[Intel GPU] Fix xpu decode input (#9145)

vllm/worker/xpu_model_runner.py

@@ -15,6 +15,7 @@
 from vllm.distributed import get_pp_group
 from vllm.inputs import INPUT_REGISTRY, InputRegistry
 from vllm.logger import init_logger
+from vllm.model_executor import SamplingMetadataCache
 from vllm.model_executor.layers.sampler import SamplerOutput
 from vllm.model_executor.model_loader import get_model
 from vllm.multimodal import (MULTIMODAL_REGISTRY, BatchedTensorInputs,
@@ -136,7 +137,7 @@ def build(self) -> ModelInputForXPU:
             (input_tokens, input_positions,
              attn_metadata) = self._prepare_decode(
                  self.seq_group_metadata_list)
-            seq_lens = []
+            seq_lens = None
             multi_modal_kwargs = None
 
         return self.model_input_cls(
@@ -390,6 +391,10 @@ def __init__(
         # Lazy initialization.
         self.model: nn.Module  # Set after init_Model
 
+        self.sampling_metadata_cache: SamplingMetadataCache = \
+              SamplingMetadataCache() \
+                if self.parallel_config.pipeline_parallel_size == 1 else None
+
     def load_model(self) -> None:
         with DeviceMemoryProfiler() as m:
             self.model = get_model(
@@ -524,12 +529,14 @@ def prepare_model_input(
             seq_group_metadata_list, finished_requests_ids)
         # Sampling metadata is only required for the final pp group
         generators = self.get_generators(finished_requests_ids)
-        sampling_metadata = SamplingMetadata.prepare(seq_group_metadata_list,
-                                                     model_input.seq_lens,
-                                                     model_input.query_lens,
-                                                     self.device,
-                                                     pin_memory=False,
-                                                     generators=generators)
+        sampling_metadata = SamplingMetadata.prepare(
+            seq_group_metadata_list,
+            model_input.seq_lens,
+            model_input.query_lens,
+            self.device,
+            pin_memory=False,
+            generators=generators,
+            cache=self.sampling_metadata_cache)
 
         return dataclasses.replace(model_input,
                                    sampling_metadata=sampling_metadata,