer.inputs.values())[1]
if list(layer.inputs.values())[1] is None:
item1 = str(layer.attrs[list(layer.inputs.keys())[1]])
line = "{} = {}[{}]".format(layer.outputs[0], item0, item1)
elif layer.kernel == "prim.le":
item0 = list(layer.inputs.values())[0]
if list(layer.inputs.values())[0] is None:
item0 = str(layer.attrs[list(layer.inputs.keys())[0]])
item1 = list(layer.inputs.values())[1]
if list(layer.inputs.values())[1] is None:
item1 = str(layer.attrs[list(layer.inputs.keys())[1]])
line = "{} = {} < {}".format(layer.outputs[0], item0, item1)
elif layer.kernel == "prim.ne":
item0 = list(layer.inputs.values())[0]
item1 = list(layer.inputs.values())[1]
line = "{} = {} < {}".format(layer.outputs[0], item0, item1)
elif layer.kernel == "prim.slice":
inputs_str = ""
for v in list(layer.inputs.values())[1:]:
inputs_str += "{}:".format(v)
inputs_str = inputs_str[:-1]
line = "{} = {}[{}]".format(layer.outputs[0],
list(layer.inputs.values())[0], inputs_str)
elif layer.kernel == "prim.add":
line = "{} = {} + {}".format(layer.outputs[0],
list(layer.inputs.values())[0],
list(layer.inputs.values())[1])
elif layer.kernel == "prim.sub":
line = "{} = {} - {}".format(layer.outputs[0],
list(layer.inputs.values())[0],
list(layer.inputs.values())[1])
elif layer.kernel == "prim.mul":
line = "{} = {} * {}".format(layer.outputs[0],
list(layer.inputs.values())[0],
list(layer.inputs.values())[1])
elif layer.kernel == "prim.neg":
line = "{} = -{}".format(layer.outputs[0],
list(layer.inputs.values())[0])
else:
print(layer.kernel)
line = ""
forward_func.extend(gen_codes([line], indent=indent)) forward_func.extend(gen_codes([line], indent=indent))
def prim_tuple_unpack(layer, indent=1, init_func=[], forward_func=[]):
outputs_str = ', '.join(layer.outputs)
line = "{} = {}".format(outputs_str, layer.inputs["input"])
forward_func.extend(gen_codes([line], indent=indent))
def prim_warnings(layer, indent=1, init_func=[], forward_func=[]):
lines = ["import warnings"]
line = "warnings.warn({}, stacklevel={})".format(layer.inputs["input"],
layer.attrs["stacklevel"])
lines.append(line)
forward_func.extend(gen_codes(lines, indent=indent))
...@@ -297,7 +297,6 @@ class PaddleGraph(object): ...@@ -297,7 +297,6 @@ class PaddleGraph(object):
for output_name in layer.outputs: for output_name in layer.outputs:
if not output_name.startswith("x"): if not output_name.startswith("x"):
continue continue
print(layer.kernel)
self.outputs.append(output_name) self.outputs.append(output_name)
self.outputs = list(set(self.outputs)) self.outputs = list(set(self.outputs))
...@@ -396,12 +395,19 @@ class PaddleGraph(object): ...@@ -396,12 +395,19 @@ class PaddleGraph(object):
line += ")" line += ")"
self.forward_func.extend(gen_codes([line], indent=indent)) self.forward_func.extend(gen_codes([line], indent=indent))
elif "prim" in layer.kernel: elif "prim" in layer.kernel:
from .convert_prim import convert_prim func_name = layer.kernel.replace(".", "_")
convert_prim( from . import convert_prim
if hasattr(convert_prim, func_name):
func = getattr(convert_prim, func_name)
func(
layer, layer,
indent=indent, indent=indent,
init_func=self.init_func, init_func=self.init_func,
forward_func=self.forward_func) forward_func=self.forward_func)
else:
raise Exception(
"The kind {} in paddle model is not supported yet.".
format(layer.kernel))
else: else:
if len(layer.outputs) == 1: if len(layer.outputs) == 1:
line = layer.outputs[0] line = layer.outputs[0]
......
