def _opt_raw_graph(self, raw_graph):
graph_searcher = GraphSearcher(raw_graph)
# shufflenet
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=["ListUnpack"]),
PatternType(pattern=["TupleUnpack"])
])
OptPass.unpack_ListUnpack_op(self, raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["slice"])])
OptPass.slice_to_strided_slice(self, raw_graph, node_sets)
# yolo_v3
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["select", "copy_"])])
OptPass.select_to_slice_inplace_copy(self, raw_graph, node_sets)
# 3d pointpillar
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["strided_slice", "index_put_"])])
OptPass.stride_slice_to_index_inplace_put(self, raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["strided_slice", "copy_"])])
OptPass.create_stride_slice_inplace_copy(self, raw_graph, node_sets)
# nd(>2) linear (JIRA 2646)
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=["matmul", "add"]),
PatternType(pattern=["matmul", "add_"])
])
OptPass.merge_matmul_with_add(self, raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=["ListConstruct"]),
PatternType(pattern=["TupleConstruct"])
])
OptPass.pack_ListConstruct_op(self, raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["embedding_bag"])])
OptPass.strip_reduantant_tensors_in_embedding_bag(
self, raw_graph, node_sets)
return raw_graph
def collect_layer_act_pair(self):
graph_searcher = GraphSearcher(self.graph)
patterns = []
tuning_ops = [NNDCT_OP.CONV2D, NNDCT_OP.DEPTHWISE_CONV2D, NNDCT_OP.CONVTRANSPOSE2D,
NNDCT_OP.CONV3D, NNDCT_OP.DEPTHWISE_CONV3D, NNDCT_OP.CONVTRANSPOSE3D]
tail_act_ops = [NNDCT_OP.RELU, NNDCT_OP.RELU6, NNDCT_OP.HSWISH, NNDCT_OP.HSIGMOID]
for tuning_op in tuning_ops:
for act_op in tail_act_ops:
patterns.append(PatternType(pattern=[tuning_op, act_op]))
node_sets = graph_searcher.find_nodes_from_type(patterns)
layer_act_group = {}
for _, node_list in node_sets.items():
for nodeset in node_list:
conv, act = nodeset
layer_act_group[conv] = act
return layer_act_group
def FuseBnToConv(self):
# find fusable bathnorm node
fuse_bn_handler = ConvBnHandler()
graph_searcher = GraphSearcher(self._graph)
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=[NNDCT_OP.CONV2D, NNDCT_OP.BATCH_NORM],
action=fuse_bn_handler),
PatternType(
pattern=[NNDCT_OP.DEPTHWISE_CONV2D, NNDCT_OP.BATCH_NORM],
action=fuse_bn_handler),
PatternType(
pattern=[NNDCT_OP.CONVTRANSPOSE2D, NNDCT_OP.BATCH_NORM],
action=fuse_bn_handler)
])
for id, node_list in node_sets.items():
for nodeset in node_list:
_, bn_node = nodeset
self._graph.remove_node(bn_node)
def _execute_optimize(self, raw_graph):
graph_searcher = GraphSearcher(raw_graph)
# classification
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["t", "addmm"])])
OptPass.merge_param_transpose_with_addmm(raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["TupleConstruct", "TupleUnpack"])])
OptPass.penetrate_pack_unpack(raw_graph, node_sets)
# shufflenet
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=["ListUnpack"]),
PatternType(pattern=["TupleUnpack"])
])
OptPass.unpack_ListUnpack_op(raw_graph, node_sets)
# node_sets = graph_searcher.find_nodes_from_type([PatternType(pattern=["slice"])])
# OptPass.slice_to_strided_slice(raw_graph, node_sets)
# yolo_v3
# node_sets = graph_searcher.find_nodes_from_type([PatternType(pattern=["select", "copy_"])])
# OptPass.select_to_slice_inplace_copy(raw_graph, node_sets)
# 3d pointpillar
# node_sets = graph_searcher.find_nodes_from_type([PatternType(pattern=["strided_slice", "index_put_"])])
# OptPass.stride_slice_to_index_inplace_put(raw_graph, node_sets)
# node_sets = graph_searcher.find_nodes_from_type([PatternType(pattern=["strided_slice", "copy_"])])
# OptPass.create_stride_slice_inplace_copy(raw_graph, node_sets)
# nd(>2) linear (JIRA 2646)
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=["matmul", "add"]),
PatternType(pattern=["matmul", "add_"])
])
OptPass.merge_matmul_with_add(raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=["ListConstruct"]),
PatternType(pattern=["TupleConstruct"])
])
OptPass.pack_ListConstruct_op(raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["embedding_bag"])])
OptPass.strip_reduantant_tensors_in_embedding_bag(raw_graph, node_sets)
return raw_graph
def finetune(self, run_fn, run_args):
if self.quantizer.quant_mode == 2:
NndctScreenLogger().warning(
f"Finetune function will be ignored in test mode!")
