Merge pull request #58 from yokomotod/tdidf-back-compat

revert dummy tdidf.py for pickle backward compatibility

redshells/app/word_item_similarity/dimension_reduction_model.py

@@ -12,6 +12,7 @@ class DimensionReductionModel(object):
     """ Reduce the dimension of vector values with respect to its importance.
     The importance is calculated by sum of squared values.
     """
+
     def __init__(self, dimension_size: int) -> None:
         self.dimension_size = dimension_size
         self.top_n_indices = None

redshells/contrib/model/early_stopping.py

@@ -11,6 +11,7 @@
 
 
 class EarlyStopping(object):
+
     def __init__(self, try_count=1, learning_rate=0., decay_speed=2.0, threshold=0.001, save_directory: str = None):
         self._save_path = os.path.join(save_directory, 'model.ckpt') if save_directory else None
         self._try_count = try_count

redshells/contrib/model/factorization_machine.py

@@ -18,6 +18,7 @@
 
 
 class FactorizationMachineGraph(object):
+
     def __init__(self, input_size: int, feature_kind_size: int, embedding_size: int, l2_weight: float, learning_rate: float, scope_name: str = '') -> None:
 
         with tf.variable_scope(scope_name, reuse=tf.AUTO_REUSE):
@@ -59,6 +60,7 @@ class FactorizationMachine(sklearn.base.BaseEstimator):
     
     For details of the algorithm, see "Factorization Machines" which is available at https://www.csie.ntu.edu.tw/~b97053/paper/Rendle2010FM.pdf
     """
+
     def __init__(self,
                  embedding_size: int,
                  l2_weight: float,

redshells/contrib/model/feature_aggregation_similarity_model.py

@@ -14,6 +14,7 @@
 
 
 class FeatureAggregationSimilarityDataset(tf.keras.utils.Sequence):
+
     def __init__(self,
                  x_item_indices: np.ndarray,
                  y_item_indices: np.ndarray,
@@ -71,6 +72,7 @@ def __len__(self):
 
 
 class Average(tf.keras.layers.Layer):
+
     def __init__(self, **kwargs):
         super(Average, self).__init__(**kwargs)
 
@@ -94,6 +96,7 @@ def from_config(cls, config):
 
 
 class Clip(tf.keras.layers.Layer):
+
     def __init__(self, **kwargs):
         super(Clip, self).__init__(**kwargs)
 
@@ -116,6 +119,7 @@ def from_config(cls, config):
 
 
 class FeatureAggregationSimilarityGraph(object):
+
     def __init__(self,
                  feature_size: int,
                  embedding_size: int,
@@ -165,6 +169,7 @@ def clip(x):
 
 
 class FeatureAggregationSimilarityModel(object):
+
     def __init__(
         self,
         embedding_size: int,

redshells/contrib/model/gcmc_dataset.py

@@ -12,6 +12,7 @@
 
 
 class GcmcDataset(object):
+
     def __init__(self,
                  user_ids: np.ndarray,
                  item_ids: np.ndarray,
@@ -28,6 +29,7 @@ def __init__(self,
         self.item_features = item_features
 
     def filter(self, user_ids: Set, item_ids: Set, ratings: Set):
+
         def _is_valid(u, i, r):
             return u in user_ids and i in item_ids and r in ratings
 
@@ -39,6 +41,7 @@ def _is_valid(u, i, r):
 
 
 class GcmcIdMap(object):
+
     def __init__(self,
                  ids: np.ndarray,
                  features: Optional[List[Dict[Any, np.ndarray]]] = None,
@@ -150,6 +153,7 @@ def _sort_features(cls, features: List[Dict[Any, np.ndarray]], order_map: Dict)
 
 
 class GcmcGraphDataset(object):
+
     def __init__(self, dataset: GcmcDataset, test_size: float, min_user_click_count: int = 0, max_user_click_count: int = sys.maxsize) -> None:
         self._user = GcmcIdMap(dataset.user_ids, features=dataset.user_features, min_count=min_user_click_count, max_count=max_user_click_count)
         self._item = GcmcIdMap(dataset.item_ids, features=dataset.item_features)

