[`CLeanup`] Revert SDPA attention changes that got in the static kv cache PR (#29027)

* revert unrelated changes that got in

* style

src/transformers/models/mistral/modeling_mistral.py

@@ -659,34 +659,28 @@ class MistralSdpaAttention(MistralAttention):
         value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
 
         kv_seq_len = key_states.shape[-2]
-        past_key_value = getattr(self, "past_key_value", past_key_value)
         if past_key_value is not None:
-            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)  # add what was seen
+            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
         cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
 
         query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
 
-        past_seen_tokens = kv_seq_len - key_states.shape[-2]
-        new_cache_positions = torch.arange(past_seen_tokens, past_seen_tokens + q_len, device=key_states.device)
         if past_key_value is not None:
-            cache_kwargs = {"sin": sin, "cos": cos, "position_ids": new_cache_positions}  # Specific to RoPE models
+            cache_kwargs = {"sin": sin, "cos": cos}  # Specific to RoPE models
             key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
 
         key_states = repeat_kv(key_states, self.num_key_value_groups)
         value_states = repeat_kv(value_states, self.num_key_value_groups)
 
-        if (
-            attention_mask is not None and not torch.all(attention_mask[..., 0] == 1) and q_len != 1
-        ):  # user defined causal mask
-            causal_mask = attention_mask[:, :, past_seen_tokens : past_seen_tokens + q_len, : key_states.shape[-2]]
-            # this one liner is equivalent to the pad_unpad function
-            causal_mask.mul_(~torch.eq(causal_mask, causal_mask.min()).all(dim=-1)[..., None])
-        else:
-            causal_mask = None
+        if attention_mask is not None:
+            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
+                raise ValueError(
+                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
+                )
 
         # SDPA with memory-efficient backend is currently (torch==2.1.2) bugged with non-contiguous inputs with custom attn_mask,
         # Reference: https://github.com/pytorch/pytorch/issues/112577.
-        if query_states.device.type == "cuda" and causal_mask is not None:
+        if query_states.device.type == "cuda" and attention_mask is not None:
             query_states = query_states.contiguous()
             key_states = key_states.contiguous()
             value_states = value_states.contiguous()
@@ -695,9 +689,10 @@ class MistralSdpaAttention(MistralAttention):
             query_states,
             key_states,
             value_states,
-            attn_mask=causal_mask,
+            attn_mask=attention_mask,
             dropout_p=self.attention_dropout if self.training else 0.0,
-            is_causal=causal_mask is None and q_len > 1,
+            # The q_len > 1 is necessary to match with AttentionMaskConverter.to_causal_4d that does not create a causal mask in case q_len == 1.
+            is_causal=self.is_causal and attention_mask is None and q_len > 1,
         )
 
         attn_output = attn_output.transpose(1, 2).contiguous()

src/transformers/models/mixtral/modeling_mixtral.py

@@ -736,34 +736,28 @@ class MixtralSdpaAttention(MixtralAttention):
         value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
 
         kv_seq_len = key_states.shape[-2]
-        past_key_value = getattr(self, "past_key_value", past_key_value)
         if past_key_value is not None:
-            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)  # add what was seen
+            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
         cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
 
         query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
 
-        past_seen_tokens = kv_seq_len - key_states.shape[-2]
-        new_cache_positions = torch.arange(past_seen_tokens, past_seen_tokens + q_len, device=key_states.device)
         if past_key_value is not None:
-            cache_kwargs = {"sin": sin, "cos": cos, "position_ids": new_cache_positions}  # Specific to RoPE models
+            cache_kwargs = {"sin": sin, "cos": cos}  # Specific to RoPE models
             key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
 
         key_states = repeat_kv(key_states, self.num_key_value_groups)
         value_states = repeat_kv(value_states, self.num_key_value_groups)
 
-        if (
-            attention_mask is not None and not torch.all(attention_mask[..., 0] == 1) and q_len != 1
-        ):  # user defined causal mask
-            causal_mask = attention_mask[:, :, past_seen_tokens : past_seen_tokens + q_len, : key_states.shape[-2]]
-            # this one liner is equivalent to the pad_unpad function
-            causal_mask.mul_(~torch.eq(causal_mask, causal_mask.min()).all(dim=-1)[..., None])
-        else:
-            causal_mask = None
+        if attention_mask is not None:
+            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
+                raise ValueError(
+                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
+                )
 