return
NndctScreenLogger().info(
f"=>Finetuning module parameters for better quantization accuracy... "
)
# backup option value
opt_bak_param_corr = NndctOption.nndct_param_corr.value
set_option_value("nndct_param_corr", 0)
# cache input and output
#print("**** cache input and output")
last_quant_nodes = self.collect_last_quant_nodes()
with torch.no_grad():
hook_mods = []
for node in self.graph.nodes:
if node.op.type == NNDCT_OP.INPUT or \
node in last_quant_nodes:
# (self.quantizer.configer.is_node_quantizable(node, False) and
# len(node.op.params) > 0):
hook_mods.append(node.module)
handlers = self.hook_cache_output(hook_mods)
set_option_value("nndct_quant_off", True)
run_fn(*run_args)
self.clean_hooks(handlers)
# for mod in self.quant_model.modules():
# if hasattr(mod, "node") and mod.node.op.type in [NNDCT_OP.DENSE, NNDCT_OP.CONV2D, NNDCT_OP.DEPTHWISE_CONV2D, NNDCT_OP.CONVTRANSPOSE2D]:
# self._float_weights[mod.node].append(mod.weight.detach().cpu())
torch.cuda.empty_cache()
# calibration to get a set of quantization steps
#print("****calibration to get float model tensor values")
for mod in self.quant_model.modules():
if hasattr(mod, "param_quantized"):
setattr(mod, "param_quantized", False)
# evaluation to get float model tensors
set_option_value("nndct_quant_off", False)
with torch.no_grad():
run_fn(*run_args)
torch.cuda.empty_cache()
#print("****Parameter finetuning")
device = GLOBAL_MAP.get_ele(NNDCT_KEYS.QUANT_DEVICE)
graph_searcher = GraphSearcher(self.graph)
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=[NNDCT_OP.CONV2D, NNDCT_OP.RELU]),
PatternType(pattern=[NNDCT_OP.CONV2D, NNDCT_OP.RELU6]),
PatternType(pattern=[NNDCT_OP.DEPTHWISE_CONV2D, NNDCT_OP.RELU]),
PatternType(pattern=[NNDCT_OP.DEPTHWISE_CONV2D, NNDCT_OP.RELU6]),
PatternType(pattern=[NNDCT_OP.CONVTRANSPOSE2D, NNDCT_OP.RELU])
])
layer_act_group = {}
for _, node_list in node_sets.items():
for nodeset in node_list:
conv, act = nodeset
layer_act_group[conv] = act
# to avoid quantization steps change among parameter finetuning
self.quantizer.quant_mode = 2
net_inputs = []
for node in self.input_nodes:
cached_net_input = [
out for out in self.cached_outputs[node.module]
]
net_inputs.append(cached_net_input)
# last_quant_nodes = self.collect_last_quant_nodes()
last_quant_mods = [node.module for node in last_quant_nodes]
handlers = self.hook_cache_output(last_quant_mods, hook_type="single")
net_loss = self.eval_loss(net_inputs, last_quant_mods, device)
self.clean_hooks(handlers)
# model.clean_hooks()
torch.cuda.empty_cache()
finetune_group = {}
# hook_mods = []
for qmod, fmod in zip(self._quant_model.modules(),
self._float_model.modules()):
if hasattr(qmod, "node"):
if (self.quantizer.configer.is_node_quantizable(
qmod.node, False) and len(qmod.node.op.params) > 0):
finetune_group[qmod.node] = [qmod, fmod]
# hook_mods.append(fmod)
# self.hook_cache_output(hook_mods, hook_type="single")
for node, module_pair in finetune_group.items():
