[Bugfix] Fix LoRA with PP (vllm-project#7292)

jeejeelee · Aug 8, 2024 · 6dffa4b · 6dffa4b
1 parent 48abee9
commit 6dffa4b
Showing 1 changed file with 3 additions and 0 deletions.
diff --git a/vllm/lora/models.py b/vllm/lora/models.py
@@ -25,6 +25,7 @@
 from vllm.lora.utils import (from_layer, from_layer_logits_processor,
                              parse_fine_tuned_lora_name, replace_submodule)
 from vllm.model_executor.models.interfaces import SupportsLoRA
+from vllm.model_executor.models.utils import PPMissingLayer
 from vllm.utils import is_pin_memory_available
 
 logger = init_logger(__name__)
@@ -432,6 +433,8 @@ def remove_all_adapters(self):
     def _create_lora_modules(self):
         for module_name, module in self.model.named_modules(
                 remove_duplicate=False):
+            if isinstance(module, PPMissingLayer):
+                continue
             if not self._match_target_modules(module_name):
                 continue
             parts = module_name.split(".")[-1]