[Misc][Refactor] Introduce ExecuteModelData (#4540)

vllm/engine/async_llm_engine.py

@@ -16,7 +16,7 @@ from vllm.logger import init_logger
 from vllm.lora.request import LoRARequest
 from vllm.outputs import RequestOutput
 from vllm.sampling_params import SamplingParams
-from vllm.sequence import MultiModalData, SamplerOutput
+from vllm.sequence import ExecuteModelRequest, MultiModalData, SamplerOutput
 from vllm.usage.usage_lib import UsageContext
 
 logger = init_logger(__name__)
@@ -210,12 +210,16 @@ class _AsyncLLMEngine(LLMEngine):
 
         if not scheduler_outputs.is_empty():
             # Execute the model.
+            execute_model_req = ExecuteModelRequest(
+                seq_group_metadata_list=seq_group_metadata_list,
+                blocks_to_swap_in=scheduler_outputs.blocks_to_swap_in,
+                blocks_to_swap_out=scheduler_outputs.blocks_to_swap_out,
+                blocks_to_copy=scheduler_outputs.blocks_to_copy,
+                num_lookahead_slots=scheduler_outputs.num_lookahead_slots,
+                running_queue_size=scheduler_outputs.running_queue_size,
+            )
             output = await self.model_executor.execute_model_async(
-                seq_group_metadata_list,
-                scheduler_outputs.blocks_to_swap_in,
-                scheduler_outputs.blocks_to_swap_out,
-                scheduler_outputs.blocks_to_copy,
-                num_lookahead_slots=scheduler_outputs.num_lookahead_slots)
+                execute_model_req)
         else:
             output = []