[Spec Decode] Reduce TP communication for speculative decoding draft token generation (#34049)

Signed-off-by: qizixi <qizixi@meta.com>
Co-authored-by: Lu Fang <30275821+houseroad@users.noreply.github.com>

vllm/model_executor/models/llama4_eagle.py

@@ -208,6 +208,23 @@ class EagleLlama4ForCausalLM(Llama4ForCausalLM):
     ) -> tuple[torch.Tensor, torch.Tensor]:
         return self.model(input_ids, positions, hidden_states, inputs_embeds)
 
+    def get_top_tokens(
+        self,
+        hidden_states: torch.Tensor,
+    ) -> torch.Tensor:
+        """Vocab-parallel argmax without all-gathering full logits.
+
+        Falls back to full logits when draft_id_to_target_id remapping is
+        active, since the shared lm_head covers the full target vocab but
+        the draft model only predicts over a subset (draft_vocab_size).
+        """
+        if (
+            hasattr(self, "draft_id_to_target_id")
+            and self.draft_id_to_target_id is not None
+        ):
+            return self.compute_logits(hidden_states).argmax(dim=-1)
+        return self.logits_processor.get_top_tokens(self.lm_head, hidden_states)
+
     def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> None:
         def transform(inputs):
             name, loaded_weight = inputs