[Core] Pipeline parallel with Ray ADAG (#6837)

Support pipeline-parallelism with Ray accelerated DAG.

Signed-off-by: Rui Qiao <ruisearch42@gmail.com>

Dockerfile

@@ -42,6 +42,7 @@ WORKDIR /workspace
 
 # install build and runtime dependencies
 COPY requirements-common.txt requirements-common.txt
+COPY requirements-adag.txt requirements-adag.txt
 COPY requirements-cuda.txt requirements-cuda.txt
 RUN --mount=type=cache,target=/root/.cache/pip \
     python3 -m pip install -r requirements-cuda.txt
@@ -78,6 +79,7 @@ COPY setup.py setup.py
 COPY cmake cmake
 COPY CMakeLists.txt CMakeLists.txt
 COPY requirements-common.txt requirements-common.txt
+COPY requirements-adag.txt requirements-adag.txt
 COPY requirements-cuda.txt requirements-cuda.txt
 COPY pyproject.toml pyproject.toml
 COPY vllm vllm

MANIFEST.in

@@ -1,4 +1,5 @@
 include LICENSE
+include requirements-adag.txt
 include requirements-common.txt
 include requirements-cuda.txt
 include requirements-rocm.txt

requirements-adag.txt

@@ -0,0 +1,3 @@
+# Dependencies for Ray accelerated DAG
+cupy-cuda12x
+ray >= 2.32

requirements-test.txt

@@ -1,3 +1,6 @@
+# Needed for Ray accelerated DAG tests
+-r requirements-adag.txt
+
 # testing
 pytest
 tensorizer>=2.9.0

tests/distributed/test_pipeline_parallel.py

@@ -15,22 +15,31 @@ VLLM_MULTI_NODE = os.getenv("VLLM_MULTI_NODE", "0") == "1"
 
 
 @pytest.mark.parametrize(
-    "TP_SIZE, PP_SIZE, EAGER_MODE, CHUNKED_PREFILL, MODEL_NAME, DIST_BACKEND",
+    ("TP_SIZE, PP_SIZE, EAGER_MODE, CHUNKED_PREFILL, "
-    [
+     "MODEL_NAME, DIST_BACKEND, USE_RAY_ADAG, USE_RAY_ADAG_NCCL"), [
-        (2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray"),
+         (2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", False, False),
-        (2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray"),
+         (2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", False, False),
-        (1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "ray"),
+         (1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "ray", False, False),
-        (1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray"),
+         (1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", False, False),
-        (1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray"),
+         (1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", False, False),
-        (2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "mp"),
+         (2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", True, False),
-        (2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "mp"),
+         (2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, False),
-        (1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "mp"),
+         (1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, False),
-        (1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "mp"),
+         (1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", True, False),
-        (1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "mp"),
+         (1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, False),
-    ])
+         (2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", True, True),
-@fork_new_process_for_each_test
+         (2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, True),
+         (1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, True),
+         (1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "ray", True, True),
+         (1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "ray", True, True),
+         (2, 2, 0, 1, "meta-llama/Meta-Llama-3-8B", "mp", False, False),
+         (2, 2, 1, 0, "meta-llama/Meta-Llama-3-8B", "mp", False, False),
+         (1, 3, 0, 0, "meta-llama/Meta-Llama-3-8B", "mp", False, False),
+         (1, 4, 0, 1, "meta-llama/Meta-Llama-3-8B", "mp", False, False),
+         (1, 4, 1, 0, "meta-llama/Meta-Llama-3-8B", "mp", False, False),
+     ])
 def test_compare_tp(TP_SIZE, PP_SIZE, EAGER_MODE, CHUNKED_PREFILL, MODEL_NAME,
-                    DIST_BACKEND):
+                    DIST_BACKEND, USE_RAY_ADAG, USE_RAY_ADAG_NCCL):
     if VLLM_MULTI_NODE and DIST_BACKEND == "mp":
         pytest.skip("Skipping multi-node pipeline parallel test for "
                     "multiprocessing distributed backend")
@@ -67,8 +76,18 @@ def test_compare_tp(TP_SIZE, PP_SIZE, EAGER_MODE, CHUNKED_PREFILL, MODEL_NAME,
     if EAGER_MODE:
         pp_args.append("--enforce-eager")
         tp_args.append("--enforce-eager")
+    pp_env = None
+    if USE_RAY_ADAG:
+        assert DIST_BACKEND == "ray", (
+            "Ray ADAG is only supported with Ray distributed backend")
+        pp_env = {
+            "VLLM_USE_RAY_COMPILED_DAG": "1",
+            "VLLM_USE_RAY_SPMD_WORKER": "1",
+            "VLLM_USE_RAY_COMPILED_DAG_NCCL_CHANNEL":
+            str(int(USE_RAY_ADAG_NCCL)),
+        }
 
