使用from paddlets.models.model_loader import load
load()方法加载InformerModel报错
AttributeError: 'InformerModel' object has no attribute '_use_revin',为什么只有加载模型的时候用RevinWrapper,训练的时候好像没用到
def revin_norm(func):
@functools.wraps(func)
def wrapper(obj: BaseModel, *args, **kwargs):
"""
The core logic. The base modelout is been wrappered by the RevinWrapper.
"""
model = func(obj, *args, **kwargs)
if obj._use_revin:
logger.warning(
"Using reversible instance normalization (revin) to remove and restore the statistical information of a time-series instance"
)
model = RevinWrapper(model, obj._fit_params['target_dim'], **
obj._revin_params)
return model
使用from paddlets.models.model_loader import load
load()方法加载InformerModel报错
AttributeError: 'InformerModel' object has no attribute '_use_revin',为什么只有加载模型的时候用RevinWrapper,训练的时候好像没用到
def revin_norm(func):
@functools.wraps(func)
def wrapper(obj: BaseModel, *args, **kwargs):
"""
The core logic. The base modelout is been wrappered by the RevinWrapper.
"""
model = func(obj, *args, **kwargs)
if obj._use_revin:
logger.warning(
"Using reversible instance normalization (revin) to remove and restore the statistical information of a time-series instance"
)
model = RevinWrapper(model, obj._fit_params['target_dim'], **
obj._revin_params)
return model