         # SDPA with memory-efficient backend is currently (torch==2.1.2) bugged with non-contiguous inputs with custom attn_mask,
         # Reference: https://github.com/pytorch/pytorch/issues/112577.
-        if query_states.device.type == "cuda" and causal_mask is not None:
+        if query_states.device.type == "cuda" and attention_mask is not None:
             query_states = query_states.contiguous()
             key_states = key_states.contiguous()
             value_states = value_states.contiguous()
@@ -772,9 +766,10 @@ class MixtralSdpaAttention(MixtralAttention):
             query_states,
             key_states,
             value_states,
-            attn_mask=causal_mask,
+            attn_mask=attention_mask,
             dropout_p=self.attention_dropout if self.training else 0.0,
-            is_causal=causal_mask is None and q_len > 1,
+            # The q_len > 1 is necessary to match with AttentionMaskConverter.to_causal_4d that does not create a causal mask in case q_len == 1.
+            is_causal=self.is_causal and attention_mask is None and q_len > 1,
         )
 
         attn_output = attn_output.transpose(1, 2).contiguous()

src/transformers/models/qwen2/modeling_qwen2.py

@@ -669,34 +669,28 @@ class Qwen2SdpaAttention(Qwen2Attention):
         value_states = value_states.view(bsz, q_len, self.num_key_value_heads, self.head_dim).transpose(1, 2)
 
         kv_seq_len = key_states.shape[-2]
-        past_key_value = getattr(self, "past_key_value", past_key_value)
         if past_key_value is not None:
-            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)  # add what was seen
+            kv_seq_len += past_key_value.get_usable_length(kv_seq_len, self.layer_idx)
         cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
 
         query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
 
-        past_seen_tokens = kv_seq_len - key_states.shape[-2]
-        new_cache_positions = torch.arange(past_seen_tokens, past_seen_tokens + q_len, device=key_states.device)
         if past_key_value is not None:
-            cache_kwargs = {"sin": sin, "cos": cos, "position_ids": new_cache_positions}  # Specific to RoPE models
+            cache_kwargs = {"sin": sin, "cos": cos}  # Specific to RoPE models
             key_states, value_states = past_key_value.update(key_states, value_states, self.layer_idx, cache_kwargs)
 
         key_states = repeat_kv(key_states, self.num_key_value_groups)
         value_states = repeat_kv(value_states, self.num_key_value_groups)
 
-        if (
-            attention_mask is not None and not torch.all(attention_mask[..., 0] == 1) and q_len != 1
-        ):  # user defined causal mask
-            causal_mask = attention_mask[:, :, past_seen_tokens : past_seen_tokens + q_len, : key_states.shape[-2]]
-            # this one liner is equivalent to the pad_unpad function
-            causal_mask.mul_(~torch.eq(causal_mask, causal_mask.min()).all(dim=-1)[..., None])
-        else:
-            causal_mask = None
+        if attention_mask is not None:
+            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
+                raise ValueError(
+                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
+                )
 
         # SDPA with memory-efficient backend is currently (torch==2.1.2) bugged with non-contiguous inputs with custom attn_mask,
         # Reference: https://github.com/pytorch/pytorch/issues/112577.
-        if query_states.device.type == "cuda" and causal_mask is not None:
+        if query_states.device.type == "cuda" and attention_mask is not None:
             query_states = query_states.contiguous()
             key_states = key_states.contiguous()
             value_states = value_states.contiguous()
@@ -705,9 +699,10 @@ class Qwen2SdpaAttention(Qwen2Attention):
             query_states,
             key_states,
             value_states,
-            attn_mask=causal_mask,
+            attn_mask=attention_mask,
             dropout_p=self.attention_dropout if self.training else 0.0,
-            is_causal=causal_mask is None and q_len > 1,
+            # The q_len > 1 is necessary to match with AttentionMaskConverter.to_causal_4d that does not create a causal mask in case q_len == 1.
+            is_causal=self.is_causal and attention_mask is None and q_len > 1,
         )
 
         attn_output = attn_output.transpose(1, 2).contiguous()