# if self.quantizer.configer.is_node_quantizable(node, False) and \
# len(node.op.params) > 0:
quant_layer, float_layer = module_pair
pn_node = self.graph.parents(node)[0]
handlers = self.hook_cache_output([pn_node.module],
hook_type="single")
layer_inputs = []
with torch.no_grad():
for input_args in zip(*net_inputs):
new_input_args = []
for ip in input_args:
if isinstance(ip, torch.Tensor):
new_input_args.append(ip.to(device))
_ = self.quant_model(*new_input_args)
layer_inputs.append(
self.cached_output[pn_node.module].detach().cpu())
self.clean_hooks(handlers)
del self.cached_output[pn_node.module]
#print(f"Tuning {node.name}")
net_loss = self.optimize_layer(node, float_layer, layer_inputs,
layer_act_group, net_inputs,
net_loss, last_quant_mods, device)
del layer_inputs
torch.cuda.empty_cache()
# recover quantizer status
for node in self.graph.nodes:
for _, fp_history in self.quantizer.fp_history.items():
if node.name in fp_history:
fp_history[node.name].clear()
for mod in self.quant_model.modules():
if hasattr(mod, "param_quantized"):
setattr(mod, "param_quantized", False)
for mod in self.quant_model.modules():
if hasattr(mod, "param_saved"):
setattr(mod, "param_saved", False)
self.quantizer.quant_mode = 1
set_option_value("nndct_param_corr", opt_bak_param_corr)
# export finetuned parameters
self.quantizer.export_param()
def FuseBnToConv(self):
# find fusable bathnorm node
fuse_bn_handler = ConvBnHandler()
graph_searcher = GraphSearcher(self._graph)
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=[NNDCT_OP.CONV2D, NNDCT_OP.BATCH_NORM],
action=fuse_bn_handler),
PatternType(
pattern=[NNDCT_OP.DEPTHWISE_CONV2D, NNDCT_OP.BATCH_NORM],
action=fuse_bn_handler),
PatternType(
pattern=[NNDCT_OP.CONVTRANSPOSE2D, NNDCT_OP.BATCH_NORM],
action=fuse_bn_handler),
PatternType(pattern=[
NNDCT_OP.DEPTHWISE_CONVTRANSPOSE2D, NNDCT_OP.BATCH_NORM
],
action=fuse_bn_handler),
PatternType(pattern=[NNDCT_OP.CONV3D, NNDCT_OP.BATCH_NORM],
action=fuse_bn_handler),
PatternType(
pattern=[NNDCT_OP.DEPTHWISE_CONV3D, NNDCT_OP.BATCH_NORM],
action=fuse_bn_handler),
PatternType(
pattern=[NNDCT_OP.CONVTRANSPOSE3D, NNDCT_OP.BATCH_NORM],
action=fuse_bn_handler),
PatternType(pattern=[
NNDCT_OP.DEPTHWISE_CONVTRANSPOSE3D, NNDCT_OP.BATCH_NORM
],
action=fuse_bn_handler),
PatternType(pattern=[
NNDCT_OP.CONV2D, NNDCT_OP.CONCAT, NNDCT_OP.BATCH_NORM
],
action=fuse_bn_handler),
PatternType(pattern=[
NNDCT_OP.DEPTHWISE_CONV2D, NNDCT_OP.CONCAT, NNDCT_OP.BATCH_NORM
],
action=fuse_bn_handler),
PatternType(pattern=[
NNDCT_OP.CONVTRANSPOSE2D, NNDCT_OP.CONCAT, NNDCT_OP.BATCH_NORM
],
action=fuse_bn_handler),
PatternType(pattern=[
NNDCT_OP.DEPTHWISE_CONVTRANSPOSE2D, NNDCT_OP.CONCAT,
NNDCT_OP.BATCH_NORM
],
action=fuse_bn_handler),
PatternType(pattern=[
NNDCT_OP.CONV3D, NNDCT_OP.CONCAT, NNDCT_OP.BATCH_NORM
],
action=fuse_bn_handler),