...@@ -183,6 +183,36 @@ def aten_add_(mapper, graph, node): ...@@ -183,6 +183,36 @@ def aten_add_(mapper, graph, node):
return current_inputs, current_outputs return current_inputs, current_outputs
def aten___and__(mapper, graph, node):
""" 构造与计算的PaddleLayer。
TorchScript示例:
%361 : bool = aten::__and__(%360, %358)
参数含义:
%361 (bool): 输出,与计算结果。
%360 (-): 输入 x。
%358 (-): 输入 y。
"""
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name]
layer_inputs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = [output_name]
# 处理输入0,即%i.12
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["x"] = inputs_name[0]
# 处理输入1,即%288
mapper._check_input(
graph, inputs_node[1], inputs_name[1], current_outputs, add_dim=True)
layer_inputs["y"] = inputs_name[1]
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
graph.add_layer("prim.and", inputs=layer_inputs, outputs=layer_outputs)
return current_inputs, current_outputs
def aten_append(mapper, graph, node): def aten_append(mapper, graph, node):
""" 构造对list进行append的PaddleLayer。 """ 构造对list进行append的PaddleLayer。
...@@ -193,12 +223,11 @@ def aten_append(mapper, graph, node): ...@@ -193,12 +223,11 @@ def aten_append(mapper, graph, node):
%_output_size.1 (list): 需要进行append的list。 %_output_size.1 (list): 需要进行append的list。
%v.1 (-): append的元素。 %v.1 (-): append的元素。
""" """
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name]
layer_inputs = {} layer_inputs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node) inputs_name, inputs_node = mapper._get_inputs_name(node)
layer_outputs = [inputs_name[0]]
# 获取当前节点输出的list # 获取当前节点输出的list
current_outputs = [output_name] current_outputs = [inputs_name[0]]
# 处理输入0,即_output_size.1 # 处理输入0,即_output_size.1
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs) mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["list"] = inputs_name[0] layer_inputs["list"] = inputs_name[0]
...@@ -212,6 +241,66 @@ def aten_append(mapper, graph, node): ...@@ -212,6 +241,66 @@ def aten_append(mapper, graph, node):
return current_inputs, current_outputs return current_inputs, current_outputs
def aten_avg_pool2d(mapper, graph, node):
""" 构造最大池化的PaddleLayer。
TorchScript示例:
%branch_pool.2 : Tensor = aten::avg_pool2d(%x.43, %538, %539, %540, %273, %272, %271)
参数含义:
%branch_pool.2 (Tensor): 输出,池化后的结果。
%x.43 (Tensor): 需要池化的Tensor。
%538 (list): 池化kernel的大小。
%539 (list): 步长大小。
%540 (list): 填充大小。
%273 (bool): 是否用ceil函数计算输出高度和宽度。
%272 (bool): 是否在平均池化模式不忽略填充值,False为忽略。
%271 (int): 如果指定,它将用作除数,否则将使用池化区域的大小。
"""
if "pool" in mapper.dygraph_name_id:
mapper.dygraph_name_id["pool"] += 1
else:
mapper.dygraph_name_id["pool"] = 0
pool_name = "pool" + str(mapper.dygraph_name_id["pool"])
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [pool_name, output_name]
layer_inputs = {}
layer_attrs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = [output_name]
# 处理输入0,即%x.34
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["input"] = inputs_name[0]
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
# 处理输入1,即%538
layer_attrs["pool_size"] = mapper.attrs[inputs_name[1]]
# 处理输入2,即%539
layer_attrs["pool_stride"] = mapper.attrs[inputs_name[2]]
# 处理输入3,即%540
layer_attrs["pool_padding"] = mapper.attrs[inputs_name[3]]
# 处理输入4,即%273
layer_attrs["ceil_mode"] = mapper.attrs[inputs_name[4]]
