Deeptable model can not work after loading model

[BruceQD]

As shown in the figure, the model can successfully predict X_valid. However, after loading the model, the model does not have the predict_proba function, and using the predict function for prediction also results in an error.

The error message for newmodel.predict(X_valid) is as follows:

KeyError Traceback (most recent call last)
Cell In[19], line 1
----> 1 newmodel.predict(X_valid)

File /opt/conda/lib/python3.10/site-packages/deeptables/models/deepmodel.py:127, in DeepModel.predict(self, X, batch_size, verbose)
126 def predict(self, X, batch_size=128, verbose=0):
--> 127 return self.__predict(self.model, X, batch_size=batch_size, verbose=verbose)

File /opt/conda/lib/python3.10/site-packages/deeptables/models/deepmodel.py:131, in DeepModel.__predict(self, model, X, batch_size, verbose)
129 def __predict(self, model, X, batch_size=128, verbose=0):
130 logger.info("Performing predictions...")
--> 131 ds = self.__get_prediction_data(X, batch_size=batch_size)
132 steps = math.ceil(len(X) / batch_size)
133 return model.predict(ds, steps=steps, verbose=verbose)

File /opt/conda/lib/python3.10/site-packages/deeptables/models/deepmodel.py:207, in DeepModel.__get_prediction_data(self, X, y, batch_size)
206 def __get_prediction_data(self, X, y=None, *, batch_size):
--> 207 ds = to_dataset(self.config, self.task, self.num_classes, X, y,
208 batch_size=batch_size, shuffle=False, drop_remainder=False,
209 categorical_columns=self.categorical_columns,
210 continuous_columns=self.continuous_columns,
211 var_len_categorical_columns=self.var_len_categorical_columns)
213 return ds

File /opt/conda/lib/python3.10/site-packages/deeptables/utils/dataset_generator.py:246, in to_dataset(config, task, num_classes, X, y, batch_size, shuffle, drop_remainder, categorical_columns, continuous_columns, var_len_categorical_columns)
240 logger.info(f'create dataset generator with {cls.name}, '
241 f'batch_size={batch_size}, shuffle={shuffle}, drop_remainder={drop_remainder}')
243 dg = cls(config, task, num_classes, categorical_columns=categorical_columns,
244 continuous_columns=continuous_columns,
245 var_len_categorical_columns=var_len_categorical_columns)
--> 246 ds = dg(X, y, batch_size=batch_size, shuffle=shuffle, drop_remainder=drop_remainder)
247 return ds

File /opt/conda/lib/python3.10/site-packages/deeptables/utils/dataset_generator.py:44, in _TFDGForPandas.call(self, X, y, batch_size, shuffle, drop_remainder)
41 # add categorical data
42 if self.categorical_columns is not None and len(self.categorical_columns) > 0:
43 train_data['input_categorical_vars_all'] =
---> 44 X[[c.name for c in self.categorical_columns]].values.astype(consts.DATATYPE_TENSOR_FLOAT)
46 # add continuous data
47 if self.continuous_columns is not None and len(self.continuous_columns) > 0:

File /opt/conda/lib/python3.10/site-packages/pandas/core/frame.py:3899, in DataFrame.getitem(self, key)
3897 if is_iterator(key):
3898 key = list(key)
-> 3899 indexer = self.columns._get_indexer_strict(key, "columns")[1]
3901 # take() does not accept boolean indexers
3902 if getattr(indexer, "dtype", None) == bool:

File /opt/conda/lib/python3.10/site-packages/pandas/core/indexes/base.py:6114, in Index._get_indexer_strict(self, key, axis_name)
6111 else:
6112 keyarr, indexer, new_indexer = self._reindex_non_unique(keyarr)
-> 6114 self._raise_if_missing(keyarr, indexer, axis_name)
6116 keyarr = self.take(indexer)
6117 if isinstance(key, Index):
6118 # GH 42790 - Preserve name from an Index

File /opt/conda/lib/python3.10/site-packages/pandas/core/indexes/base.py:6178, in Index._raise_if_missing(self, key, indexer, axis_name)
6175 raise KeyError(f"None of [{key}] are in the [{axis_name}]")
6177 not_found = list(ensure_index(key)[missing_mask.nonzero()[0]].unique())
-> 6178 raise KeyError(f"{not_found} not in index")

KeyError: "

[oaksharks]

@BruceQD Please refer to this example: https://github.com/DataCanvasIO/DeepTables/blob/master/deeptables/examples/quickstart.ipynb

from hypernets.utils import fs

model_dir = "model_dir"
dt.save(model_dir)

print(f"Absolute path for the persisted model: {fs.remote_root_}/{model_dir}")

model_load = deeptable.DeepTable.load(model_dir)  # KEYPOINT
model_load.evaluate(df_test,y_test, batch_size=512, verbose=0)