[Model] Support Qwen-VL and Qwen-VL-Chat models with text-only inputs (#5710)

Co-authored-by: Roger Wang <ywang@roblox.com>

vllm/model_executor/models/qwen.py

@@ -28,6 +28,7 @@ from vllm.model_executor.layers.vocab_parallel_embedding import (
 from vllm.model_executor.model_loader.weight_utils import default_weight_loader
 from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.sequence import SamplerOutput
+from vllm.utils import print_warning_once
 
 
 class QWenMLP(nn.Module):
@@ -288,6 +289,15 @@ class QWenLMHeadModel(nn.Module):
                 # Skip loading extra bias for GPTQ models.
                 if name.endswith(".bias") and name not in params_dict:
                     continue
+                # Skip loading visual weights to support Qwen-VL models
+                # in cases with text-only inputs
+                # TODO: add support for Qwen-VL
+                if (name not in params_dict
+                        and name.startswith("transformer.visual.")):
+                    print_warning_once(
+                        "Only text inputs are allowed. Images won't be handled "
+                        "until Qwen-VL models are fully supported.")
+                    continue
                 param = params_dict[name]
                 weight_loader = getattr(param, "weight_loader",
                                         default_weight_loader)