remove extra code

tests/models/falcon/test_modeling_falcon.py

@@ -29,10 +29,8 @@
 from transformers.testing_utils import (
     is_flaky,
     require_bitsandbytes,
-    require_flash_attn,
     require_torch,
     require_torch_sdpa,
-    require_torch_gpu,
     slow,
     torch_device,
 )
@@ -583,11 +581,6 @@ def test_eager_matches_sdpa_generate(self):
 
                 self.assertTrue(torch.allclose(res_eager, res_sdpa))
 
-    @require_flash_attn
-    @require_torch_gpu
-    @slow
-    def test_flash_attention_2_padding_matches_padding_free_with_position_ids(self):
-        super().test_flash_attention_2_padding_matches_padding_free_with_position_ids()
 
 @require_torch
 class FalconLanguageGenerationTest(unittest.TestCase):

tests/models/gemma/test_modeling_gemma.py

@@ -526,12 +526,6 @@ def test_flash_attn_2_equivalence(self):
                 assert torch.allclose(logits_fa, logits, atol=3e-3)
 
 
-    @require_flash_attn
-    @require_torch_gpu
-    @slow
-    def test_flash_attention_2_padding_matches_padding_free_with_position_ids(self):
-        super().test_flash_attention_2_padding_matches_padding_free_with_position_ids()
-
 @slow
 @require_torch_gpu
 class GemmaIntegrationTest(unittest.TestCase):

tests/models/llama/test_modeling_llama.py

@@ -555,12 +555,6 @@ def test_use_flash_attention_2_true(self):
                 if not has_flash:
                     raise ValueError("The flash model should have flash attention layers")
 
-    @require_flash_attn
-    @require_torch_gpu
-    @slow
-    def test_flash_attention_2_padding_matches_padding_free_with_position_ids(self):
-        super().test_flash_attention_2_padding_matches_padding_free_with_position_ids()
-
     @require_torch_sdpa
     @slow
     def test_eager_matches_sdpa_generate(self):

tests/models/mistral/test_modeling_mistral.py

@@ -490,12 +490,7 @@ def test_flash_attn_2_generate_use_cache(self):
     @slow
     def test_flash_attn_2_inference_equivalence_right_padding(self):
         self.skipTest(reason="Mistral flash attention does not support right padding")
-
-    @require_flash_attn
-    @require_torch_gpu
-    @slow
-    def test_flash_attention_2_padding_matches_padding_free_with_position_ids(self):
-        super().test_flash_attention_2_padding_matches_padding_free_with_position_ids()
+
 
 @require_torch_gpu
 class MistralIntegrationTest(unittest.TestCase):

tests/models/mixtral/test_modeling_mixtral.py

@@ -492,12 +492,6 @@ def test_flash_attn_2_generate_use_cache(self):
     def test_flash_attn_2_inference_equivalence_right_padding(self):
         self.skipTest(reason="Mixtral flash attention does not support right padding")
 
-    @require_flash_attn
-    @require_torch_gpu
-    @slow
-    def test_flash_attention_2_padding_matches_padding_free_with_position_ids(self):
-        super().test_flash_attention_2_padding_matches_padding_free_with_position_ids()
-
     # Ignore copy
     def test_load_balancing_loss(self):
         r"""

tests/models/phi/test_modeling_phi.py

@@ -495,12 +495,6 @@ def test_flash_attn_2_generate_padding_right(self):
         self.assertListEqual(output_native, output_fa_2)
 
 
-    @require_flash_attn
-    @require_torch_gpu
-    @slow
-    def test_flash_attention_2_padding_matches_padding_free_with_position_ids(self):
-        super().test_flash_attention_2_padding_matches_padding_free_with_position_ids()
-
 @slow
 @require_torch
 class PhiIntegrationTest(unittest.TestCase):

tests/models/starcoder2/test_modeling_starcoder2.py

@@ -484,12 +484,6 @@ def test_flash_attn_2_inference_equivalence_right_padding(self):
         self.skipTest(reason="Starcoder2 flash attention does not support right padding")
 
 
-    @require_flash_attn
-    @require_torch_gpu
-    @slow
-    def test_flash_attention_2_padding_matches_padding_free_with_position_ids(self):
-        super().test_flash_attention_2_padding_matches_padding_free_with_position_ids()
-
 @slow
 @require_torch_gpu
 class Starcoder2IntegrationTest(unittest.TestCase):