debug on v0.8.5

vllm/zero_overhead/llm_engine.py

@@ -52,7 +52,6 @@ class ZeroOverheadEngine(LLMEngine):
         log_stats: bool,
         usage_context: UsageContext = UsageContext.ENGINE_CONTEXT,
         stat_loggers: Optional[Dict[str, StatLoggerBase]] = None,
-        input_registry: InputRegistry = INPUT_REGISTRY,
         mm_registry: MultiModalRegistry = MULTIMODAL_REGISTRY,
         use_cached_outputs: bool = False,
     ) -> None:
@@ -113,10 +112,6 @@ class ZeroOverheadEngine(LLMEngine):
                                                     self.tokenizer,
                                                     mm_registry)
 
-        self.input_registry = input_registry
-        self.input_processor = input_registry.create_input_processor(
-            self.model_config)
-
         self.model_executor = executor_class(vllm_config=vllm_config, )
 
         if self.model_config.runner_type != "pooling":
@@ -159,11 +154,6 @@ class ZeroOverheadEngine(LLMEngine):
                     self.parallel_config.disable_custom_all_reduce,
                 })
 
-        if self.tokenizer:
-            # Ping the tokenizer to ensure liveness if it runs in a
-            # different process.
-            self.tokenizer.ping()
-
         self.cached_scheduler_outputs = [
             SchedulerOutputState()
             for _ in range(self.parallel_config.pipeline_parallel_size)