-    compare_two_settings(MODEL_NAME, pp_args, tp_args)
+    compare_two_settings(MODEL_NAME, pp_args, tp_args, pp_env)
 
 
 @pytest.mark.parametrize("PP_SIZE, MODEL_NAME", [

tests/utils.py

@@ -7,7 +7,7 @@ import time
 import warnings
 from contextlib import contextmanager
 from pathlib import Path
-from typing import Any, Dict, List
+from typing import Any, Dict, List, Optional
 
 import openai
 import ray
@@ -57,6 +57,7 @@ class RemoteOpenAIServer:
         model: str,
         cli_args: List[str],
         *,
+        env_dict: Optional[Dict[str, str]] = None,
         auto_port: bool = True,
     ) -> None:
         if auto_port:
@@ -77,6 +78,8 @@ class RemoteOpenAIServer:
         # the current process might initialize cuda,
         # to be safe, we should use spawn method
         env['VLLM_WORKER_MULTIPROC_METHOD'] = 'spawn'
+        if env_dict is not None:
+            env.update(env_dict)
         self.proc = subprocess.Popen(["vllm", "serve"] + [model] + cli_args,
                                      env=env,
                                      stdout=sys.stdout,
@@ -89,6 +92,11 @@ class RemoteOpenAIServer:
 
     def __exit__(self, exc_type, exc_value, traceback):
         self.proc.terminate()
+        try:
+            self.proc.wait(3)
+        except subprocess.TimeoutExpired:
+            # force kill if needed
+            self.proc.kill()
 
     def _wait_for_server(self, *, url: str, timeout: float):
         # run health check
@@ -127,10 +135,21 @@ class RemoteOpenAIServer:
         )
 
 
-def compare_two_settings(model: str, arg1: List[str], arg2: List[str]):
+def compare_two_settings(model: str,
+                         arg1: List[str],
+                         arg2: List[str],
+                         env1: Optional[Dict[str, str]] = None,
+                         env2: Optional[Dict[str, str]] = None):
     """
-    Launch API server with two different sets of arguments and compare the
+    Launch API server with two different sets of arguments/environments
-    results of the API calls. The arguments are after the model name.
+    and compare the results of the API calls.
+
+    Args:
+        model: The model to test.
+        arg1: The first set of arguments to pass to the API server.
+        arg2: The second set of arguments to pass to the API server.
+        env1: The first set of environment variables to pass to the API server.
+        env2: The second set of environment variables to pass to the API server.
     """
 
     tokenizer = AutoTokenizer.from_pretrained(model)
@@ -138,8 +157,8 @@ def compare_two_settings(model: str, arg1: List[str], arg2: List[str]):
     prompt = "Hello, my name is"
     token_ids = tokenizer(prompt)["input_ids"]
     results = []
-    for args in (arg1, arg2):
+    for args, env in ((arg1, env1), (arg2, env2)):
-        with RemoteOpenAIServer(model, args) as server:
+        with RemoteOpenAIServer(model, args, env_dict=env) as server:
             client = server.get_client()
 
