remove all copied from attentions

src/transformers/modeling_flash_attention_utils.py

@@ -276,7 +276,7 @@ def _flash_attention_forward(
     if not use_top_left_mask:
         causal = is_causal
     else:
-        # TODO: Remove the `query_length != 1` check once Flash Attention for RoCm is bumped to 2.1. For details, please see the comment in transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__.
+        # TODO: Remove the `query_length != 1` check once Flash Attention for RoCm is bumped to 2.1.
         causal = is_causal and query_length != 1
 
     # Assuming 4D tensors, key_states.shape[1] is the key/value sequence length (source length).

src/transformers/models/bark/modeling_bark.py

@@ -197,7 +197,6 @@ class BarkSelfFlashAttention2(BarkSelfAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/bart/modeling_bart.py

@@ -294,7 +294,6 @@ class BartFlashAttention2(BartAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/chameleon/modeling_chameleon.py

@@ -362,7 +362,7 @@ def forward(
         return attn_output, attn_weights, past_key_value
 
 
-# copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2 with Llama->Chameleon
+# NO LONGER EXIST copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2 with Llama->Chameleon
 # TODO(joao): add me back asap :)
 class ChameleonFlashAttention2(ChameleonAttention):
     """

src/transformers/models/clip/modeling_clip.py

@@ -401,7 +401,6 @@ class CLIPFlashAttention2(CLIPAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/cohere/modeling_cohere.py

@@ -351,7 +351,8 @@ def forward(
         return attn_output, attn_weights, past_key_value
 
 
-# Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2 with Llama->Cohere
+# NO LONGER EXIST Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2 with Llama->Cohere
+# TODO cyril: modular
 class CohereFlashAttention2(CohereAttention):
     """
     Cohere flash attention module. This module inherits from `CohereAttention` as the weights of the module stays

src/transformers/models/data2vec/modeling_data2vec_audio.py

@@ -489,7 +489,6 @@ class Data2VecAudioFlashAttention2(Data2VecAudioAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/dbrx/modeling_dbrx.py

@@ -318,7 +318,6 @@ class DbrxFlashAttention2(DbrxAttention):
     calls the public API of flash attention.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/distilbert/modeling_distilbert.py

@@ -245,7 +245,6 @@ class DistilBertFlashAttention2(MultiHeadSelfAttention):
     API of flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/falcon/modeling_falcon.py

@@ -492,7 +492,6 @@ class FalconFlashAttention2(FalconAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/gpt_bigcode/modeling_gpt_bigcode.py

@@ -278,7 +278,6 @@ class GPTBigCodeFlashAttention2(GPTBigCodeAttention):
     API of flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/gpt_neo/modeling_gpt_neo.py

@@ -278,7 +278,6 @@ class GPTNeoFlashAttention2(GPTNeoSelfAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/gptj/modeling_gptj.py

@@ -266,7 +266,6 @@ class GPTJFlashAttention2(GPTJAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/granitemoe/modeling_granitemoe.py

@@ -510,7 +510,8 @@ def forward(
         return attn_output, attn_weights, past_key_value
 
 
-# Copied from transformers.models.granite.modeling_granite.GraniteFlashAttention2 with Granite->GraniteMoe
+# NO LONGER EXIST Copied from transformers.models.granite.modeling_granite.GraniteFlashAttention2 with Granite->GraniteMoe
+# TODO cyril: modular
 class GraniteMoeFlashAttention2(GraniteMoeAttention):
     """
     GraniteMoe flash attention module. This module inherits from `GraniteMoeAttention` as the weights of the module stays
@@ -617,7 +618,8 @@ def forward(
         return attn_output, attn_weights, past_key_value
 
 
-# Copied from transformers.models.granite.modeling_granite.GraniteSdpaAttention with Granite->GraniteMoe
+# NO LONGER EXIST Copied from transformers.models.granite.modeling_granite.GraniteSdpaAttention with Granite->GraniteMoe
+# TODO cyril: modular
 class GraniteMoeSdpaAttention(GraniteMoeAttention):
     """
     GraniteMoe attention module using torch.nn.functional.scaled_dot_product_attention. This module inherits from

src/transformers/models/hubert/modeling_hubert.py

@@ -563,7 +563,6 @@ class HubertFlashAttention2(HubertAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/idefics2/modeling_idefics2.py

@@ -272,7 +272,6 @@ class Idefics2VisionFlashAttention2(Idefics2VisionAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 
@@ -859,15 +858,15 @@ def forward(
         return attn_output, attn_weights, past_key_value
 
 
-# Copied from transformers.models.mistral.modeling_mistral.MistralFlashAttention2 with MistralAttention->Idefics2PerceiverAttention,MistralFlashAttention->Idefics2PerceiverFlashAttention,Mistral->Idefics2
+# NO LONGER EXIST Copied from transformers.models.mistral.modeling_mistral.MistralFlashAttention2 with MistralAttention->Idefics2PerceiverAttention,MistralFlashAttention->Idefics2PerceiverFlashAttention,Mistral->Idefics2
+# TODO cyril: modular
 class Idefics2PerceiverFlashAttention2(Idefics2PerceiverAttention):
     """
     Idefics2 flash attention module. This module inherits from `Idefics2PerceiverAttention` as the weights of the module stays
     untouched. The only required change would be on the forward pass where it needs to correctly call the public API of
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/idefics3/modeling_idefics3.py

@@ -273,7 +273,6 @@ class Idefics3VisionFlashAttention2(Idefics3VisionAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/jamba/modeling_jamba.py

@@ -384,7 +384,6 @@ class JambaFlashAttention2(JambaAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/jetmoe/modeling_jetmoe.py

@@ -641,7 +641,6 @@ def forward(
 
 
 class JetMoeFlashAttention2(JetMoeAttention):
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/m2m_100/modeling_m2m_100.py

@@ -348,7 +348,6 @@ class M2M100FlashAttention2(M2M100Attention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/mbart/modeling_mbart.py

@@ -291,7 +291,6 @@ class MBartFlashAttention2(MBartAttention):
     flash attention and deal with padding tokens in case the input contains any of them.
     """
 
-    # Copied from transformers.models.llama.modeling_llama.LlamaFlashAttention2.__init__
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
 

src/transformers/models/mimi/modeling_mimi.py

@@ -559,7 +559,8 @@ def forward(
         return attn_output, attn_weights, past_key_value
 
 
-# Copied from transformers.models.gemma.modeling_gemma.GemmaFlashAttention2 with Gemma->Mimi
+# NO LONGER EXIST Copied from transformers.models.gemma.modeling_gemma.GemmaFlashAttention2 with Gemma->Mimi
+# TODO cyril: modular
 class MimiFlashAttention2(MimiAttention):
     """
     Mimi flash attention module. This module inherits from `MimiAttention` as the weights of the module stays
@@ -670,7 +671,8 @@ def forward(
         return attn_output, attn_weights, past_key_value
 
 
-# Copied from transformers.models.gemma.modeling_gemma.GemmaSdpaAttention with Gemma->Mimi
+# NO LONGER EXIST Copied from transformers.models.gemma.modeling_gemma.GemmaSdpaAttention with Gemma->Mimi
+# TODO cyril: modular
 class MimiSdpaAttention(MimiAttention):
     """
     Mimi attention module using torch.nn.functional.scaled_dot_product_attention. This module inherits from