PatternType(pattern=[
NNDCT_OP.DEPTHWISE_CONV3D, NNDCT_OP.CONCAT, NNDCT_OP.BATCH_NORM
],
action=fuse_bn_handler),
PatternType(pattern=[
NNDCT_OP.CONVTRANSPOSE3D, NNDCT_OP.CONCAT, NNDCT_OP.BATCH_NORM
],
action=fuse_bn_handler),
PatternType(pattern=[
NNDCT_OP.DEPTHWISE_CONVTRANSPOSE3D, NNDCT_OP.CONCAT,
NNDCT_OP.BATCH_NORM
],
action=fuse_bn_handler),
])
removed_bn = set()
for id, node_list in node_sets.items():
for nodeset in node_list:
bn_node = nodeset[-1]
if bn_node.merged and bn_node not in removed_bn:
self._graph.remove_node(bn_node)
removed_bn.add(bn_node)
def finetune(self, run_fn, run_args):
if self.quantizer.quant_mode == 2:
NndctScreenLogger().warning(
f"Finetune function will be ignored in test mode!")
return
NndctScreenLogger().info(
f"=>Preparing data for fast finetuning module parameters ...")
# memory status
total_m, *_, available_m = list(
map(lambda x: x / 1024,
map(int,
os.popen('free -t -m').readlines()[1].split()[1:])))
NndctScreenLogger().info(
f"Mem status(total mem: {total_m:.2f}G, available mem: {available_m:.2f}G)."
)
NndctScreenLogger().info(
f"=>Preparing data for fast finetuning module parameters ...")
# backup option value
opt_bak_param_corr = NndctOption.nndct_param_corr.value
set_option_value("nndct_param_corr", 0)
# cache input and output
#print("**** cache input and output")
last_quant_nodes = self.collect_last_quant_nodes()
with torch.no_grad():
cache_layers = []
monitor_layers = []
for node in self.graph.nodes:
if node.op.type == NNDCT_OP.INPUT or node in last_quant_nodes:
cache_layers.append(node.module)
elif self.quantizer.configer.is_conv_like(node):
monitor_layers.append(node.module)
monitor_handlers = self.hook_memory_monitor(monitor_layers)
cache_handlers = self.hook_cache_output(cache_layers,
monitor_mem=True)
set_option_value("nndct_quant_off", True)
run_fn(*run_args)
# memory statistics
total_memory_cost = 0.0
for layer in cache_layers:
total_memory_cost += self._mem_count[layer]
del self._mem_count[layer]
total_memory_cost += 2 * max(self._mem_count.values())
self.clean_hooks(monitor_handlers + cache_handlers)
torch.cuda.empty_cache()
NndctScreenLogger().info(
f"Mem cost by fast finetuning: {total_memory_cost:.2f}G.")
if total_memory_cost > 0.8 * available_m:
NndctScreenLogger().warning(
f"There is not enought memory for fast finetuning and this process will be ignored!.Try to use a smaller calibration dataset."
)
return
# calibration to get a set of quantization steps
#print("****calibration to get float model tensor values")
for mod in self.quant_model.modules():
if hasattr(mod, "param_quantized"):
setattr(mod, "param_quantized", False)
# evaluation to get float model tensors
set_option_value("nndct_quant_off", False)
with torch.no_grad():
run_fn(*run_args)
torch.cuda.empty_cache()
#print("****Parameter finetuning")
NndctScreenLogger().info(
f"=>Fast finetuning module parameters for better quantization accuracy..."
