Cleanup tensorflow dependencies (#530)

* Remove tensorflow from requiremnents extras

* Remove import of tensorflow from `transformers4rec/config/transformer`

* Move tensorflow-metadata from dev to base requirements.

This is used in the `merlin_standard_lib` package which is used by `transformers4rec`

* Remove unused tensorflow-estimator and tensorflow-ranking from dev requirements

* Remove test of `to_huggingface_tf_model`

Co-authored-by: Sara Rabhi <sara.rabhi@gmail.com>

requirements/base.txt

@@ -3,3 +3,4 @@ transformers<4.19
 tqdm>=4.27
 betterproto<2.0.0
 pyarrow>=1.0
+tensorflow-metadata

requirements/dev.txt

@@ -4,8 +4,5 @@ flake8
 isort
 bandit
 mypy==0.971
-tensorflow-metadata
-tensorflow-estimator==2.6.*
-tensorflow-ranking>=0.4
 codespell
 click<8.1.0

setup.py

@@ -41,7 +41,6 @@ def read_requirements(filename):
 
 requirements = {
     "base": read_requirements("requirements/base.txt"),
-    "tensorflow": read_requirements("requirements/tensorflow.txt"),
     "pytorch": read_requirements("requirements/pytorch.txt"),
     "nvtabular": read_requirements("requirements/nvtabular.txt"),
     "docs": read_requirements("requirements/docs.txt"),

tests/config/test_transformer.py

@@ -15,7 +15,7 @@
 #
 
 import pytest
-from transformers import PreTrainedModel, TFPreTrainedModel
+from transformers import PreTrainedModel
 
 from transformers4rec.config import transformer as tconf
 
@@ -35,12 +35,3 @@ def test_to_hugginface_torch_model(config_cls):
     model = config.to_huggingface_torch_model()
 
     assert isinstance(model, PreTrainedModel)
-
-
-@pytest.mark.parametrize("config_cls", list(set(config_classes) - {tconf.ReformerConfig}))
-def test_to_hugginface_tf_model(config_cls):
-    config = config_cls.build(100, 4, 2, 20)
-
-    model = config.to_huggingface_tf_model()
-
-    assert isinstance(model, TFPreTrainedModel)

transformers4rec/config/transformer.py

@@ -18,11 +18,6 @@
 
 from merlin_standard_lib import Registry
 
-try:
-    import tensorflow as tf
-except ImportError:
-    tf = None
-
 transformer_registry: Registry = Registry("transformers")
 
 
@@ -63,48 +58,6 @@ def to_torch_model(
             loss_reduction=loss_reduction,
         ).to_model(**kwargs)
 
-    if tf:
-
-        def to_tf_model(
-            self,
-            input_features,
-            *prediction_task,
-            task_blocks=None,
-            task_weights=None,
-            loss_reduction=tf.reduce_mean,
-            **kwargs
-        ):
-            from .. import tf as tf4rec
-
-            if not isinstance(input_features, tf4rec.TabularSequenceFeatures):
-                raise ValueError("`input_features` must an instance of SequentialTabularFeatures")
-            if not all(isinstance(t, tf4rec.PredictionTask) for t in prediction_task):
-                raise ValueError(
-                    "`task` is of the wrong type, please provide one or multiple "
-                    "instance(s) of PredictionTask"
-                )
-
-            body = tf4rec.SequentialBlock(
-                [input_features, tf4rec.TransformerBlock(self, masking=input_features.masking)]
-            )
-
-            return tf4rec.Model(
-                tf4rec.Head(
-                    body,
-                    *prediction_task,
-                    task_blocks=task_blocks,
-                    task_weights=task_weights,
-                    loss_reduction=loss_reduction,
-                    inputs=input_features,
-                ),
-                **kwargs,
-            )
-
-    def to_huggingface_tf_model(self):
-        model_cls = transformers.TF_MODEL_MAPPING[self.transformers_config_cls]
-
-        return model_cls(self)
-
     @property
     def transformers_config_cls(self):
         return self.__class__.__bases__[1]