# 处理输入5,即%272
layer_attrs["exclusive"] = not mapper.attrs[inputs_name[5]]
# 处理输入6,即%271
graph.add_layer(
"prim.assert",
inputs={},
outputs=[inputs_name[6]],
type="eq",
key=mapper.attrs[inputs_name[6]],
value=None)
layer_attrs["pool_type"] = string("avg")
graph.add_layer(
"fluid.dygraph.Pool2D",
inputs=layer_inputs,
outputs=layer_outputs,
**layer_attrs)
return current_inputs, current_outputs
def aten_batch_norm(mapper, graph, node): def aten_batch_norm(mapper, graph, node):
""" 构造BatchNorm的PaddleLayer。 """ 构造BatchNorm的PaddleLayer。
...@@ -278,6 +367,44 @@ def aten_batch_norm(mapper, graph, node): ...@@ -278,6 +367,44 @@ def aten_batch_norm(mapper, graph, node):
return current_inputs, current_outputs return current_inputs, current_outputs
def aten_cat(mapper, graph, node):
""" 构造连接Tensor的PaddleLayer。
TorchScript示例:
%x.222 : Tensor = aten::cat(%32, %7)
参数含义:
%x.222 (Tensor): 输出,连接后的结果。
%i.12 (list): 需要连接的Tensor组成的list。
%7 (int): 连接的轴。
"""
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name]
layer_inputs = {}
layer_attrs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = [output_name]
# 处理输入0,即%13
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["input"] = inputs_name[0]
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
# 处理输入1,即%12
if inputs_name[1] in mapper.attrs:
layer_attrs["axis"] = mapper.attrs[inputs_name[1]]
else:
mapper._check_input(graph, inputs_node[1], inputs_name[1],
current_outputs)
layer_attrs["axis"] = inputs_name[1]
current_inputs.append(inputs_name[1])
graph.add_layer(
"fluid.layers.concat",
inputs=layer_inputs,
outputs=layer_outputs,
**layer_attrs)
return current_inputs, current_outputs
def aten_conv2d(mapper, graph, node): def aten_conv2d(mapper, graph, node):
""" 构造conv2d的PaddleLayer。 """ 构造conv2d的PaddleLayer。
...@@ -512,6 +639,35 @@ def aten___getitem__(mapper, graph, node): ...@@ -512,6 +639,35 @@ def aten___getitem__(mapper, graph, node):
return current_inputs, current_outputs return current_inputs, current_outputs
def aten_gt(mapper, graph, node):
""" 构造对比大小的PaddleLayer。
TorchScript示例:
%83 : bool = aten::gt(%82, %78)
参数含义:
%83 (bool): 输出,第一个元素是否大于第二个元素。
%82 (-): 需对比的输入1。
%78 (-): 需对比的输入2。
"""
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name]
layer_inputs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = [output_name]
# 处理输入0,即%82
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["x"] = inputs_name[0]
# 处理输入1,即%78
mapper._check_input(graph, inputs_node[1], inputs_name[1], current_outputs)
layer_inputs["y"] = inputs_name[1]
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
graph.add_layer("prim.gt", inputs=layer_inputs, outputs=layer_outputs)
return current_inputs, current_outputs
def aten_hardtanh_(mapper, graph, node): def aten_hardtanh_(mapper, graph, node):
""" 构造hardtanh激活的PaddleLayer。 """ 构造hardtanh激活的PaddleLayer。
...@@ -565,7 +721,7 @@ def aten_le(mapper, graph, node): ...@@ -565,7 +721,7 @@ def aten_le(mapper, graph, node):
TorchScript示例: TorchScript示例:
%80 : bool = aten::le(%78, %79) %80 : bool = aten::le(%78, %79)
参数含义: 参数含义:
%80 (bool): 输出,第一个元素是否小于第二个元素。 %80 (bool): 输出,第一个元素是否小于等于第二个元素。
%78 (-): 需对比的输入1。 %78 (-): 需对比的输入1。
%79 (-): 需对比的输入2。 %79 (-): 需对比的输入2。
""" """
...@@ -577,10 +733,10 @@ def aten_le(mapper, graph, node): ...@@ -577,10 +733,10 @@ def aten_le(mapper, graph, node):