)
device = GLOBAL_MAP.get_ele(NNDCT_KEYS.QUANT_DEVICE)
graph_searcher = GraphSearcher(self.graph)
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=[NNDCT_OP.CONV2D, NNDCT_OP.HSWISH]),
PatternType(pattern=[NNDCT_OP.CONV2D, NNDCT_OP.HSIGMOID]),
PatternType(pattern=[NNDCT_OP.CONV2D, NNDCT_OP.RELU]),
PatternType(pattern=[NNDCT_OP.CONV2D, NNDCT_OP.RELU6]),
PatternType(pattern=[NNDCT_OP.DEPTHWISE_CONV2D, NNDCT_OP.HSWISH]),
PatternType(
pattern=[NNDCT_OP.DEPTHWISE_CONV2D, NNDCT_OP.HSIGMOID]),
PatternType(pattern=[NNDCT_OP.DEPTHWISE_CONV2D, NNDCT_OP.RELU]),
PatternType(pattern=[NNDCT_OP.DEPTHWISE_CONV2D, NNDCT_OP.RELU6]),
PatternType(pattern=[NNDCT_OP.CONVTRANSPOSE2D, NNDCT_OP.HSWISH]),
PatternType(pattern=[NNDCT_OP.CONVTRANSPOSE2D, NNDCT_OP.HSIGMOID]),
PatternType(pattern=[NNDCT_OP.CONVTRANSPOSE2D, NNDCT_OP.RELU]),
PatternType(pattern=[NNDCT_OP.CONVTRANSPOSE2D, NNDCT_OP.RELU6]),
PatternType(pattern=[NNDCT_OP.CONV3D, NNDCT_OP.HSWISH]),
PatternType(pattern=[NNDCT_OP.CONV3D, NNDCT_OP.HSIGMOID]),
PatternType(pattern=[NNDCT_OP.CONV3D, NNDCT_OP.RELU]),
PatternType(pattern=[NNDCT_OP.CONV3D, NNDCT_OP.RELU6]),
PatternType(pattern=[NNDCT_OP.DEPTHWISE_CONV3D, NNDCT_OP.HSWISH]),
PatternType(
pattern=[NNDCT_OP.DEPTHWISE_CONV3D, NNDCT_OP.HSIGMOID]),
PatternType(pattern=[NNDCT_OP.DEPTHWISE_CONV3D, NNDCT_OP.RELU]),
PatternType(pattern=[NNDCT_OP.DEPTHWISE_CONV3D, NNDCT_OP.RELU6]),
PatternType(pattern=[NNDCT_OP.CONVTRANSPOSE3D, NNDCT_OP.HSWISH]),
PatternType(pattern=[NNDCT_OP.CONVTRANSPOSE3D, NNDCT_OP.HSIGMOID]),
PatternType(pattern=[NNDCT_OP.CONVTRANSPOSE3D, NNDCT_OP.RELU]),
PatternType(pattern=[NNDCT_OP.CONVTRANSPOSE3D, NNDCT_OP.RELU6]),
])
layer_act_group = {}
for _, node_list in node_sets.items():
for nodeset in node_list:
conv, act = nodeset
layer_act_group[conv] = act
# to avoid quantization steps change among parameter finetuning
self.quantizer.quant_mode = 2
net_inputs = []
for node in self.input_nodes:
cached_net_input = [
out for out in self.cached_outputs[node.module]
]
net_inputs.append(cached_net_input)
# last_quant_nodes = self.collect_last_quant_nodes()
last_quant_mods = [node.module for node in last_quant_nodes]
handlers = self.hook_cache_output(last_quant_mods, hook_type="single")
net_loss = self.eval_loss(net_inputs, last_quant_mods, device)
self.clean_hooks(handlers)
# model.clean_hooks()
torch.cuda.empty_cache()
finetune_group = {}
# hook_mods = []
for qmod, fmod in zip(self._quant_model.modules(),
self._float_model.modules()):
if hasattr(qmod, "node"):
if (self.quantizer.configer.is_node_quantizable(
qmod.node, False) and len(qmod.node.op.params) > 0):
finetune_group[qmod.node] = [qmod, fmod]
# hook_mods.append(fmod)
# self.hook_cache_output(hook_mods, hook_type="single")
#for node, module_pair in finetune_group.items():
for idx, (node,
module_pair) in tqdm(enumerate(finetune_group.items()),
total=len(finetune_group.items())):
# if self.quantizer.configer.is_node_quantizable(node, False) and \
# len(node.op.params) > 0:
quant_layer, float_layer = module_pair
pn_node = self.graph.parents(node)[0]
handlers = self.hook_cache_output([pn_node.module],
hook_type="single")
layer_inputs = []
with torch.no_grad():
for input_args in zip(*net_inputs):
new_input_args = []
for ip in input_args:
if isinstance(ip, torch.Tensor):
new_input_args.append(ip.to(device))
_ = self.quant_model(*new_input_args)
layer_inputs.append(
self.cached_output[pn_node.module].detach().cpu())
self.clean_hooks(handlers)
del self.cached_output[pn_node.module]