             # test models list

vllm/envs.py

@@ -38,6 +38,7 @@ if TYPE_CHECKING:
     VLLM_FUSED_MOE_CHUNK_SIZE: int = 64 * 1024
     VLLM_USE_RAY_SPMD_WORKER: bool = False
     VLLM_USE_RAY_COMPILED_DAG: bool = False
+    VLLM_USE_RAY_COMPILED_DAG_NCCL_CHANNEL: bool = True
     VLLM_WORKER_MULTIPROC_METHOD: str = "fork"
     VLLM_ASSETS_CACHE: str = os.path.join(VLLM_CACHE_ROOT, "assets")
     VLLM_IMAGE_FETCH_TIMEOUT: int = 5
@@ -273,13 +274,20 @@ environment_variables: Dict[str, Callable[[], Any]] = {
     # execution on all workers.
     # Run vLLM with VLLM_USE_RAY_SPMD_WORKER=1 to enable it.
     "VLLM_USE_RAY_SPMD_WORKER":
-    lambda: bool(os.getenv("VLLM_USE_RAY_SPMD_WORKER", 0)),
+    lambda: bool(int(os.getenv("VLLM_USE_RAY_SPMD_WORKER", "0"))),
 
     # If the env var is set, it uses the Ray's compiled DAG API
     # which optimizes the control plane overhead.
     # Run vLLM with VLLM_USE_RAY_COMPILED_DAG=1 to enable it.
     "VLLM_USE_RAY_COMPILED_DAG":
-    lambda: bool(os.getenv("VLLM_USE_RAY_COMPILED_DAG", 0)),
+    lambda: bool(int(os.getenv("VLLM_USE_RAY_COMPILED_DAG", "0"))),
+
+    # If the env var is set, it uses NCCL for communication in
+    # Ray's compiled DAG. This flag is ignored if
+    # VLLM_USE_RAY_COMPILED_DAG is not set.
+    "VLLM_USE_RAY_COMPILED_DAG_NCCL_CHANNEL":
+    lambda: bool(int(os.getenv("VLLM_USE_RAY_COMPILED_DAG_NCCL_CHANNEL", "1"))
+                 ),
 
     # Use dedicated multiprocess context for workers.
     # Both spawn and fork work

vllm/executor/ray_gpu_executor.py

@@ -105,12 +105,19 @@ class RayGPUExecutor(DistributedGPUExecutor):
         # The remaining workers are the actual ray actors.
         self.workers: List[RayWorkerWrapper] = []
 
+        # Used in ray compiled DAG: indexed first by PP rank,
+        # and then TP rank. In other words, the inner list is
+        # the TP group of workers for a PP rank.
+        self.pp_tp_workers: List[List[RayWorkerWrapper]] = []
+
         if self.parallel_config.ray_workers_use_nsight:
             ray_remote_kwargs = self._configure_ray_workers_use_nsight(
                 ray_remote_kwargs)
 
+        logger.info("use_ray_spmd_worker: %s", self.use_ray_spmd_worker)
         # Create the workers.
         driver_ip = get_ip()
+        logger.info("driver_ip: %s", driver_ip)
         worker_wrapper_kwargs = self._get_worker_wrapper_args()
         for bundle_id, bundle in enumerate(placement_group.bundle_specs):
             if not bundle.get("GPU", 0):
@@ -142,42 +149,49 @@ class RayGPUExecutor(DistributedGPUExecutor):
                     # Else, added to the list of workers.
                     self.workers.append(worker)
 
+        logger.debug("workers: %s", self.workers)
+        logger.debug("driver_dummy_worker: %s", self.driver_dummy_worker)
         if not self.use_ray_spmd_worker and self.driver_dummy_worker is None:
             raise ValueError(
                 "Ray does not allocate any GPUs on the driver node. Consider "
                 "adjusting the Ray placement group or running the driver on a "
                 "GPU node.")
 
