[V1][Core] Remove unused speculative config from scheduler (vllm-project#15818)

Signed-off-by: Mark McLoughlin <[email protected]>
Signed-off-by: xinyuxiao <[email protected]>

Author: markmc

tests/v1/core/test_scheduler.py

@@ -71,7 +71,6 @@ def create_scheduler(
         scheduler_config,
         model_config,
         cache_config,
-        speculative_config=None,
         lora_config=None,
         log_stats=True,
         structured_output_manager=StructuredOutputManager(vllm_config),

vllm/v1/core/sched/scheduler.py

@@ -7,8 +7,7 @@
 from collections.abc import Iterable
 from typing import Optional, Union
 
-from vllm.config import (CacheConfig, LoRAConfig, ModelConfig, SchedulerConfig,
-                         SpeculativeConfig)
+from vllm.config import CacheConfig, LoRAConfig, ModelConfig, SchedulerConfig
 from vllm.logger import init_logger
 from vllm.multimodal import MULTIMODAL_REGISTRY, MultiModalRegistry
 from vllm.v1.core.encoder_cache_manager import (EncoderCacheManager,
@@ -36,7 +35,6 @@ def __init__(
         model_config: ModelConfig,
         cache_config: CacheConfig,
         lora_config: Optional[LoRAConfig],
-        speculative_config: Optional[SpeculativeConfig],
         structured_output_manager: StructuredOutputManager,
         mm_registry: MultiModalRegistry = MULTIMODAL_REGISTRY,
         include_finished_set: bool = False,
@@ -45,7 +43,6 @@ def __init__(
         self.scheduler_config = scheduler_config
         self.cache_config = cache_config
         self.lora_config = lora_config
-        self.speculative_config = speculative_config
         self.log_stats = log_stats
         self.structured_output_manager = structured_output_manager
 

vllm/v1/engine/core.py

@@ -95,7 +95,6 @@ def __init__(
             model_config=vllm_config.model_config,
             cache_config=vllm_config.cache_config,
             lora_config=vllm_config.lora_config,
-            speculative_config=vllm_config.speculative_config,
             include_finished_set=vllm_config.parallel_config.data_parallel_size
             > 1,
             log_stats=self.log_stats,