current_outputs = [output_name] current_outputs = [output_name]
# 处理输入0,即%78 # 处理输入0,即%78
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs) mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["input0"] = inputs_name[0] layer_inputs["x"] = inputs_name[0]
# 处理输入1,即%79 # 处理输入1,即%79
mapper._check_input(graph, inputs_node[1], inputs_name[1], current_outputs) mapper._check_input(graph, inputs_node[1], inputs_name[1], current_outputs)
layer_inputs["input1"] = inputs_name[1] layer_inputs["y"] = inputs_name[1]
# 获取当前节点输入的list # 获取当前节点输入的list
current_inputs = list(layer_inputs.values()) current_inputs = list(layer_inputs.values())
...@@ -613,6 +769,35 @@ def aten_len(mapper, graph, node): ...@@ -613,6 +769,35 @@ def aten_len(mapper, graph, node):
return current_inputs, current_outputs return current_inputs, current_outputs
def aten_lt(mapper, graph, node):
""" 构造对比大小的PaddleLayer。
TorchScript示例:
%80 : bool = aten::lt(%78, %79)
参数含义:
%80 (bool): 输出,第一个元素是否小于第二个元素。
%78 (-): 需对比的输入1。
%79 (-): 需对比的输入2。
"""
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name]
layer_inputs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = [output_name]
# 处理输入0,即%78
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["x"] = inputs_name[0]
# 处理输入1,即%79
mapper._check_input(graph, inputs_node[1], inputs_name[1], current_outputs)
layer_inputs["y"] = inputs_name[1]
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
graph.add_layer("prim.lt", inputs=layer_inputs, outputs=layer_outputs)
return current_inputs, current_outputs
def aten_max_pool2d(mapper, graph, node): def aten_max_pool2d(mapper, graph, node):
""" 构造最大池化的PaddleLayer。 """ 构造最大池化的PaddleLayer。
...@@ -784,6 +969,31 @@ def aten_neg(mapper, graph, node): ...@@ -784,6 +969,31 @@ def aten_neg(mapper, graph, node):
return current_inputs, current_outputs return current_inputs, current_outputs
def aten___not__(mapper, graph, node):
""" 构造对bool型取负的PaddleLayer。
TorchScript示例:
%4498 : bool = aten::__not__(%aux_defined.2)
参数含义:
%4498 (bool): 取负后结果。
%aux_defined.2 (bool): 需取负的输入。
"""
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name]
layer_inputs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = [output_name]
# 处理输入0,即%124
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["input"] = inputs_name[0]
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
graph.add_layer("prim.not", inputs=layer_inputs, outputs=layer_outputs)
return current_inputs, current_outputs
def aten_relu_(mapper, graph, node): def aten_relu_(mapper, graph, node):
""" 构造ReLU激活的PaddleLayer。 """ 构造ReLU激活的PaddleLayer。
...@@ -874,6 +1084,43 @@ def aten_reshape(mapper, graph, node): ...@@ -874,6 +1084,43 @@ def aten_reshape(mapper, graph, node):
return current_inputs, current_outputs return current_inputs, current_outputs
def aten_select(mapper, graph, node):
""" 构造选取特定维度Variable的PaddleLayer。
TorchScript示例:
%19 : Tensor = aten::select(%18, %8, %7)
参数含义:
%19 (Tensor): 输出,选取的Tensor。
%18 (Tensor): 需要选取的Tensor。
%8 (int): select的维度。
%7 (int): select的第n个向量。
"""
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name]
layer_inputs = {}
layer_attrs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = [output_name]
# 处理输入0,即%18
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["input"] = inputs_name[0]