redshells/contrib/model/graph_convolutional_matrix_completion.py

@@ -32,6 +32,7 @@ def _convert_sparse_matrix_to_sparse_tensor(x):
 
 
 class GraphConvolutionalMatrixCompletionGraph(object):
+
     def __init__(self,
                  n_rating: int,
                  n_user: int,
@@ -216,6 +217,7 @@ def _to_constant(x):
 
 
 class GraphConvolutionalMatrixCompletion(object):
+
     def __init__(self,
                  graph_dataset: GcmcGraphDataset,
                  encoder_hidden_size: int,

redshells/contrib/model/matrix_factorization_model.py

@@ -16,6 +16,7 @@
 
 
 class MatrixFactorizationGraph(object):
+
     def __init__(self, n_items: int, n_users: int, n_latent_factors: int, n_services: int, reg_item: float, reg_user: float, scope_name: str,
                  use_l2_upper_regularization: bool, average: float, standard_deviation: float) -> None:
         # placeholder
@@ -91,6 +92,7 @@ def __init__(self, n_items: int, n_users: int, n_latent_factors: int, n_services
 
 
 class MatrixFactorization(object):
+
     def __init__(self,
                  n_latent_factors: int,
                  learning_rate: float,

redshells/factory/optuna_param_factory.py

@@ -63,6 +63,7 @@ def _catboostclassifier_default(trial: optuna.trial.Trial):
 
 
 class _OptunaParamFactory(metaclass=Singleton):
+
     def __init__(self):
         self._rules = dict()
         self._rules['XGBClassifier_default'] = _xgbclassifier_default

redshells/factory/prediction_model_factory.py

@@ -4,6 +4,7 @@
 
 
 class _PredictionModelFactory(metaclass=Singleton):
+
     def __init__(self):
         self._models = dict()
         try:

redshells/model/lda_model.py

@@ -12,6 +12,7 @@
 class LdaModel(object):
     """TopicModel is a kind of wrapper of LdaModel in gensim module.
     """
+
     def __init__(self, n_topics: int, chunksize: int = 16, decay: float = 0.5, offset: int = 16, iterations: int = 3, eta: float = 1.0e-16) -> None:
         self.n_topics = n_topics
         self.chunksize = chunksize

redshells/model/scdv.py

@@ -17,6 +17,7 @@ class SCDV(object):
     See https://arxiv.org/pdf/1612.06778.pdf for details
 
     """
+
     def __init__(self, documents: List[List[str]], cluster_size: int, sparsity_percentage: float, gaussian_mixture_kwargs: Dict[Any, Any],
                  dictionary: gensim.corpora.Dictionary, w2v: Union[FastText, Word2Vec]) -> None:
         """

redshells/model/tdidf.py

@@ -0,0 +1,3 @@
+# for pickle backward compatibility
+# https://github.com/m3dev/redshells/pull/56#discussion_r711488588
+from redshells.model.tfidf import Tfidf

redshells/model/tfidf.py

@@ -9,6 +9,7 @@
 
 
 class Tfidf(object):
+
     def __init__(self, dictionary: gensim.corpora.Dictionary, tokens: List[List[str]]) -> None:
         self.dictionary = dictionary
         self.tfidf = gensim.models.TfidfModel([dictionary.doc2bow(t) for t in tokens])

redshells/train/utils/token_iterator.py

@@ -5,6 +5,7 @@
 
 
 class TokenIterator(object):
+
     def __init__(self, texts: List[str]) -> None:
         self.texts = texts
         self.i = 0

test/contrib/model/test_feature_aggregation_similarity_model.py

@@ -10,6 +10,7 @@
 
 
 class TestFeatureAggregationSimilarityModel(unittest.TestCase):
+
     def test(self):
         model = FeatureAggregationSimilarityModel(embedding_size=7, learning_rate=0.001, feature_size=2, item_size=4, max_feature_index=5)
         dataset = FeatureAggregationSimilarityDataset(x_item_indices=np.array([0, 1, 2]),

