def init_weight(self):
for layer in self.sublayers():
if isinstance(layer, nn.Conv2D):
param_init.normal_init(layer.weight, std=0.001)
elif isinstance(layer, (nn.BatchNorm, nn.SyncBatchNorm)):
param_init.constant_init(layer.weight, value=1.0)
param_init.constant_init(layer.bias, value=0.0)
def init_weight(self):
"""Initialize the parameters of model parts."""
for sublayer in self.sublayers():
if isinstance(sublayer, nn.Conv2D):
param_init.normal_init(sublayer.weight, std=0.001)
elif isinstance(sublayer, (nn.BatchNorm, nn.SyncBatchNorm)):
param_init.constant_init(sublayer.weight, value=1.0)
param_init.constant_init(sublayer.bias, value=0.0)
def init_weight(self):
for _param in self.parameters():
_param.optimize_attr['learning_rate'] = self.lr_multiple
for layer in self.sublayers():
if isinstance(layer, nn.LayerNorm):
param_init.constant_init(layer.weight, value=1.0)
param_init.constant_init(layer.bias, value=0.0)
def init_weight(self):
for layer in self.sublayers():
if isinstance(layer, nn.Conv2D):
param_init.normal_init(layer.weight, std=0.001)
elif isinstance(layer, (nn.BatchNorm, nn.SyncBatchNorm)):
param_init.constant_init(layer.weight, value=1.0)
param_init.constant_init(layer.bias, value=0.0)
if self.pretrained is not None:
utils.load_pretrained_model(self, self.pretrained)
def init_weight(self):
if self.pretrained is not None:
utils.load_entire_model(self, self.pretrained)
else:
for sublayer in self.sublayers():
if isinstance(sublayer, nn.Conv2D):
param_init.kaiming_normal_init(sublayer.weight)
elif isinstance(sublayer, (nn.BatchNorm, nn.SyncBatchNorm)):
param_init.constant_init(sublayer.weight, value=1.0)
param_init.constant_init(sublayer.bias, value=0.0)
def init_weight(self):
for name, _param in self.named_parameters():
if name.startswith('norms.') or name.startswith('mla.'):
continue
_param.optimize_attr['learning_rate'] = self.lr_multiple
for layer in self.sublayers():
if isinstance(layer, nn.LayerNorm):
param_init.constant_init(layer.weight, value=1.0)
param_init.constant_init(layer.bias, value=0.0)
def init_params(self):
for m in self.sublayers():
if isinstance(m, nn.Conv2D):
param_init.kaiming_normal_init(m.weight)
if m.bias is not None:
param_init.constant_init(m.bias, value=0.0)
elif isinstance(m, nn.BatchNorm2D):
param_init.constant_init(m.weight, value=1.0)
param_init.constant_init(m.bias, value=0.0)
elif isinstance(m, nn.Linear):
param_init.normal_init(m.weight, std=0.001)
if m.bias is not None:
param_init.constant_init(m.bias, value=0.0)