# 处理输入1,即%8
layer_attrs["dim"] = mapper.attrs[inputs_name[1]]
# 处理输入2,即%75
mapper._check_input(graph, inputs_node[2], inputs_name[2], current_outputs)
layer_inputs["index"] = inputs_name[2]
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
graph.add_layer(
"prim.select",
inputs=layer_inputs,
outputs=current_outputs,
**layer_attrs)
return current_inputs, current_outputs
def aten_size(mapper, graph, node): def aten_size(mapper, graph, node):
""" 构造获取shape的PaddleLayer。 """ 构造获取shape的PaddleLayer。
...@@ -900,13 +1147,13 @@ def aten_size(mapper, graph, node): ...@@ -900,13 +1147,13 @@ def aten_size(mapper, graph, node):
def aten_slice(mapper, graph, node): def aten_slice(mapper, graph, node):
""" 构造切分list的PaddleLayer。 """ 构造切分list或Variable的PaddleLayer。
TorchScript示例: TorchScript示例:
%83 : int[] = aten::slice(%73, %82, %75, %77) %83 : int[] = aten::slice(%73, %82, %75, %77)
参数含义: 参数含义:
%83 (list): 输出,切分后的list。 %83 (list/Tensor): 输出,切分后的list。
%73 (list): 需要切分的list。 %73 (list/Tensor): 需要切分的list。
%82 (int): 切分的开始索引。 %82 (int): 切分的开始索引。
%75 (int): 切分的结束索引。 %75 (int): 切分的结束索引。
%77 (int): 切分的步长。 %77 (int): 切分的步长。
...@@ -993,3 +1240,79 @@ def aten_t(mapper, graph, node): ...@@ -993,3 +1240,79 @@ def aten_t(mapper, graph, node):
outputs=layer_outputs, outputs=layer_outputs,
perm=[1, 0]) perm=[1, 0])
return current_inputs, current_outputs return current_inputs, current_outputs
def aten_unsqueeze(mapper, graph, node):
""" 构造插入维度的PaddleLayer。
TorchScript示例:
%13 : Tensor = aten::unsqueeze(%12, %7)
参数含义:
%13 (Tensor): 输出,插入维度后的Tensor。
%12 (Tensor): 需要插入维度的Tensor。
%7 (int): 维度。
"""
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name]
layer_inputs = {}
layer_attrs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = [output_name]
# 处理输入0,即%13
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["input"] = inputs_name[0]
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
# 处理输入1,即%12
if inputs_name[1] in mapper.attrs:
layer_attrs["axes"] = mapper.attrs[inputs_name[1]]
else:
mapper._check_input(graph, inputs_node[1], inputs_name[1],
current_outputs)
layer_attrs["axes"] = inputs_name[1]
current_inputs.append(inputs_name[1])
graph.add_layer(
"fluid.layers.unsqueeze",
inputs=layer_inputs,
outputs=layer_outputs,
**layer_attrs)
return current_inputs, current_outputs
def aten_warn(mapper, graph, node):
""" 构造warning的PaddleLayer。
TorchScript示例:
= aten::warn(%3, %2)
参数含义:
%3 (str): warning的提示字符串。
%2 (int): warning的stacklevel。
"""
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name]
layer_inputs = {}
layer_attrs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = [output_name]
# 处理输入0,即%3
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["input"] = inputs_name[0]
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
# 处理输入1,即%2
if inputs_name[1] in mapper.attrs:
layer_attrs["stacklevel"] = mapper.attrs[inputs_name[1]]
else:
mapper._check_input(graph, inputs_node[1], inputs_name[1],
current_outputs)
layer_attrs["stacklevel"] = inputs_name[1]
current_inputs.append(inputs_name[1])
graph.add_layer(
"prim.warnings",
inputs=layer_inputs,
outputs=layer_outputs,
**layer_attrs)
return current_inputs, current_outputs
...@@ -74,9 +74,9 @@ def prim_ListConstruct(mapper, graph, node): ...@@ -74,9 +74,9 @@ def prim_ListConstruct(mapper, graph, node):
