Stop storing references to bound methods via tf.function (#24146)

* Stop storing references to bound methods in tf.functions

* Remove the gc.collect calls now that we resolved the underlying problem

* Remove the default signature from model.serving entirely, big cleanup

* Remove _prune_signature as self.input_signature can prune itself

* Restore serving docstring

* Update int support test to check the input signature

* Make sure other tests also use model.input_signature and not serving.input_signature

* Restore _prune_signature

* Remove the doctest GC now it's no longer needed

* Correct core tests to use the pruned sig

* order lines correctly in core tests

* Add eager_serving back with a deprecation warning

tests/models/rag/test_modeling_tf_rag.py

@@ -1,6 +1,5 @@
 from __future__ import annotations
 
-import gc
 import json
 import os
 import shutil
@@ -551,11 +550,6 @@ class TFRagDPRBartTest(TFRagTestMixin, unittest.TestCase):
 @require_sentencepiece
 @require_tokenizers
 class TFRagModelIntegrationTests(unittest.TestCase):
-    def tearDown(self):
-        super().tearDown()
-        # clean-up as much as possible GPU memory occupied by PyTorch
-        gc.collect()
-
     @cached_property
     def token_model(self):
         return TFRagTokenForGeneration.from_pretrained_question_encoder_generator(

tests/models/sam/test_modeling_tf_sam.py

@@ -17,7 +17,6 @@
 
 from __future__ import annotations
 
-import gc
 import inspect
 import unittest
 
@@ -431,11 +430,6 @@ def prepare_dog_img():
 @require_tf
 @slow
 class TFSamModelIntegrationTest(unittest.TestCase):
-    def tearDown(self):
-        super().tearDown()
-        # clean-up as much as possible GPU memory occupied by PyTorch
-        gc.collect()
-
     def test_inference_mask_generation_no_point(self):
         model = TFSamModel.from_pretrained("facebook/sam-vit-base")
         processor = SamProcessor.from_pretrained("facebook/sam-vit-base")

tests/models/xglm/test_modeling_tf_xglm.py

@@ -15,7 +15,6 @@
 
 from __future__ import annotations
 
-import gc
 import unittest
 
 from transformers import XGLMConfig, XGLMTokenizer, is_tf_available
@@ -173,11 +172,6 @@ class TFXGLMModelTest(TFModelTesterMixin, PipelineTesterMixin, unittest.TestCase
 
 @require_tf
 class TFXGLMModelLanguageGenerationTest(unittest.TestCase):
-    def tearDown(self):
-        super().tearDown()
-        # clean-up as much as possible GPU memory occupied by PyTorch
-        gc.collect()
-
     @slow
     def test_lm_generate_xglm(self, verify_outputs=True):
         model = TFXGLMForCausalLM.from_pretrained("facebook/xglm-564M")