Fix undeterministic order in modular dependencies (#39005)

* sort correctly

* Update modeling_minimax.py

* Update modular_model_converter.py

examples/modular-transformers/modeling_switch_function.py

@@ -14,13 +14,9 @@ from ...cache_utils import Cache
 from ...modeling_flash_attention_utils import FlashAttentionKwargs
 from ...modeling_utils import ALL_ATTENTION_FUNCTIONS
 from ...processing_utils import Unpack
-from ...utils import logging
 from .configuration_switch_function import SwitchFunctionConfig
 
 
-logger = logging.get_logger(__name__)
-
-
 def rotate_half(x):
     # Split and rotate. Note that this function is different from e.g. Llama.
     x1 = x[..., ::2]
@@ -145,15 +141,8 @@ class SwitchFunctionAttention(nn.Module):
             key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
 
         attention_interface: Callable = eager_attention_forward
-
         if self.config._attn_implementation != "eager":
-            if self.config._attn_implementation == "sdpa" and kwargs.get("output_attentions", False):
-                logger.warning_once(
-                    "`torch.nn.functional.scaled_dot_product_attention` does not support `output_attentions=True`. Falling back to "
-                    'eager attention. This warning can be removed using the argument `attn_implementation="eager"` when loading the model.'
-                )
-            else:
-                attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation]
+            attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation]
 
         attn_output, attn_weights = attention_interface(
             self,