diff --git a/vllm_omni/utils/mm_outputs.py b/vllm_omni/utils/mm_outputs.py index 3009e21d4..656ced316 100644 --- a/vllm_omni/utils/mm_outputs.py +++ b/vllm_omni/utils/mm_outputs.py @@ -20,6 +20,9 @@ def build_mm_cpu(multimodal_outputs: dict) -> dict[str, object]: Args: multimodal_outputs: Multimodal dict mapping strings to objects. """ + if not multimodal_outputs: + return {} + # Pre-copy multimodal tensors to CPU once (not per-request) to avoid # redundant D2H transfers when gpu_resident_buffer_keys keeps them on GPU. mm_cpu: dict[str, object] = {} @@ -28,11 +31,10 @@ def build_mm_cpu(multimodal_outputs: dict) -> dict[str, object]: if not isinstance(multimodal_outputs, Mapping): logger.warning("Multimodal outputs are not a dict and will not be passed") - if multimodal_outputs: - for k, v in multimodal_outputs.items(): - cpu_v = _to_cpu(v) - if cpu_v is not None: - mm_cpu[k] = cpu_v + for k, v in multimodal_outputs.items(): + cpu_v = _to_cpu(v) + if cpu_v is not None: + mm_cpu[k] = cpu_v return mm_cpu diff --git a/vllm_omni/worker/omni_connector_model_runner_mixin.py b/vllm_omni/worker/omni_connector_model_runner_mixin.py index 205a28c6d..b173f81ac 100644 --- a/vllm_omni/worker/omni_connector_model_runner_mixin.py +++ b/vllm_omni/worker/omni_connector_model_runner_mixin.py @@ -110,7 +110,7 @@ class OmniConnectorModelRunnerMixin: self._custom_process_func_path, self._custom_process_func = self._load_custom_func(model_config) self._custom_process_supports_is_finished = self._custom_process_supports_is_finished_kwarg() - logger.info( + logger.debug( "[Stage-%s] init_omni_connectors: async_chunk=%s, custom_process_func=%s, connector=%s, func_path=%s", self._stage_id, self._async_chunk, @@ -1154,7 +1154,7 @@ class OmniConnectorModelRunnerMixin: connector_put_key = f"{request_id}_{self._stage_id}_{chunk_id}" if chunk_id == 0: - logger.info( + logger.debug( "[Stage-%s] send_chunk: first chunk enqueued, req=%s key=%s", self._stage_id, request_id, @@ -1252,7 +1252,7 @@ class OmniConnectorModelRunnerMixin: active_requests = getattr(self, "requests", None) if active_requests is not None and request_id not in active_requests: - logger.info("Skip receiving KV cache for inactive request %s", request_id) + logger.debug("Skip receiving KV cache for inactive request %s", request_id) return False primary_ok = False @@ -1272,7 +1272,7 @@ class OmniConnectorModelRunnerMixin: if cfg_kvs and hasattr(req, "sampling_params") and req.sampling_params is not None: for key, value in cfg_kvs.items(): setattr(req.sampling_params, key, value) - logger.info("Applied CFG KV caches: %s", list(cfg_kvs.keys())) + logger.debug("Applied CFG KV caches: %s", list(cfg_kvs.keys())) except Exception: logger.exception("Failed to collect CFG KV caches for %s", request_id) @@ -1657,7 +1657,7 @@ class OmniConnectorModelRunnerMixin: _recv_poll_count += 1 if _recv_poll_count % 5000 == 1: - logger.info( + logger.debug( "[Stage-%s] _recv_loop: polling %s pending reqs: %s (poll#%s)", self._stage_id, len(pending_ids),