How to predict on unseen data using pymc_marketing predict function

Question

I am trying to predict on test data using pymc_markeing delayedsaturatedmmm model and using the below code :

import datetime
from sklearn.model_selection import train_test_split
sampler_config= {"progressbar": True}


# Define Model parameters
X = df_mmm.drop(["A","Date"],axis=1)
y = df_mmm["A"]

# Another way to split
X_train, X_test, y_train, y_test = train_test_split(X,y,test_size=0.2, shuffle=False)

train_df = pd.DataFrame(X_train,y_train)


mmm = DelayedSaturatedMMM(
    model_config = mynew_model_config,#dummy_model.default_model_config,
    sampler_config = sampler_config,
    data=train_df,
    target_column='Joins',
    date_column='Date',
    validate_data = True,
    channel_columns=['Audio_s','Cable_s','CTV_s','Display_s','META_s','Radio_s','Snapchat_s','TikTok_s','TV_s','YouTube_s'],
    control_columns = ['google_trends'],
    adstock_max_lag = 1,
    yearly_seasonality = 2
)
mmm.model_config

The model builds allright without any error etc. but when I try to predict using X_test I get an error which says coord values need to be updated. Not sure what that means but I know that the data_setter method in pymc is the one where the error originated. How can I predict using the pymc_marketing library!!