TorchScript示例: TorchScript示例:
%86 : int[] = prim::ListConstruct(%84, %85) %86 : int[] = prim::ListConstruct(%84, %85)
参数含义: 参数含义:
%86 (list): list节点输出。
%84 (int/其他): list第一个元素信息。 %84 (int/其他): list第一个元素信息。
%85 (int/其他): list第二个元素信息。 %85 (int/其他): list第二个元素信息。
%86 (list): list节点输出。
""" """
output_name = mapper._get_outputs_name(node)[0] output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name] layer_outputs = [output_name]
...@@ -247,6 +247,32 @@ def prim_min(mapper, graph, node): ...@@ -247,6 +247,32 @@ def prim_min(mapper, graph, node):
return current_inputs, current_outputs return current_inputs, current_outputs
def prim_requires_grad(mapper, graph, node):
""" 构造是否计算梯度的PaddleLayer。
TorchScript示例:
%356 : bool = prim::requires_grad(%tensor.31)
参数含义:
%356 (bool): 输出,当前Tensor是否计算梯度。
%tensor.31 (Tensor): 输入的Tensor。
"""
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name]
layer_inputs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = [output_name]
# 处理输入0,即%86
mapper._check_input(graph, inputs_node[0], inputs_name[0], current_outputs)
layer_inputs["input"] = inputs_name[0]
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
graph.add_layer(
"prim.requires_grad", inputs=layer_inputs, outputs=layer_outputs)
return current_inputs, current_outputs
def prim_SetAttr(mapper, graph, node): def prim_SetAttr(mapper, graph, node):
""" 设置attribute信息。 """ 设置attribute信息。
...@@ -297,3 +323,70 @@ def prim_shape(mapper, graph, node): ...@@ -297,3 +323,70 @@ def prim_shape(mapper, graph, node):
graph.add_layer("prim.shape", inputs=layer_inputs, outputs=layer_outputs) graph.add_layer("prim.shape", inputs=layer_inputs, outputs=layer_outputs)
return current_inputs, current_outputs return current_inputs, current_outputs
def prim_TupleConstruct(mapper, graph, node):
""" 构造tuple的PaddleLayer。
TorchScript示例:
%4492 : (Tensor, Tensor?) = prim::TupleConstruct(%x.46, %aux)
参数含义:
%4492 (tuple): 输出,tuple。
%x.46 (Tensor/其他): tuple第一个元素信息。
%aux (Tensor/其他): tuple第二个元素信息。
"""
output_name = mapper._get_outputs_name(node)[0]
layer_outputs = [output_name]
layer_inputs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = [output_name]
# 处理每个输入
for i, input_name in enumerate(inputs_name):
layer_inputs["input{}".format(i)] = input_name
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
graph.add_layer("prim.tuple", inputs=layer_inputs, outputs=layer_outputs)
return current_inputs, current_outputs
def prim_TupleUnpack(mapper, graph, node):
""" 构造获取tuple元素的PaddleLayer。
TorchScript示例:
%x.223 : Tensor, %aux.3 : Tensor? = prim::TupleUnpack(%4492)
参数含义:
%x.223 (Tensor/其他): 输出,tuple第一个元素信息。
%aux.3 (Tensor/其他): 输出,tuple第二个元素信息。
%4492 (tuple): 需要获取元素的tuple。
"""
outputs_name = mapper._get_outputs_name(node)
layer_outputs = outputs_name
layer_inputs = {}
inputs_name, inputs_node = mapper._get_inputs_name(node)
# 获取当前节点输出的list
current_outputs = outputs_name
layer_inputs["input"] = inputs_name[0]
# 获取当前节点输入的list
current_inputs = list(layer_inputs.values())
graph.add_layer(
"prim.tuple_unpack", inputs=layer_inputs, outputs=layer_outputs)
return current_inputs, current_outputs
def prim_Uninitialized(mapper, graph, node):
""" 构造表示编译器永远不会使用的值的PaddleLayer,该节点转换为None。
TorchScript示例:
%345 : bool = prim::Uninitialized()
参数含义:
