[tests] remove tests from libraries with deprecated support (flax, tensorflow_text, ...) (#39051)

* rm tf/flax tests

* more flax deletions

* revert fixture change

* reverted test that should not be deleted; rm tf/flax test

* revert

* fix a few add-model-like tests

* fix add-model-like checkpoint source

* a few more

* test_get_model_files_only_pt fix

* fix test_retrieve_info_for_model_with_xxx

* fix test_retrieve_model_classes

* relative paths are the devil

* add todo

tests/tokenization/test_tokenization_utils.py

@@ -37,7 +37,6 @@ from transformers import (
 from transformers.models.gpt2.tokenization_gpt2 import GPT2Tokenizer
 from transformers.testing_utils import (
     CaptureStderr,
-    require_flax,
     require_sentencepiece,
     require_tokenizers,
     require_torch,
@@ -98,8 +97,6 @@ class TokenizerUtilsTest(unittest.TestCase):
 
     @require_tokenizers
     def test_batch_encoding_pickle(self):
-        import numpy as np
-
         tokenizer_p = BertTokenizer.from_pretrained("google-bert/bert-base-cased")
         tokenizer_r = BertTokenizerFast.from_pretrained("google-bert/bert-base-cased")
 
@@ -189,22 +186,6 @@ class TokenizerUtilsTest(unittest.TestCase):
         self.assertEqual(tensor_batch["inputs"].shape, (1, 3))
         self.assertEqual(tensor_batch["labels"].shape, (1,))
 
-    @require_flax
-    def test_batch_encoding_with_labels_jax(self):
-        batch = BatchEncoding({"inputs": [[1, 2, 3], [4, 5, 6]], "labels": [0, 1]})
-        tensor_batch = batch.convert_to_tensors(tensor_type="jax")
-        self.assertEqual(tensor_batch["inputs"].shape, (2, 3))
-        self.assertEqual(tensor_batch["labels"].shape, (2,))
-        # test converting the converted
-        with CaptureStderr() as cs:
-            tensor_batch = batch.convert_to_tensors(tensor_type="jax")
-        self.assertFalse(len(cs.err), msg=f"should have no warning, but got {cs.err}")
-
-        batch = BatchEncoding({"inputs": [1, 2, 3], "labels": 0})
-        tensor_batch = batch.convert_to_tensors(tensor_type="jax", prepend_batch_axis=True)
-        self.assertEqual(tensor_batch["inputs"].shape, (1, 3))
-        self.assertEqual(tensor_batch["labels"].shape, (1,))
-
     def test_padding_accepts_tensors(self):
         features = [{"input_ids": np.array([0, 1, 2])}, {"input_ids": np.array([0, 1, 2, 3])}]
         tokenizer = BertTokenizer.from_pretrained("google-bert/bert-base-cased")