CI: avoid human error, automatically infer generative models (#33212)

* tmp commit

* move tests to the right class

* remove ALL all_generative_model_classes = ...

* skip tf roberta

* skip InstructBlipForConditionalGenerationDecoderOnlyTest

* videollava

* reduce diff

* reduce diff

* remove  on vlms

* fix a few more

* manual rebase bits

* more manual rebase

* remove all manual generative model class test entries

* fix up to ernie

* a few more removals

* handle remaining cases

* recurrent gemma

* it's better here

* make fixup

* tf idefics is broken

* tf bert + generate is broken

* don't touch tf :()

* don't touch tf :(

* make fixup

* better comments for test skips

* revert tf changes

* remove empty line removal

* one more

* missing one

tests/models/musicgen/test_modeling_musicgen.py

@@ -176,6 +176,8 @@ class MusicgenDecoderTester:
 @require_torch
 class MusicgenDecoderTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin, unittest.TestCase):
     all_model_classes = (MusicgenModel, MusicgenForCausalLM) if is_torch_available() else ()
+    # Doesn't run generation tests. See `greedy_sample_model_classes` below
+    all_generative_model_classes = ()
     greedy_sample_model_classes = (
         (MusicgenForCausalLM,) if is_torch_available() else ()
     )  # we don't want to run all the generation tests, only a specific subset
@@ -801,6 +803,8 @@ class MusicgenTester:
 @require_torch
 class MusicgenTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMixin, unittest.TestCase):
     all_model_classes = (MusicgenForConditionalGeneration,) if is_torch_available() else ()
+    # Doesn't run generation tests. See `greedy_sample_model_classes` below
+    all_generative_model_classes = ()
     greedy_sample_model_classes = (MusicgenForConditionalGeneration,) if is_torch_available() else ()
     pipeline_model_mapping = {"text-to-audio": MusicgenForConditionalGeneration} if is_torch_available() else {}
     test_pruning = False  # training is not supported yet for MusicGen