test/contrib/model/test_gcmc_graph_dataset.py

@@ -13,6 +13,7 @@ def _make_sparse_matrix(n, m, n_values):
 
 
 class TestGcmcGraphDataset(unittest.TestCase):
+
     def test(self):
         # This tests that GraphConvolutionalMatrixCompletion runs without error, and its loss and rmse are small enough.
         n_users = 101

test/contrib/model/test_gcmc_graph_dataset_map.py

@@ -9,6 +9,7 @@
 
 
 class TestGcmcGraphDataset(unittest.TestCase):
+
     def setUp(self) -> None:
         dataset = GcmcDataset(user_ids=np.array([0, 1, 2]), item_ids=np.array([10, 11, 12]), ratings=np.array([100, 101, 102]))
         self.graph_dataset = GcmcGraphDataset(dataset=dataset, test_size=0.1)

test/contrib/model/test_gcmc_id_map.py

@@ -7,6 +7,7 @@
 
 
 class TestGcmcIdMap(unittest.TestCase):
+
     def test_initialize(self):
         ids = np.array([0, 0, 1, 2, 3, 3])
         features = [{0: np.array([0]), 1: np.array([1])}]

test/contrib/model/test_graph_convolutional_matrix_completion.py

@@ -15,6 +15,7 @@ def _make_sparse_matrix(n, m, n_values):
 
 
 class GraphConvolutionalMatrixCompletionTest(unittest.TestCase):
+
     def test_run(self):
         # This tests that GraphConvolutionalMatrixCompletion runs without error, and its loss and rmse are small enough.
         n_users = 101

test/contrib/train/test_gcmc_dataset.py

@@ -10,6 +10,7 @@
 
 
 class TestGCMCDataset(unittest.TestCase):
+
     def test_without_information(self):
         user_ids = np.array([1, 1, 2, 2, 2])
         item_ids = np.array([1, 2, 1, 2, 3])

test/contrib/train/test_train_feature_aggregation_similarity_model.py

@@ -18,6 +18,7 @@ class _DummyTask(gokart.TaskOnKart):
 
 
 class TestTrainFeatureAggregationSimilarityModel(unittest.TestCase):
+
     def setUp(self):
         self.input_data = None
         self.dump_data = None

test/model/test_tfidf.py

@@ -11,6 +11,7 @@
 
 
 class TestTfidf(unittest.TestCase):
+
     def test_apply_with_empty(self):
         texts = [random.choices(string.ascii_letters, k=100) for _ in range(100)]
         dictionary = gensim.corpora.Dictionary(texts)

test/train/test_train_doc2vec.py

@@ -12,6 +12,7 @@ class _DummyTask(luigi.Task):
 
 
 class TrainDoc2VecTest(unittest.TestCase):
+
     def setUp(self):
         self.input_data = None
         self.dump_data = None

test/train/test_train_fasttext.py

@@ -11,6 +11,7 @@ class _DummyTask(gokart.TaskOnKart):
 
 
 class TrainFastTextTest(unittest.TestCase):
+
     def setUp(self):
         self.input_data = None
         self.dump_data = None

test/train/test_train_lda_model.py

@@ -12,6 +12,7 @@ class _DummyTask(luigi.Task):
 
 
 class TrainLdaModelTest(unittest.TestCase):
+
     def setUp(self):
         self.input_data = None
         self.dump_data = None

test/train/test_train_pairwise_similarity_model.py

@@ -13,6 +13,7 @@ class _DummyTask(luigi.Task):
 
 
 class TrainPairwiseSimilarityModelTest(unittest.TestCase):
+
     def setUp(self):
         self.input_data = dict()
         self.dump_data = None

test/train/test_train_word2vec.py

@@ -12,6 +12,7 @@ class _DummyTask(luigi.Task):
 
 
 class TrainWord2VecTest(unittest.TestCase):
+
     def setUp(self):
         self.input_data = None
         self.dump_data = None