+        worker_ips = [
+            ray.get(worker.get_node_ip.remote())  # type: ignore[attr-defined]
+            for worker in self.workers
+        ]
+        ip_counts: Dict[str, int] = {}
+        for ip in worker_ips:
+            ip_counts[ip] = ip_counts.get(ip, 0) + 1
+
+        def sort_by_driver_then_worker_ip(worker):
+            """
+            Sort the workers based on 3 properties:
+            1. If the worker is on the same node as the driver (vllm engine),
+                it should be placed first.
+            2. Then, if the worker is on a node with fewer workers, it should
+                be placed first.
+            3. Finally, if the work is on a node with smaller IP address, it
+                should be placed first.
+            """
+            ip = ray.get(worker.get_node_ip.remote())
+            return (ip != driver_ip, ip_counts[ip], ip)
+
+        # After sorting, the workers on the same node will be
+        # close to each other, and the workers on the driver
+        # node will be placed first.
+        self.workers = sorted(self.workers, key=sort_by_driver_then_worker_ip)
+
         # Get the set of GPU IDs used on each node.
         worker_node_and_gpu_ids = self._run_workers("get_node_and_gpu_ids",
                                                     use_dummy_driver=True)
 
-        # the order in `worker_node_and_gpu_ids` does not necessarily match
-        # the machine boundaries. We need to make sure that workers in the
-        # same node are assigned consecutive ranks.
-        # examples:
-        # [('852a09a13c7503ef126d7c828454c741494b1be33a8627a5206604d9', [0]), ('dfaad7adfdae57a694cc74490db45bd112c9f31243523e43ddc2e7f0', [0]), ('dfaad7adfdae57a694cc74490db45bd112c9f31243523e43ddc2e7f0', [1]), ('dfaad7adfdae57a694cc74490db45bd112c9f31243523e43ddc2e7f0', [2]), ('dfaad7adfdae57a694cc74490db45bd112c9f31243523e43ddc2e7f0', [3]), ('852a09a13c7503ef126d7c828454c741494b1be33a8627a5206604d9', [1]), ('852a09a13c7503ef126d7c828454c741494b1be33a8627a5206604d9', [2]), ('852a09a13c7503ef126d7c828454c741494b1be33a8627a5206604d9', [3])] # noqa
-
-        # initialize worker ranks with -1 (unassigned)
-        worker_ranks = [-1 for x in worker_node_and_gpu_ids]
-        current_rank = 0
-        while -1 in worker_ranks:
-            # whenever we find an unassigned worker, find the node
-            index = worker_ranks.index(-1)
-            current_node_id = worker_node_and_gpu_ids[index][0]
-            # assign ranks to all workers in the same node
-            for i, (node_id, _) in enumerate(worker_node_and_gpu_ids):
-                if node_id == current_node_id:
-                    worker_ranks[i] = current_rank
-                    current_rank += 1
-        # with the above example, worker_ranks will be [0, 4, 5, 6, 7, 1, 2, 3]
-
         node_workers = defaultdict(list)  # node id -> list of worker ranks
         node_gpus = defaultdict(list)  # node id -> list of gpu ids
 
-        for worker_rank, (node_id, gpu_ids) in zip(worker_ranks,
+        for i, (node_id, gpu_ids) in enumerate(worker_node_and_gpu_ids):
-                                                   worker_node_and_gpu_ids):
+            node_workers[node_id].append(i)
-            node_workers[node_id].append(worker_rank)
             # `gpu_ids` can be a list of strings or integers.
             # convert them to integers for consistency.
             # NOTE: gpu_ids can be larger than 9 (e.g. 16 GPUs),
@@ -202,16 +216,6 @@ class RayGPUExecutor(DistributedGPUExecutor):
         self._run_workers("update_environment_variables",
                           all_args=all_args_to_update_environment_variables)
 
-        if len(node_gpus) == 1:
-            # in single node case, we don't need to get the IP address.
-            # the loopback address is sufficient
-            # NOTE: a node may have several IP addresses, one for each
-            # network interface. `get_ip()` might return any of them,
-            # while they might not work for communication inside the node
-            # if the network setup is complicated. Using the loopback address
-            # solves this issue, as it always works for communication inside
-            # the node.
-            driver_ip = "127.0.0.1"
         distributed_init_method = get_distributed_init_method(
             driver_ip, get_open_port())
 
