[bugfix] fix siglip batch text output error (#28365)

Signed-off-by: piood <2477084691@qq.com>

vllm/model_executor/models/siglip.py

@@ -19,6 +19,7 @@ from transformers import (
 )
 
 from vllm.attention.layer import MultiHeadAttention
+from vllm.attention.layers.encoder_only_attention import EncoderOnlyAttention
 from vllm.config import VllmConfig
 from vllm.config.multimodal import BaseDummyOptions
 from vllm.distributed import divide, get_tensor_model_parallel_world_size
@@ -379,6 +380,7 @@ class SiglipAttention(nn.Module):
         quant_config: QuantizationConfig | None = None,
         *,
         prefix: str = "",
+        attn_cls: type[EncoderOnlyAttention] | type[MultiHeadAttention],
     ) -> None:
         super().__init__()
 
@@ -413,8 +415,11 @@ class SiglipAttention(nn.Module):
         self.tp_size = get_tensor_model_parallel_world_size()
         self.num_heads_per_partition = divide(self.num_heads, self.tp_size)
 
-        self.attn = MultiHeadAttention(
-            self.num_heads_per_partition, self.head_dim, self.scale
+        self.attn = attn_cls(
+            self.num_heads_per_partition,
+            self.head_dim,
+            self.scale,
+            prefix=f"{prefix}.attn",
         )
 
     def forward(
@@ -424,25 +429,7 @@ class SiglipAttention(nn.Module):
         """Input shape: Batch x Time x Channel"""
         qkv_states, _ = self.qkv_proj(hidden_states)
         query_states, key_states, value_states = qkv_states.chunk(3, dim=-1)
-
-        needs_unsqueeze = query_states.ndim == 2
-        if needs_unsqueeze:
-            query_states, key_states, value_states = (
-                query_states.unsqueeze(0),
-                key_states.unsqueeze(0),
-                value_states.unsqueeze(0),
-            )
-
         out = self.attn(query_states, key_states, value_states)
-
-        if needs_unsqueeze:
-            out, query_states, key_states, value_states = (
-                out.squeeze(0),
-                query_states.squeeze(0),
-                key_states.squeeze(0),
-                value_states.squeeze(0),
-            )
-
         attn_output, _ = self.out_proj(out)
 
         return attn_output, None
@@ -495,6 +482,7 @@ class SiglipEncoderLayer(nn.Module):
         quant_config: QuantizationConfig | None = None,
         *,
         prefix: str = "",
+        attn_cls: type[EncoderOnlyAttention] | type[MultiHeadAttention],
     ) -> None:
         super().__init__()
 
@@ -504,6 +492,7 @@ class SiglipEncoderLayer(nn.Module):
             config,
             quant_config=quant_config,
             prefix=f"{prefix}.self_attn",
+            attn_cls=attn_cls,
         )
         self.layer_norm1 = nn.LayerNorm(self.embed_dim, eps=config.layer_norm_eps)
         self.mlp = SiglipMLP(
@@ -539,6 +528,7 @@ class SiglipEncoder(nn.Module):
         num_hidden_layers_override: int | None = None,
         *,
         prefix: str = "",
+        attn_cls: type[EncoderOnlyAttention] | type[MultiHeadAttention],
     ) -> None:
         super().__init__()
 
@@ -555,6 +545,7 @@ class SiglipEncoder(nn.Module):
                     config,
                     quant_config=quant_config,
                     prefix=f"{prefix}.layers.{layer_idx}",
+                    attn_cls=attn_cls,
                 )
                 for layer_idx in range(num_hidden_layers)
             ]
@@ -598,6 +589,7 @@ class SiglipTextTransformer(nn.Module):
             config=config,
             quant_config=quant_config,
             prefix=f"{prefix}.encoder",
+            attn_cls=EncoderOnlyAttention,
         )
 
         self.final_layer_norm = nn.LayerNorm(embed_dim, eps=config.layer_norm_eps)
@@ -709,6 +701,7 @@ class SiglipVisionTransformer(nn.Module):
             quant_config=quant_config,
             num_hidden_layers_override=num_hidden_layers_override,
             prefix=f"{prefix}.encoder",
+            attn_cls=MultiHeadAttention,
         )
 
         num_hidden_layers = config.num_hidden_layers
@@ -1034,10 +1027,56 @@ class SiglipEmbeddingModel(nn.Module, SupportsMultiModal, SupportsQuant):
             inputs_embeds=inputs_embeds,
         )
         text_features = self.text_model.head(last_hidden_state)
-        # Flip to extract CLS token (first token after reversal) for pooling
-        text_features = text_features.flip(0)
+
+        # SigLIP uses reversed position_ids;
+        # flip sequences to move EOS token to first position
+        text_features = self._flip_sequences_by_position_ids(
+            text_features, position_ids
+        )
+
         return text_features
 
+    def _flip_sequences_by_position_ids(
+        self,
+        features: torch.Tensor,
+        position_ids: torch.Tensor,
+    ) -> torch.Tensor:
+        """Flip sequences so EOS token moves to first position for CLS pooling.
+
+        SigLIP position_ids are reversed within each sequence. This method detects
+        sequence boundaries and flips each sequence individually.
+        """
+        if len(features) == 1:
+            return features
+
+        # Detect sequence boundaries where position_ids decrease
+        position_diffs = position_ids[1:] - position_ids[:-1]
+        boundary_mask = position_diffs <= 0
+
+        boundary_indices = torch.cat(
+            [
+                torch.tensor([0], device=features.device),
+                torch.where(boundary_mask)[0] + 1,
+                torch.tensor([len(features)], device=features.device),
+            ]
+        )
+
+        # For each sequence [start, end), position i flips to: start + end - 1 - i
+        lengths = boundary_indices[1:] - boundary_indices[:-1]
+        starts = boundary_indices[:-1]
+        ends = boundary_indices[1:]
+
+        # Assign sequence ID to each element
+        sequence_ids = torch.arange(
+            len(lengths), device=features.device
+        ).repeat_interleave(lengths)
+
+        # Calculate flipped indices for all positions at once
+        current_positions = torch.arange(len(features), device=features.device)
+        flip_indices = starts[sequence_ids] + ends[sequence_ids] - 1 - current_positions
+
+        return features[flip_indices]
+
     def get_image_features(
         self,
         pixel_values: torch.Tensor,