revert changes to v0 model runner

vllm-project · Nov 22, 2024 · 3cdc02d · 3cdc02d
1 parent 191afc8
commit 3cdc02d
Showing 1 changed file with 1 addition and 5 deletions.
diff --git a/vllm/worker/model_runner.py b/vllm/worker/model_runner.py
@@ -917,8 +917,7 @@ def build(self) -> ModelInputForGPU:
             lora_mapping = LoRAMapping(
                 **dict(index_mapping=lora_index_mapping,
                        prompt_mapping=lora_prompt_mapping,
-                       is_prefill = False))
-                       #is_prefill=not self.decode_only))
+                       is_prefill=not self.decode_only))
 
         # Prompt adapter data.
         prompt_adapter_requests: Set[PromptAdapterRequest] = set()
@@ -1613,9 +1612,6 @@ def execute_model(
         if self.lora_config:
             assert model_input.lora_requests is not None
             assert model_input.lora_mapping is not None
-            #print (f" index mapping : {len(model_input.lora_mapping.index_mapping)} {model_input.lora_mapping.index_mapping}")
-            #print (f" prompt mapping : {len(model_input.lora_mapping.prompt_mapping)} {model_input.lora_mapping.prompt_mapping}")
-            #print (f" # lora requests : {len(model_input.lora_requests)}")
             self.set_active_loras(model_input.lora_requests,
                                   model_input.lora_mapping)