#print(f"Tuning {node.name}")
net_loss = self.optimize_layer(node, float_layer, layer_inputs,
layer_act_group, net_inputs,
net_loss, last_quant_mods, device)
# print(f"{node.name}:{net_loss}")
del layer_inputs
torch.cuda.empty_cache()
# recover quantizer status
for node in self.graph.nodes:
for _, config_history in self.quantizer.config_history.items():
if node.name in config_history:
config_history[node.name].clear()
for mod in self.quant_model.modules():
if hasattr(mod, "param_quantized"):
setattr(mod, "param_quantized", False)
for mod in self.quant_model.modules():
if hasattr(mod, "param_saved"):
setattr(mod, "param_saved", False)
self.quantizer.quant_mode = 1
set_option_value("nndct_param_corr", opt_bak_param_corr)
NndctScreenLogger().info(f"=>Export fast finetuned parameters ...")
# export finetuned parameters
self.quantizer.export_param()
def _opt_raw_graph(self, raw_graph):
graph_searcher = GraphSearcher(raw_graph)
# torch 1.8.x will generate redundant type_as op before element-wise add.
if get_torch_version() >= 180 and get_torch_version() < 190:
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["type_as", "add"])])
OptPass.merge_internal_type_as(raw_graph, node_sets)
# classification
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=["t", "addmm"]),
PatternType(pattern=["t", "matmul"])
])
OptPass.merge_param_transpose_with_addmm(raw_graph, node_sets)
# shufflenet
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=["ListUnpack"]),
PatternType(pattern=["TupleUnpack"])
])
OptPass.unpack_ListUnpack_op(raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["slice"])])
OptPass.slice_to_strided_slice(raw_graph, node_sets)
# yolo_v3
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["select", "copy_"])])
OptPass.select_to_slice_inplace_copy(raw_graph, node_sets)
# FADnet
while True:
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["select", "strided_slice"])])
if not OptPass.merge_select_to_strided_slice(raw_graph, node_sets):
break
# FADnet
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["strided_slice", "strided_slice"])])
OptPass.merge_consecutive_strided_slice(raw_graph, node_sets)
# 3d pointpillar
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["strided_slice", "index_put_"])])
OptPass.stride_slice_to_index_inplace_put(raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["strided_slice", "copy_"])])
OptPass.create_stride_slice_inplace_copy(raw_graph, node_sets)
# nd(>2) linear (JIRA 2646)
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=["matmul", "add"]),
PatternType(pattern=["matmul", "add_"])
])
OptPass.merge_matmul_with_add(raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=["ListConstruct"]),
PatternType(pattern=["TupleConstruct"])
])
OptPass.pack_ListConstruct_op(raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["embedding_bag"])])
OptPass.strip_reduantant_tensors_in_embedding_bag(raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["empty", "zero_"])])
OptPass.merge_empty_with_zero(raw_graph, node_sets)
# delete node should be done after merge stride_slice
# delete reduantant view FADnet.
node_sets = graph_searcher.find_nodes_from_type(
[PatternType(pattern=["view"])])
OptPass.remove_reduantant_view(raw_graph, node_sets)
node_sets = graph_searcher.find_nodes_from_type([
PatternType(pattern=["add"]),
PatternType(pattern=["sub"]),
PatternType(pattern=["mul"]),
PatternType(pattern=["div"])
])
OptPass.transform_const_scalar_to_const_tensor(raw_graph, node_sets)
return raw_graph