@@ -221,8 +225,7 @@ class RayGPUExecutor(DistributedGPUExecutor):
                 local_rank=node_workers[node_id].index(rank),
                 rank=rank,
                 distributed_init_method=distributed_init_method,
-            ) for rank, (node_id,
+            ) for rank, (node_id, _) in enumerate(worker_node_and_gpu_ids)
-                         _) in zip(worker_ranks, worker_node_and_gpu_ids)
         ]
         self._run_workers("init_worker", all_kwargs=init_worker_all_kwargs)
 
@@ -231,6 +234,19 @@ class RayGPUExecutor(DistributedGPUExecutor):
                           max_concurrent_workers=self.parallel_config.
                           max_parallel_loading_workers)
 
+        if self.use_ray_spmd_worker:
+            for pp_rank in range(self.parallel_config.pipeline_parallel_size):
+                self.pp_tp_workers.append([])
+                for tp_rank in range(
+                        self.parallel_config.tensor_parallel_size):
+                    # PP=2, TP=4
+                    # pp_tp_workers = [[0, 1, 2, 3], [4, 5, 6, 7]]
+                    rank = (pp_rank * self.parallel_config.tensor_parallel_size
+                            ) + tp_rank
+                    assert len(self.pp_tp_workers[pp_rank]) == tp_rank
+                    assert pp_rank < len(self.pp_tp_workers)
+                    self.pp_tp_workers[pp_rank].append(self.workers[rank])
+
         # This is the list of workers that are rank 0 of each TP group EXCEPT
         # global rank 0. These are the workers that will broadcast to the
         # rest of the workers.
@@ -241,9 +257,9 @@ class RayGPUExecutor(DistributedGPUExecutor):
         self.non_driver_workers: List[RayWorkerWrapper] = []
 
         # Enforce rank order for correct rank to return final output.
-        for rank, worker in sorted(zip(worker_ranks[1:], self.workers)):
+        for index, worker in enumerate(self.workers):
-            # We need to skip the driver worker, which we
+            # The driver worker is rank 0 and not in self.workers.
-            # do by skipping worker_ranks[0] which is always 0.
+            rank = index + 1
             if rank % self.parallel_config.tensor_parallel_size == 0:
                 self.tp_driver_workers.append(worker)
             else:
@@ -376,16 +392,47 @@ class RayGPUExecutor(DistributedGPUExecutor):
             raise ValueError(f"Ray version {required_version} or greater is "
                              f"required, but found {current_version}")
 
-        from ray.dag import InputNode, MultiOutputNode
         assert self.parallel_config.use_ray
+        from ray.dag import InputNode, MultiOutputNode
+        from ray.experimental.channel.torch_tensor_type import TorchTensorType
 
-        # Right now, compiled DAG requires at least 1 arg. We send
+        logger.info("VLLM_USE_RAY_COMPILED_DAG_NCCL_CHANNEL = %s",
-        # a dummy value for now. It will be fixed soon.
+                    envs.VLLM_USE_RAY_COMPILED_DAG_NCCL_CHANNEL)
         with InputNode() as input_data:
-            forward_dag = MultiOutputNode([
+            # Example DAG: PP=2, TP=4
-                worker.execute_model_spmd.bind(  # type: ignore[attr-defined]
+            # (ExecuteModelReq, None) -> 0 -> (ExecuteModelReq, IntermediateOutput) -> 4 -> SamplerOutput   # noqa: E501
-                    input_data) for worker in self.workers
+            #                         -> 1 -> (ExecuteModelReq, IntermediateOutput) -> 5 -> SamplerOutput   # noqa: E501
-            ])
+            #                         -> 2 -> (ExecuteModelReq, IntermediateOutput) -> 6 -> SamplerOutput   # noqa: E501
+            #                         -> 3 -> (ExecuteModelReq, IntermediateOutput) -> 7 -> SamplerOutput   # noqa: E501
+
+            # All workers in the first TP group will take in the
+            # ExecuteModelRequest as input.
+            outputs = [input_data for _ in self.pp_tp_workers[0]]
+            for pp_rank, tp_group in enumerate(self.pp_tp_workers):
+                # Each PP worker takes in the output of the previous PP worker,
+                # and the TP group executes in SPMD fashion.
+                outputs = [
+                    worker.execute_model_spmd.
+                    bind(  # type: ignore[attr-defined]
+                        outputs[i]) for i, worker in enumerate(tp_group)
+                ]
+
+                last_pp_rank = len(self.pp_tp_workers) - 1
+                if pp_rank < last_pp_rank:
+                    # Specify how intermediate tensors should be passed
+                    # between pp stages, no need to specify for the last
+                    # pp stage.
+                    transport = "nccl" \
+                        if envs.VLLM_USE_RAY_COMPILED_DAG_NCCL_CHANNEL \
+                        else "auto"
+                    outputs = [
+                        output.with_type_hint(
+                            TorchTensorType(transport=transport))
+                        for output in outputs
+                    ]
+
+            forward_dag = MultiOutputNode(outputs)
+
         return forward_dag.experimental_compile(enable_asyncio=enable_asyncio)
 
