TimeXer cannot load after saving

[alemkakti]

What happened + What you expected to happen

I trained some models in neuralforecast library and I save them each time with nf.save. Everything works with nf.load, but then I got an issue with TimeXer, where it saved fine, but it couldn't load. I got this error:
TypeError: neuralforecast.common._base_model.BaseModel.__init__() got multiple values for keyword argument 'exclude_insample_y'

I tried other models on AirPassengers example and they work there, but this one fails. So, there is something wrong in way of saving and loading TimeXer. What exactly is it, I don't really know, but I would like to learn how to save and load it correctly.

Versions / Dependencies

I am using Google Cloud notebook on Vertex AI.

Reproduction script

`from neuralforecast import NeuralForecast
from neuralforecast.models import TimeXer
from neuralforecast.losses.pytorch import MSE, MAE
from neuralforecast.utils import AirPassengersPanel, AirPassengersStatic, augment_calendar_df

AirPassengersPanel, calendar_cols = augment_calendar_df(df=AirPassengersPanel, freq='M')

Y_train_df = AirPassengersPanel[AirPassengersPanel.ds<AirPassengersPanel['ds'].values[-12]] # 132 train
Y_test_df = AirPassengersPanel[AirPassengersPanel.ds>=AirPassengersPanel['ds'].values[-12]].reset_index(drop=True) # 12 test

model = TimeXer(h=12,
input_size=24,
n_series=2,
futr_exog_list=["trend", "month"],
patch_len=12,
hidden_size=128,
n_heads=16,
e_layers=2,
d_ff=256,
factor=1,
dropout=0.1,
use_norm=True,
loss=MSE(),
valid_loss=MAE(),
early_stop_patience_steps=3,
batch_size=32)

fcst = NeuralForecast(models=[model], freq='ME')
fcst.fit(df=Y_train_df, static_df=AirPassengersStatic, val_size=12)

Define save path

save_path = f"airpassengers_2/"
os.makedirs(save_path)

Save the trained model and configuration

fcst.save(path=save_path, save_dataset=True, overwrite=True, model_index=None) # this works
print(f"Model saved to {save_path}")

nf = NeuralForecast.load(save_path) # this fails with previous error message
`

Issue Severity

High: It blocks me from completing my task.

[marcopeix]

Thanks for reporting it! I was able to reproduce it, and found a fix! Will patch this soon.

[alemkakti]

@marcopeix Sounds great, but will it be able to work with installing through pip?

[marcopeix]

To access the latest changes, you have to upgrade the package through pip. If you want to access the change before the next release, then you can pip install from the main branch.