[Bugfix] Correct max tokens for non-contiguous embeds (#21798)

Signed-off-by: Alexandre Milesi <30204471+milesial@users.noreply.github.com>
Co-authored-by: Alexandre Milesi <30204471+milesial@users.noreply.github.com>

Author: milesial

vllm/multimodal/profiling.py

@@ -180,11 +180,14 @@ def _get_dummy_mm_inputs(
     def _get_mm_num_tokens(
         self,
         mm_inputs: MultiModalInputs,
+        mm_embeddings_only: bool = True,
     ) -> Mapping[str, int]:
         placeholders_by_modality = mm_inputs["mm_placeholders"]
 
         return {
-            modality: sum(item.get_num_embeds() for item in placeholders)
+            modality:
+            sum(item.get_num_embeds() if mm_embeddings_only else item.length
+                for item in placeholders)
             for modality, placeholders in placeholders_by_modality.items()
         }
 
@@ -253,10 +256,11 @@ def get_decoder_dummy_data(
             multi_modal_placeholders=mm_inputs["mm_placeholders"],
         )
 
-    def get_mm_max_tokens(
+    def _get_mm_max_tokens(
         self,
         seq_len: int,
         mm_counts: Optional[Mapping[str, int]] = None,
+        mm_embeddings_only: bool = True,
     ) -> Mapping[str, int]:
         if mm_counts is None:
             mm_counts = self.get_mm_limits()
@@ -285,4 +289,25 @@ def get_mm_max_tokens(
             return max_tokens_per_item
 
         mm_inputs = self._get_dummy_mm_inputs(seq_len, mm_counts)
-        return self._get_mm_num_tokens(mm_inputs)
+        return self._get_mm_num_tokens(mm_inputs,
+                                       mm_embeddings_only=mm_embeddings_only)
+
+    def get_mm_max_contiguous_tokens(
+        self,
+        seq_len: int,
+        mm_counts: Optional[Mapping[str, int]] = None,
+    ):
+        """
+        Returns the maximum length of the multimodal (image placeholders+text)
+        tokens, including any break/text tokens in-between image embeddings.
+
+        <im_start> [IMG] [IMG] [IMG] <row_break> [IMG] [IMG] [IMG] <im_end>
+        Returns 9, even when the number of image embeddings is 6.
+        
+        This is important to take into account when profiling and
+        initializing the encoder cache size.
+        """
+
+        return self._get_mm_max_tokens(seq_len,
+                                       mm_counts,
+                                       mm_embeddings_only=False)

vllm/multimodal/registry.py

@@ -129,7 +129,7 @@ def get_max_tokens_per_item_by_modality(
         seq_len = model_config.max_model_len
         mm_limits = self.get_mm_limits_per_prompt(model_config)
 
-        return profiler.get_mm_max_tokens(
+        return profiler.get_mm_max_contiguous_tokens(
             seq_len,
             {
                 modality: 1