%345 (bool): 输出,为赋值的bool。
"""
output_name = mapper._get_outputs_name(node)[0]
output = list(node.outputs())[0]
mapper.attrs[output_name] = None
graph.add_layer(
"prim.constant", inputs={}, outputs=[output_name], value=None)
return [], [output_name]
...@@ -56,11 +56,19 @@ class PyTorchOpMapper(OpMapper): ...@@ -56,11 +56,19 @@ class PyTorchOpMapper(OpMapper):
def traverse(self, script_graph, parent_layer=None): def traverse(self, script_graph, parent_layer=None):
# 用于获取graph的输入 # 用于获取graph的输入
def _update_graph_inputs(inputs, outputs): def _update_graph_inputs(kind, inputs, outputs):
# extend只能放更新graph_inputs之前的情况:
# 1. loop的输出i也是输入;i是输入的原因是:子图中为父图得到的。
# 2. 在_check_input中需要使用to_variable。
# extend只能放更新graph_inputs之后的情况:
# 使用了append。
if kind != "aten::append":
current_node_outputs.extend(outputs) current_node_outputs.extend(outputs)
for name in inputs: for name in inputs:
if name not in current_node_outputs: if name not in current_node_outputs:
graph_inputs.append(name) graph_inputs.append(name)
if kind == "aten::append":
current_node_outputs.extend(outputs)
# 初始化 # 初始化
graph = PaddleGraph(parent_layer) graph = PaddleGraph(parent_layer)
...@@ -80,11 +88,11 @@ class PyTorchOpMapper(OpMapper): ...@@ -80,11 +88,11 @@ class PyTorchOpMapper(OpMapper):
if hasattr(prim, func_name): if hasattr(prim, func_name):
func = getattr(prim, func_name) func = getattr(prim, func_name)
inputs, outputs = func(self, graph, node) inputs, outputs = func(self, graph, node)
_update_graph_inputs(inputs, outputs) _update_graph_inputs(kind, inputs, outputs)
elif hasattr(aten, func_name): elif hasattr(aten, func_name):
func = getattr(aten, func_name) func = getattr(aten, func_name)
inputs, outputs = func(self, graph, node) inputs, outputs = func(self, graph, node)
_update_graph_inputs(inputs, outputs) _update_graph_inputs(kind, inputs, outputs)
# 转换输出节点 # 转换输出节点
if hasattr(script_graph, 'returnNode'): if hasattr(script_graph, 'returnNode'):
...@@ -99,7 +107,7 @@ class PyTorchOpMapper(OpMapper): ...@@ -99,7 +107,7 @@ class PyTorchOpMapper(OpMapper):
uid=script_unique_id, uid=script_unique_id,
parent_layer=parent_layer, parent_layer=parent_layer,
index=i) index=i)
_update_graph_inputs(inputs, outputs) _update_graph_inputs("equal", inputs, outputs)
# 设置graph的参数 # 设置graph的参数
if isinstance(script_graph, torch._C.Graph): if isinstance(script_graph, torch._C.Graph):
graph.set_parameters(self.paddle_params) graph.set_parameters(self.paddle_params)
...@@ -190,8 +198,10 @@ class PyTorchOpMapper(OpMapper): ...@@ -190,8 +198,10 @@ class PyTorchOpMapper(OpMapper):
if parent_layer.kernel == "prim.loop": if parent_layer.kernel == "prim.loop":
control_output_id = index - 1 control_output_id = index - 1
output_node_name = parent_layer.outputs[control_output_id] output_node_name = parent_layer.outputs[control_output_id]
current_outputs = [output_node_name]
self._check_input(graph, node, input_node_name, current_outputs)
graph.add_layer( graph.add_layer(
"prim.equal", "prim.equal",
inputs={'input': input_node_name}, inputs={'input': input_node_name},
outputs=[output_node_name]) outputs=[output_node_name])
return [input_node_name], [output_node_name] return [input_node_name], current_outputs
Markdown is supported
0% .
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