     def __del__(self):

vllm/executor/ray_utils.py

@@ -1,8 +1,8 @@
-from typing import List, Optional, Tuple
+from typing import List, Optional, Tuple, Union
 
 from vllm.config import ParallelConfig
 from vllm.logger import init_logger
-from vllm.sequence import ExecuteModelRequest
+from vllm.sequence import ExecuteModelRequest, IntermediateTensors
 from vllm.utils import get_ip, is_hip, is_tpu, is_xpu
 from vllm.worker.worker_base import WorkerWrapperBase
 
@@ -31,9 +31,17 @@ try:
             gpu_ids = ray.get_gpu_ids()
             return node_id, gpu_ids
 
-        def execute_model_spmd(self, execute_model_req: ExecuteModelRequest):
+        def execute_model_spmd(
-            """Used only when SPMD worker and compiled DAG are both
+            self, req_or_tuple: Union[ExecuteModelRequest,
-            enabled."""
+                                      Tuple[ExecuteModelRequest,
+                                            IntermediateTensors]]):
+            """Execute model in SPMD fashion: used only when SPMD worker and
+            compiled DAG are both enabled.
+
+            Args:
+                req_or_tuple: The request to execute the model, or a tuple
+                    containing the request and intermediate tensors.
+            """
             # TODO(swang): This is needed right now because Ray aDAG executes
             # on a background thread, so we need to reset torch's current
             # device.
@@ -42,7 +50,17 @@ try:
                 torch.cuda.set_device(self.worker.device)
                 self.compiled_dag_cuda_device_set = True
 
-            return self.worker._execute_model_spmd(execute_model_req)
+            if isinstance(req_or_tuple, tuple):
+                execute_model_req, intermediate_tensors = req_or_tuple
+            else:
+                execute_model_req = req_or_tuple
+                intermediate_tensors = None
+
+            output = self.worker._execute_model_spmd(execute_model_req,
+                                                     intermediate_tensors)
+            if isinstance(output, IntermediateTensors):
+                return execute_model_req, output
+            return output
 
     ray_import_err = None
 

vllm/worker/worker_base.py

@@ -285,7 +285,9 @@ class LocalOrDistributedWorkerBase(WorkerBase):
         return output
 
     def _execute_model_spmd(
-        self, execute_model_req: ExecuteModelRequest
+        self,
+        execute_model_req: ExecuteModelRequest,
+        intermediate_tensors: Optional[IntermediateTensors] = None
     ) -> Optional[List[SamplerOutput]]:
         """
         Execute model in Single Program Multiple Data (SPMD) fashion.
@@ -309,7 +311,7 @@ class LocalOrDistributedWorkerBase(WorkerBase):
 
         return self.model_runner.execute_model(
             model_input, self.kv_cache[worker_input.virtual_engine]
-            if self.kv_cache is not None else None)
+            if self.kv_cache is not None else None, intermediate_tensors)
 
 
 class WorkerWrapperBase: